diff --git a/.github/scripts/package-lock.json b/.github/scripts/package-lock.json
new file mode 100644
index 000000000..7a9cfb168
--- /dev/null
+++ b/.github/scripts/package-lock.json
@@ -0,0 +1,557 @@
+{
+  "name": "docs-search-excerpt-smoke",
+  "lockfileVersion": 3,
+  "requires": true,
+  "packages": {
+    "": {
+      "name": "docs-search-excerpt-smoke",
+      "devDependencies": {
+        "jsdom": "29.1.1"
+      }
+    },
+    "node_modules/@asamuzakjp/css-color": {
+      "version": "5.1.11",
+      "resolved": "https://registry.npmjs.org/@asamuzakjp/css-color/-/css-color-5.1.11.tgz",
+      "integrity": "sha512-KVw6qIiCTUQhByfTd78h2yD1/00waTmm9uy/R7Ck/ctUyAPj+AEDLkQIdJW0T8+qGgj3j5bpNKK7Q3G+LedJWg==",
+      "dev": true,
+      "license": "MIT",
+      "dependencies": {
+        "@asamuzakjp/generational-cache": "^1.0.1",
+        "@csstools/css-calc": "^3.2.0",
+        "@csstools/css-color-parser": "^4.1.0",
+        "@csstools/css-parser-algorithms": "^4.0.0",
+        "@csstools/css-tokenizer": "^4.0.0"
+      },
+      "engines": {
+        "node": "^20.19.0 || ^22.12.0 || >=24.0.0"
+      }
+    },
+    "node_modules/@asamuzakjp/dom-selector": {
+      "version": "7.1.1",
+      "resolved": "https://registry.npmjs.org/@asamuzakjp/dom-selector/-/dom-selector-7.1.1.tgz",
+      "integrity": "sha512-67RZDnYRc8H/8MLDgQCDE//zoqVFwajkepHZgmXrbwybzXOEwOWGPYGmALYl9J2DOLfFPPs6kKCqmbzV895hTQ==",
+      "dev": true,
+      "license": "MIT",
+      "dependencies": {
+        "@asamuzakjp/generational-cache": "^1.0.1",
+        "@asamuzakjp/nwsapi": "^2.3.9",
+        "bidi-js": "^1.0.3",
+        "css-tree": "^3.2.1",
+        "is-potential-custom-element-name": "^1.0.1"
+      },
+      "engines": {
+        "node": "^20.19.0 || ^22.12.0 || >=24.0.0"
+      }
+    },
+    "node_modules/@asamuzakjp/generational-cache": {
+      "version": "1.0.1",
+      "resolved": "https://registry.npmjs.org/@asamuzakjp/generational-cache/-/generational-cache-1.0.1.tgz",
+      "integrity": "sha512-wajfB8KqzMCN2KGNFdLkReeHncd0AslUSrvHVvvYWuU8ghncRJoA50kT3zP9MVL0+9g4/67H+cdvBskj9THPzg==",
+      "dev": true,
+      "license": "MIT",
+      "engines": {
+        "node": "^20.19.0 || ^22.12.0 || >=24.0.0"
+      }
+    },
+    "node_modules/@asamuzakjp/nwsapi": {
+      "version": "2.3.9",
+      "resolved": "https://registry.npmjs.org/@asamuzakjp/nwsapi/-/nwsapi-2.3.9.tgz",
+      "integrity": "sha512-n8GuYSrI9bF7FFZ/SjhwevlHc8xaVlb/7HmHelnc/PZXBD2ZR49NnN9sMMuDdEGPeeRQ5d0hqlSlEpgCX3Wl0Q==",
+      "dev": true,
+      "license": "MIT"
+    },
+    "node_modules/@bramus/specificity": {
+      "version": "2.4.2",
+      "resolved": "https://registry.npmjs.org/@bramus/specificity/-/specificity-2.4.2.tgz",
+      "integrity": "sha512-ctxtJ/eA+t+6q2++vj5j7FYX3nRu311q1wfYH3xjlLOsczhlhxAg2FWNUXhpGvAw3BWo1xBcvOV6/YLc2r5FJw==",
+      "dev": true,
+      "license": "MIT",
+      "dependencies": {
+        "css-tree": "^3.0.0"
+      },
+      "bin": {
+        "specificity": "bin/cli.js"
+      }
+    },
+    "node_modules/@csstools/color-helpers": {
+      "version": "6.1.0",
+      "resolved": "https://registry.npmjs.org/@csstools/color-helpers/-/color-helpers-6.1.0.tgz",
+      "integrity": "sha512-064IFJdjTfUqnjpCVpMOdbr8FLQBhinbZj6yRv2An2E41O/pLEXqfFRWqGq/SxlE5PEUYTlvWsG2r8MswAVvkg==",
+      "dev": true,
+      "funding": [
+        {
+          "type": "github",
+          "url": "https://github.com/sponsors/csstools"
+        },
+        {
+          "type": "opencollective",
+          "url": "https://opencollective.com/csstools"
+        }
+      ],
+      "license": "MIT-0",
+      "engines": {
+        "node": ">=20.19.0"
+      }
+    },
+    "node_modules/@csstools/css-calc": {
+      "version": "3.2.1",
+      "resolved": "https://registry.npmjs.org/@csstools/css-calc/-/css-calc-3.2.1.tgz",
+      "integrity": "sha512-DtdHlgXh5ZkA43cwBcAm+huzgJiwx3ZTWVjBs94kwz2xKqSimDA3lBgCjphYgwgVUMWatSM0pDd8TILB1yrVVg==",
+      "dev": true,
+      "funding": [
+        {
+          "type": "github",
+          "url": "https://github.com/sponsors/csstools"
+        },
+        {
+          "type": "opencollective",
+          "url": "https://opencollective.com/csstools"
+        }
+      ],
+      "license": "MIT",
+      "engines": {
+        "node": ">=20.19.0"
+      },
+      "peerDependencies": {
+        "@csstools/css-parser-algorithms": "^4.0.0",
+        "@csstools/css-tokenizer": "^4.0.0"
+      }
+    },
+    "node_modules/@csstools/css-color-parser": {
+      "version": "4.1.9",
+      "resolved": "https://registry.npmjs.org/@csstools/css-color-parser/-/css-color-parser-4.1.9.tgz",
+      "integrity": "sha512-paQcIaOO53Rk5+YrBaBjm/SgrV4INImjo2BT1DtQRYr+XeTRbeAYlS+jxXp9drqvKmtFnWRJKIalDLhZZDu42A==",
+      "dev": true,
+      "funding": [
+        {
+          "type": "github",
+          "url": "https://github.com/sponsors/csstools"
+        },
+        {
+          "type": "opencollective",
+          "url": "https://opencollective.com/csstools"
+        }
+      ],
+      "license": "MIT",
+      "dependencies": {
+        "@csstools/color-helpers": "^6.1.0",
+        "@csstools/css-calc": "^3.2.1"
+      },
+      "engines": {
+        "node": ">=20.19.0"
+      },
+      "peerDependencies": {
+        "@csstools/css-parser-algorithms": "^4.0.0",
+        "@csstools/css-tokenizer": "^4.0.0"
+      }
+    },
+    "node_modules/@csstools/css-parser-algorithms": {
+      "version": "4.0.0",
+      "resolved": "https://registry.npmjs.org/@csstools/css-parser-algorithms/-/css-parser-algorithms-4.0.0.tgz",
+      "integrity": "sha512-+B87qS7fIG3L5h3qwJ/IFbjoVoOe/bpOdh9hAjXbvx0o8ImEmUsGXN0inFOnk2ChCFgqkkGFQ+TpM5rbhkKe4w==",
+      "dev": true,
+      "funding": [
+        {
+          "type": "github",
+          "url": "https://github.com/sponsors/csstools"
+        },
+        {
+          "type": "opencollective",
+          "url": "https://opencollective.com/csstools"
+        }
+      ],
+      "license": "MIT",
+      "peer": true,
+      "engines": {
+        "node": ">=20.19.0"
+      },
+      "peerDependencies": {
+        "@csstools/css-tokenizer": "^4.0.0"
+      }
+    },
+    "node_modules/@csstools/css-syntax-patches-for-csstree": {
+      "version": "1.1.6",
+      "resolved": "https://registry.npmjs.org/@csstools/css-syntax-patches-for-csstree/-/css-syntax-patches-for-csstree-1.1.6.tgz",
+      "integrity": "sha512-TcJCWFbXLPpJYq6z7bfOyjWYJDiDg2/I4gyUC9pqPNqHFRIey0EB0q0L5cSnQDfWJg8Jd6VadakxdIez/3zkqQ==",
+      "dev": true,
+      "funding": [
+        {
+          "type": "github",
+          "url": "https://github.com/sponsors/csstools"
+        },
+        {
+          "type": "opencollective",
+          "url": "https://opencollective.com/csstools"
+        }
+      ],
+      "license": "MIT-0",
+      "peerDependencies": {
+        "css-tree": "^3.2.1"
+      },
+      "peerDependenciesMeta": {
+        "css-tree": {
+          "optional": true
+        }
+      }
+    },
+    "node_modules/@csstools/css-tokenizer": {
+      "version": "4.0.0",
+      "resolved": "https://registry.npmjs.org/@csstools/css-tokenizer/-/css-tokenizer-4.0.0.tgz",
+      "integrity": "sha512-QxULHAm7cNu72w97JUNCBFODFaXpbDg+dP8b/oWFAZ2MTRppA3U00Y2L1HqaS4J6yBqxwa/Y3nMBaxVKbB/NsA==",
+      "dev": true,
+      "funding": [
+        {
+          "type": "github",
+          "url": "https://github.com/sponsors/csstools"
+        },
+        {
+          "type": "opencollective",
+          "url": "https://opencollective.com/csstools"
+        }
+      ],
+      "license": "MIT",
+      "peer": true,
+      "engines": {
+        "node": ">=20.19.0"
+      }
+    },
+    "node_modules/@exodus/bytes": {
+      "version": "1.15.1",
+      "resolved": "https://registry.npmjs.org/@exodus/bytes/-/bytes-1.15.1.tgz",
+      "integrity": "sha512-S6mL0yNB/Abt9Ei4tq8gDhcczc4S3+vQ4ra7vxnAf+YHC02srtqxKKZghx2Dq6p0e66THKwR6r8N6P95wEty7Q==",
+      "dev": true,
+      "license": "MIT",
+      "engines": {
+        "node": "^20.19.0 || ^22.12.0 || >=24.0.0"
+      },
+      "peerDependencies": {
+        "@noble/hashes": "^1.8.0 || ^2.0.0"
+      },
+      "peerDependenciesMeta": {
+        "@noble/hashes": {
+          "optional": true
+        }
+      }
+    },
+    "node_modules/bidi-js": {
+      "version": "1.0.3",
+      "resolved": "https://registry.npmjs.org/bidi-js/-/bidi-js-1.0.3.tgz",
+      "integrity": "sha512-RKshQI1R3YQ+n9YJz2QQ147P66ELpa1FQEg20Dk8oW9t2KgLbpDLLp9aGZ7y8WHSshDknG0bknqGw5/tyCs5tw==",
+      "dev": true,
+      "license": "MIT",
+      "dependencies": {
+        "require-from-string": "^2.0.2"
+      }
+    },
+    "node_modules/css-tree": {
+      "version": "3.2.1",
+      "resolved": "https://registry.npmjs.org/css-tree/-/css-tree-3.2.1.tgz",
+      "integrity": "sha512-X7sjQzceUhu1u7Y/ylrRZFU2FS6LRiFVp6rKLPg23y3x3c3DOKAwuXGDp+PAGjh6CSnCjYeAul8pcT8bAl+lSA==",
+      "dev": true,
+      "license": "MIT",
+      "dependencies": {
+        "mdn-data": "2.27.1",
+        "source-map-js": "^1.2.1"
+      },
+      "engines": {
+        "node": "^10 || ^12.20.0 || ^14.13.0 || >=15.0.0"
+      }
+    },
+    "node_modules/data-urls": {
+      "version": "7.0.0",
+      "resolved": "https://registry.npmjs.org/data-urls/-/data-urls-7.0.0.tgz",
+      "integrity": "sha512-23XHcCF+coGYevirZceTVD7NdJOqVn+49IHyxgszm+JIiHLoB2TkmPtsYkNWT1pvRSGkc35L6NHs0yHkN2SumA==",
+      "dev": true,
+      "license": "MIT",
+      "dependencies": {
+        "whatwg-mimetype": "^5.0.0",
+        "whatwg-url": "^16.0.0"
+      },
+      "engines": {
+        "node": "^20.19.0 || ^22.12.0 || >=24.0.0"
+      }
+    },
+    "node_modules/decimal.js": {
+      "version": "10.6.0",
+      "resolved": "https://registry.npmjs.org/decimal.js/-/decimal.js-10.6.0.tgz",
+      "integrity": "sha512-YpgQiITW3JXGntzdUmyUR1V812Hn8T1YVXhCu+wO3OpS4eU9l4YdD3qjyiKdV6mvV29zapkMeD390UVEf2lkUg==",
+      "dev": true,
+      "license": "MIT"
+    },
+    "node_modules/entities": {
+      "version": "8.0.0",
+      "resolved": "https://registry.npmjs.org/entities/-/entities-8.0.0.tgz",
+      "integrity": "sha512-zwfzJecQ/Uej6tusMqwAqU/6KL2XaB2VZ2Jg54Je6ahNBGNH6Ek6g3jjNCF0fG9EWQKGZNddNjU5F1ZQn/sBnA==",
+      "dev": true,
+      "license": "BSD-2-Clause",
+      "engines": {
+        "node": ">=20.19.0"
+      },
+      "funding": {
+        "url": "https://github.com/fb55/entities?sponsor=1"
+      }
+    },
+    "node_modules/html-encoding-sniffer": {
+      "version": "6.0.0",
+      "resolved": "https://registry.npmjs.org/html-encoding-sniffer/-/html-encoding-sniffer-6.0.0.tgz",
+      "integrity": "sha512-CV9TW3Y3f8/wT0BRFc1/KAVQ3TUHiXmaAb6VW9vtiMFf7SLoMd1PdAc4W3KFOFETBJUb90KatHqlsZMWV+R9Gg==",
+      "dev": true,
+      "license": "MIT",
+      "dependencies": {
+        "@exodus/bytes": "^1.6.0"
+      },
+      "engines": {
+        "node": "^20.19.0 || ^22.12.0 || >=24.0.0"
+      }
+    },
+    "node_modules/is-potential-custom-element-name": {
+      "version": "1.0.1",
+      "resolved": "https://registry.npmjs.org/is-potential-custom-element-name/-/is-potential-custom-element-name-1.0.1.tgz",
+      "integrity": "sha512-bCYeRA2rVibKZd+s2625gGnGF/t7DSqDs4dP7CrLA1m7jKWz6pps0LpYLJN8Q64HtmPKJ1hrN3nzPNKFEKOUiQ==",
+      "dev": true,
+      "license": "MIT"
+    },
+    "node_modules/jsdom": {
+      "version": "29.1.1",
+      "resolved": "https://registry.npmjs.org/jsdom/-/jsdom-29.1.1.tgz",
+      "integrity": "sha512-ECi4Fi2f7BdJtUKTflYRTiaMxIB0O6zfR1fX0GXpUrf6flp8QIYn1UT20YQqdSOfk2dfkCwS8LAFoJDEppNK5Q==",
+      "dev": true,
+      "license": "MIT",
+      "dependencies": {
+        "@asamuzakjp/css-color": "^5.1.11",
+        "@asamuzakjp/dom-selector": "^7.1.1",
+        "@bramus/specificity": "^2.4.2",
+        "@csstools/css-syntax-patches-for-csstree": "^1.1.3",
+        "@exodus/bytes": "^1.15.0",
+        "css-tree": "^3.2.1",
+        "data-urls": "^7.0.0",
+        "decimal.js": "^10.6.0",
+        "html-encoding-sniffer": "^6.0.0",
+        "is-potential-custom-element-name": "^1.0.1",
+        "lru-cache": "^11.3.5",
+        "parse5": "^8.0.1",
+        "saxes": "^6.0.0",
+        "symbol-tree": "^3.2.4",
+        "tough-cookie": "^6.0.1",
+        "undici": "^7.25.0",
+        "w3c-xmlserializer": "^5.0.0",
+        "webidl-conversions": "^8.0.1",
+        "whatwg-mimetype": "^5.0.0",
+        "whatwg-url": "^16.0.1",
+        "xml-name-validator": "^5.0.0"
+      },
+      "engines": {
+        "node": "^20.19.0 || ^22.13.0 || >=24.0.0"
+      },
+      "peerDependencies": {
+        "canvas": "^3.0.0"
+      },
+      "peerDependenciesMeta": {
+        "canvas": {
+          "optional": true
+        }
+      }
+    },
+    "node_modules/lru-cache": {
+      "version": "11.5.2",
+      "resolved": "https://registry.npmjs.org/lru-cache/-/lru-cache-11.5.2.tgz",
+      "integrity": "sha512-4pfM1Ff0x50o0tQwb5ucw/RzNyD0/YJME6IVcStalZuMWxdt3sR3huStTtxz4PUmvZfRguvDejasvQ2kifR11g==",
+      "dev": true,
+      "license": "BlueOak-1.0.0",
+      "engines": {
+        "node": "20 || >=22"
+      }
+    },
+    "node_modules/mdn-data": {
+      "version": "2.27.1",
+      "resolved": "https://registry.npmjs.org/mdn-data/-/mdn-data-2.27.1.tgz",
+      "integrity": "sha512-9Yubnt3e8A0OKwxYSXyhLymGW4sCufcLG6VdiDdUGVkPhpqLxlvP5vl1983gQjJl3tqbrM731mjaZaP68AgosQ==",
+      "dev": true,
+      "license": "CC0-1.0"
+    },
+    "node_modules/parse5": {
+      "version": "8.0.1",
+      "resolved": "https://registry.npmjs.org/parse5/-/parse5-8.0.1.tgz",
+      "integrity": "sha512-z1e/HMG90obSGeidlli3hj7cbocou0/wa5HacvI3ASx34PecNjNQeaHNo5WIZpWofN9kgkqV1q5YvXe3F0FoPw==",
+      "dev": true,
+      "license": "MIT",
+      "dependencies": {
+        "entities": "^8.0.0"
+      },
+      "funding": {
+        "url": "https://github.com/inikulin/parse5?sponsor=1"
+      }
+    },
+    "node_modules/punycode": {
+      "version": "2.3.1",
+      "resolved": "https://registry.npmjs.org/punycode/-/punycode-2.3.1.tgz",
+      "integrity": "sha512-vYt7UD1U9Wg6138shLtLOvdAu+8DsC/ilFtEVHcH+wydcSpNE20AfSOduf6MkRFahL5FY7X1oU7nKVZFtfq8Fg==",
+      "dev": true,
+      "license": "MIT",
+      "engines": {
+        "node": ">=6"
+      }
+    },
+    "node_modules/require-from-string": {
+      "version": "2.0.2",
+      "resolved": "https://registry.npmjs.org/require-from-string/-/require-from-string-2.0.2.tgz",
+      "integrity": "sha512-Xf0nWe6RseziFMu+Ap9biiUbmplq6S9/p+7w7YXP/JBHhrUDDUhwa+vANyubuqfZWTveU//DYVGsDG7RKL/vEw==",
+      "dev": true,
+      "license": "MIT",
+      "engines": {
+        "node": ">=0.10.0"
+      }
+    },
+    "node_modules/saxes": {
+      "version": "6.0.0",
+      "resolved": "https://registry.npmjs.org/saxes/-/saxes-6.0.0.tgz",
+      "integrity": "sha512-xAg7SOnEhrm5zI3puOOKyy1OMcMlIJZYNJY7xLBwSze0UjhPLnWfj2GF2EpT0jmzaJKIWKHLsaSSajf35bcYnA==",
+      "dev": true,
+      "license": "ISC",
+      "dependencies": {
+        "xmlchars": "^2.2.0"
+      },
+      "engines": {
+        "node": ">=v12.22.7"
+      }
+    },
+    "node_modules/source-map-js": {
+      "version": "1.2.1",
+      "resolved": "https://registry.npmjs.org/source-map-js/-/source-map-js-1.2.1.tgz",
+      "integrity": "sha512-UXWMKhLOwVKb728IUtQPXxfYU+usdybtUrK/8uGE8CQMvrhOpwvzDBwj0QhSL7MQc7vIsISBG8VQ8+IDQxpfQA==",
+      "dev": true,
+      "license": "BSD-3-Clause",
+      "engines": {
+        "node": ">=0.10.0"
+      }
+    },
+    "node_modules/symbol-tree": {
+      "version": "3.2.4",
+      "resolved": "https://registry.npmjs.org/symbol-tree/-/symbol-tree-3.2.4.tgz",
+      "integrity": "sha512-9QNk5KwDF+Bvz+PyObkmSYjI5ksVUYtjW7AU22r2NKcfLJcXp96hkDWU3+XndOsUb+AQ9QhfzfCT2O+CNWT5Tw==",
+      "dev": true,
+      "license": "MIT"
+    },
+    "node_modules/tldts": {
+      "version": "7.4.9",
+      "resolved": "https://registry.npmjs.org/tldts/-/tldts-7.4.9.tgz",
+      "integrity": "sha512-3kZ8wQQ/k5DrChD4X4FVvr2D7E5uoRgAqkPyLpSCGUvqOvqu+JEdr3mwMUaVWb+vMHZaKhF5fp2PBigKsui7hA==",
+      "dev": true,
+      "license": "MIT",
+      "dependencies": {
+        "tldts-core": "^7.4.9"
+      },
+      "bin": {
+        "tldts": "bin/cli.js"
+      }
+    },
+    "node_modules/tldts-core": {
+      "version": "7.4.9",
+      "resolved": "https://registry.npmjs.org/tldts-core/-/tldts-core-7.4.9.tgz",
+      "integrity": "sha512-DxKfPBI52p2msTEu7MPhdpdDTBhhVQg1a/8PjQckeyAvO13eMYElX545grIp6nnTGIMZlRvFZPvFhvI/WIz2Vg==",
+      "dev": true,
+      "license": "MIT"
+    },
+    "node_modules/tough-cookie": {
+      "version": "6.0.2",
+      "resolved": "https://registry.npmjs.org/tough-cookie/-/tough-cookie-6.0.2.tgz",
+      "integrity": "sha512-exgYmnmL/sJpR3upZfXG5PoatXQii55xAiXGXzY+sROLZ/Y+SLcp9PgJNI9Vz37HpQ74WvDcLT8eqm+kV3FzrA==",
+      "dev": true,
+      "license": "BSD-3-Clause",
+      "dependencies": {
+        "tldts": "^7.0.5"
+      },
+      "engines": {
+        "node": ">=16"
+      }
+    },
+    "node_modules/tr46": {
+      "version": "6.0.0",
+      "resolved": "https://registry.npmjs.org/tr46/-/tr46-6.0.0.tgz",
+      "integrity": "sha512-bLVMLPtstlZ4iMQHpFHTR7GAGj2jxi8Dg0s2h2MafAE4uSWF98FC/3MomU51iQAMf8/qDUbKWf5GxuvvVcXEhw==",
+      "dev": true,
+      "license": "MIT",
+      "dependencies": {
+        "punycode": "^2.3.1"
+      },
+      "engines": {
+        "node": ">=20"
+      }
+    },
+    "node_modules/undici": {
+      "version": "7.28.0",
+      "resolved": "https://registry.npmjs.org/undici/-/undici-7.28.0.tgz",
+      "integrity": "sha512-cRZYrTDwWznlnRiPjggAGxZXanty6M8RV1ff8Wm4LWXBp7/IG8v5DnOm74DtUBp9OONpK75YlPnIjQqX0dBDtA==",
+      "dev": true,
+      "license": "MIT",
+      "engines": {
+        "node": ">=20.18.1"
+      }
+    },
+    "node_modules/w3c-xmlserializer": {
+      "version": "5.0.0",
+      "resolved": "https://registry.npmjs.org/w3c-xmlserializer/-/w3c-xmlserializer-5.0.0.tgz",
+      "integrity": "sha512-o8qghlI8NZHU1lLPrpi2+Uq7abh4GGPpYANlalzWxyWteJOCsr/P+oPBA49TOLu5FTZO4d3F9MnWJfiMo4BkmA==",
+      "dev": true,
+      "license": "MIT",
+      "dependencies": {
+        "xml-name-validator": "^5.0.0"
+      },
+      "engines": {
+        "node": ">=18"
+      }
+    },
+    "node_modules/webidl-conversions": {
+      "version": "8.0.1",
+      "resolved": "https://registry.npmjs.org/webidl-conversions/-/webidl-conversions-8.0.1.tgz",
+      "integrity": "sha512-BMhLD/Sw+GbJC21C/UgyaZX41nPt8bUTg+jWyDeg7e7YN4xOM05YPSIXceACnXVtqyEw/LMClUQMtMZ+PGGpqQ==",
+      "dev": true,
+      "license": "BSD-2-Clause",
+      "engines": {
+        "node": ">=20"
+      }
+    },
+    "node_modules/whatwg-mimetype": {
+      "version": "5.0.0",
+      "resolved": "https://registry.npmjs.org/whatwg-mimetype/-/whatwg-mimetype-5.0.0.tgz",
+      "integrity": "sha512-sXcNcHOC51uPGF0P/D4NVtrkjSU2fNsm9iog4ZvZJsL3rjoDAzXZhkm2MWt1y+PUdggKAYVoMAIYcs78wJ51Cw==",
+      "dev": true,
+      "license": "MIT",
+      "engines": {
+        "node": ">=20"
+      }
+    },
+    "node_modules/whatwg-url": {
+      "version": "16.0.1",
+      "resolved": "https://registry.npmjs.org/whatwg-url/-/whatwg-url-16.0.1.tgz",
+      "integrity": "sha512-1to4zXBxmXHV3IiSSEInrreIlu02vUOvrhxJJH5vcxYTBDAx51cqZiKdyTxlecdKNSjj8EcxGBxNf6Vg+945gw==",
+      "dev": true,
+      "license": "MIT",
+      "dependencies": {
+        "@exodus/bytes": "^1.11.0",
+        "tr46": "^6.0.0",
+        "webidl-conversions": "^8.0.1"
+      },
+      "engines": {
+        "node": "^20.19.0 || ^22.12.0 || >=24.0.0"
+      }
+    },
+    "node_modules/xml-name-validator": {
+      "version": "5.0.0",
+      "resolved": "https://registry.npmjs.org/xml-name-validator/-/xml-name-validator-5.0.0.tgz",
+      "integrity": "sha512-EvGK8EJ3DhaHfbRlETOWAS5pO9MZITeauHKJyb8wyajUfQUenkIg2MvLDTZ4T/TgIcm3HU0TFBgWWboAZ30UHg==",
+      "dev": true,
+      "license": "Apache-2.0",
+      "engines": {
+        "node": ">=18"
+      }
+    },
+    "node_modules/xmlchars": {
+      "version": "2.2.0",
+      "resolved": "https://registry.npmjs.org/xmlchars/-/xmlchars-2.2.0.tgz",
+      "integrity": "sha512-JZnDKK8B0RCDw84FNdDAIpZK+JuJw+s7Lz8nksI7SIuU3UXJJslUthsi+uWBUYOwPFwW7W7PRLRfUKpxjtjFCw==",
+      "dev": true,
+      "license": "MIT"
+    }
+  }
+}
diff --git a/.github/scripts/package.json b/.github/scripts/package.json
new file mode 100644
index 000000000..c16755d9e
--- /dev/null
+++ b/.github/scripts/package.json
@@ -0,0 +1,8 @@
+{
+  "name": "docs-search-excerpt-smoke",
+  "private": true,
+  "description": "Pinned dependency set for search_excerpt_smoke.mjs (run via npm ci --prefix .github/scripts)",
+  "devDependencies": {
+    "jsdom": "29.1.1"
+  }
+}
diff --git a/.github/scripts/search_excerpt_smoke.mjs b/.github/scripts/search_excerpt_smoke.mjs
new file mode 100644
index 000000000..e8a92f5de
--- /dev/null
+++ b/.github/scripts/search_excerpt_smoke.mjs
@@ -0,0 +1,202 @@
+/* Behavioral smoke test for the search-excerpt anchor fix.
+ *
+ * Loads the BUILT Sphinx search machinery (doctools.js + searchtools.js) and
+ * the shipped shims (searchtools-css-escape.js, legacy-fragment-redirect.js)
+ * into a jsdom window, then drives the exact code path search results use -
+ * Search.htmlToText(pageHtml, anchor) - against built tutorial HTML for every
+ * CSS-special anchor class (apostrophe, parentheses+comma, colon, and a
+ * synthetic legacy digit anchor). Without the CSS.escape wrapper each of
+ * these throws SyntaxError and the excerpt is dropped; the test asserts a
+ * non-empty, section-scoped excerpt instead. Also asserts the
+ * legacy-fragment redirect rewrites a pre-rename numbered hash.
+ *
+ * Run from the repo root after a docs build (jsdom is pinned by the
+ * committed package.json + package-lock.json next to this script):
+ *   npm ci --prefix .github/scripts --ignore-scripts
+ *   node .github/scripts/search_excerpt_smoke.mjs
+ */
+import { readFileSync, readdirSync } from "node:fs";
+import { JSDOM } from "jsdom";
+
+const ROOT = "docs/_build/html";
+let failures = 0;
+
+const check = (label, ok, detail) => {
+  console.log(`${ok ? "PASS" : "FAIL"}: ${label}${detail ? ` - ${detail}` : ""}`);
+  if (!ok) failures += 1;
+};
+
+// jsdom quirks vs browsers: window.CSS is absent (Part 1's env stub inlines
+// a minimal spec-subset escaper so the wrapper's guard passes and its logic
+// runs against jsdom's real selector engine; "\3<digit> " is the correct
+// CSS hex escape for ASCII digits since '0'-'9' are 0x30-0x39), and
+// document.readyState stays "loading" until jsdom finishes its async load
+// (so shims register DOMContentLoaded listeners - await before asserting).
+const domLoaded = (window) =>
+  new Promise((resolve) => {
+    if (window.document.readyState !== "loading") resolve();
+    else window.document.addEventListener("DOMContentLoaded", () => resolve());
+  });
+
+// --- Part 0: production emit order on the built search page --------------
+// The shim relies on being emitted BEFORE searchtools.js (it wraps at
+// DOMContentLoaded, and listener order follows script order). Pin that
+// order here so a theme/template change that flips it fails CI instead of
+// silently racing the wrapper against the search runner.
+{
+  const searchPage = readFileSync(`${ROOT}/search.html`, "utf8");
+  const shimAt = searchPage.indexOf("searchtools-css-escape.js");
+  const toolsAt = searchPage.indexOf('src="_static/searchtools.js"');
+  check(
+    "search.html emits the shim before searchtools.js",
+    shimAt !== -1 && toolsAt !== -1 && shimAt < toolsAt,
+    `shim@${shimAt} searchtools@${toolsAt}`,
+  );
+}
+
+// --- Part 1: Search.htmlToText excerpt rendering (production order) ------
+// Reproduces the real page lifecycle instead of a settled-DOM eval: the
+// scripts run as inline classic <script> tags in the PRODUCTION order -
+// doctools, shim (Search undefined at its parse time), then searchtools -
+// while document.readyState is "loading", so the shim and searchtools race
+// through DOMContentLoaded exactly as on the live search page. If the
+// wrapper loses that race, the special-anchor cases below throw.
+// (The full query pipeline is not driven: performSearch defers without a
+// loaded searchindex and fetches result pages; Search.htmlToText is the
+// excerpt entry point either way.)
+{
+  const inline = (relPath) => {
+    const src = readFileSync(`${ROOT}/${relPath}`, "utf8");
+    if (src.includes("</script")) {
+      throw new Error(`${relPath} contains "</script" - cannot inline safely`);
+    }
+    return `<script>${src}</script>`;
+  };
+  const envStub =
+    "<script>window.DOCUMENTATION_OPTIONS = {}; window.CSS = { escape: " +
+    "(value) => String(value).replace(/[\\0-,./:-@[-^`{-~]|^\\d/g, " +
+    "(ch) => (ch >= '0' && ch <= '9' ? `\\\\3${ch} ` : `\\\\${ch}`)) };" +
+    "</script>";
+  const harnessHtml = [
+    "<!doctype html><html><head>",
+    envStub,
+    inline("_static/doctools.js"),
+    inline("_static/searchtools-css-escape.js"),
+    inline("_static/searchtools.js"),
+    "</head><body></body></html>",
+  ].join("\n");
+  const dom = new JSDOM(harnessHtml, {
+    url: "http://localhost/search.html",
+    runScripts: "dangerously",
+  });
+  const { window } = dom;
+  await domLoaded(window);
+  const Search = window.eval("Search");
+
+  const page = readFileSync(`${ROOT}/tutorials/02_staggered_did.html`, "utf8");
+  // Expected prefixes are the section heading TEXT (note the typographic
+  // apostrophe in rendered prose vs the straight one in the id): asserting
+  // startsWith proves the excerpt is section-scoped, not the whole-page
+  // fallback searchtools uses when the anchor lookup finds nothing.
+  const cases = [
+    ["#Callaway-Sant'Anna-Estimator", "Callaway-Sant’Anna Estimator"],
+    ["#Group-Time-Effects-ATT(g,t)", "Group-Time Effects ATT(g,t)"],
+    [
+      "#Understanding-Why-TWFE-Fails:-Goodman-Bacon-Decomposition",
+      "Understanding Why TWFE Fails: Goodman-Bacon Decomposition",
+    ],
+    ["#Aggregating-Effects", "Aggregating Effects"],
+  ];
+  for (const [anchor, expectedPrefix] of cases) {
+    try {
+      const text = Search.htmlToText(page, anchor);
+      check(
+        `section-scoped excerpt for ${anchor}`,
+        typeof text === "string" && text.trim().startsWith(expectedPrefix),
+        text ? `got ${JSON.stringify(text.trim().slice(0, 60))}` : "empty result",
+      );
+    } catch (err) {
+      check(`section-scoped excerpt for ${anchor}`, false, `threw ${err}`);
+    }
+  }
+
+  // Legacy digit anchors no longer exist in built pages (headings were
+  // renumbered) but still arrive via old inbound search/deep links; the
+  // wrapper must keep them from throwing. Synthetic page keeps this case
+  // covered independently of the notebook content.
+  const legacyPage =
+    '<div role="main"><section id="3.-Legacy-Digit-Heading"><h2>Legacy Digit Heading</h2>' +
+    "<p>legacy digit section body</p></section></div>";
+  try {
+    const text = Search.htmlToText(legacyPage, "#3.-Legacy-Digit-Heading");
+    check(
+      "excerpt for synthetic legacy digit anchor",
+      typeof text === "string" && text.includes("legacy digit section body"),
+      text ? undefined : "empty result",
+    );
+  } catch (err) {
+    check("excerpt for synthetic legacy digit anchor", false, `threw ${err}`);
+  }
+}
+
+// --- Part 2: legacy fragment redirect ------------------------------------
+{
+  const page = readFileSync(`${ROOT}/tutorials/02_staggered_did.html`, "utf8");
+  const legacyHash = "#3.-Callaway-Sant'Anna-Estimator";
+  const dom = new JSDOM(page, {
+    url: `http://localhost/tutorials/02_staggered_did.html${legacyHash}`,
+    runScripts: "outside-only",
+  });
+  const { window } = dom;
+  window.eval(readFileSync(`${ROOT}/_static/legacy-fragment-redirect.js`, "utf8"));
+  await domLoaded(window);
+  check(
+    "legacy numbered fragment redirects to renamed target",
+    window.location.hash === "#Callaway-Sant'Anna-Estimator",
+    `hash is ${JSON.stringify(window.location.hash)}`,
+  );
+
+  // A hash that still resolves (or matches nothing) must be left alone.
+  const dom2 = new JSDOM(page, {
+    url: "http://localhost/tutorials/02_staggered_did.html#Aggregating-Effects",
+    runScripts: "outside-only",
+  });
+  dom2.window.eval(
+    readFileSync(`${ROOT}/_static/legacy-fragment-redirect.js`, "utf8"),
+  );
+  await domLoaded(dom2.window);
+  check(
+    "valid fragment left untouched",
+    dom2.window.location.hash === "#Aggregating-Effects",
+    `hash is ${JSON.stringify(dom2.window.location.hash)}`,
+  );
+}
+
+// --- Part 3: redirect targets are unambiguous ----------------------------
+// The legacy redirect resolves a stripped fragment with getElementById, so
+// its correctness rests on ids being unique per page. nbsphinx guarantees
+// this today (repeated heading titles - "Background", "Summary" - get an id
+// only on their FIRST occurrence, in the pre-rename state too, so no legacy
+// numbered fragment ever pointed at a later duplicate). Pin the uniqueness
+// so a toolchain change that starts emitting duplicate ids fails here
+// instead of silently making redirects ambiguous.
+{
+  const tutorialsDir = `${ROOT}/tutorials`;
+  let dupPages = 0;
+  for (const file of readdirSync(tutorialsDir).filter((f) => f.endsWith(".html"))) {
+    const ids = [...readFileSync(`${tutorialsDir}/${file}`, "utf8").matchAll(/\sid="([^"]+)"/g)]
+      .map((m) => m[1]);
+    const dupes = ids.filter((id, i) => ids.indexOf(id) !== i);
+    if (dupes.length) {
+      dupPages += 1;
+      console.log(`FAIL: duplicate ids in ${file}: ${[...new Set(dupes)].join(", ")}`);
+    }
+  }
+  check("built tutorial pages have unique ids (unambiguous redirect targets)", dupPages === 0);
+}
+
+if (failures) {
+  console.error(`\n${failures} check(s) failed`);
+  process.exit(1);
+}
+console.log("\nall search-excerpt smoke checks passed");
diff --git a/.github/workflows/docs-tests.yml b/.github/workflows/docs-tests.yml
index d1b4fcd79..84263a18a 100644
--- a/.github/workflows/docs-tests.yml
+++ b/.github/workflows/docs-tests.yml
@@ -20,6 +20,8 @@ on:
       # (Python version, apt packages, post_install docs deps).
       - '.readthedocs.yaml'
       - '.github/workflows/docs-tests.yml'
+      # Search smoke test: script + pinned jsdom manifest/lockfile.
+      - '.github/scripts/**'
   pull_request:
     branches: [main]
     types: [opened, synchronize, reopened, labeled, unlabeled]
@@ -36,6 +38,8 @@ on:
       # (Python version, apt packages, post_install docs deps).
       - '.readthedocs.yaml'
       - '.github/workflows/docs-tests.yml'
+      # Search smoke test: script + pinned jsdom manifest/lockfile.
+      - '.github/scripts/**'
   schedule:
     # Weekly Sunday 6am UTC - smoke test that snippets still execute
     # against current upstream deps (mirrors notebooks.yml schedule).
@@ -136,6 +140,42 @@ jobs:
         # blocking any new warning from sneaking in.
         run: make -C docs html SPHINXOPTS="-W"
 
+      - name: Guard search-anchor invariants in built HTML
+        # Tripwires for the search-excerpt fix (searchtools-css-escape.js):
+        # 1. No digit-leading section ids may reappear in tutorial pages -
+        #    heading numbers were stripped because searchtools.js crashed
+        #    on them before the shim existed, and numberless headings are
+        #    the documented convention.
+        # 2. The CSS.escape shim must be emitted on the built search page.
+        # 3. The shim wraps Search.htmlToText; if a Sphinx upgrade renames
+        #    that hook the shim's feature-guard silently no-ops, so fail
+        #    loudly here and revisit the shim instead.
+        run: |
+          test -d docs/_build/html/tutorials
+          ! grep -rE 'id="[0-9]' docs/_build/html/tutorials/
+          grep -q 'searchtools-css-escape.js' docs/_build/html/search.html
+          grep -q 'htmlToText' docs/_build/html/_static/searchtools.js
+
+      - name: Set up Node for the search smoke test
+        uses: actions/setup-node@249970729cb0ef3589644e2896645e5dc5ba9c38 # v6
+        with:
+          # Current LTS; jsdom's floor is far below this. Pinned so the
+          # smoke test doesn't float with the runner image default.
+          node-version: '24'
+
+      - name: Behavioral smoke test - search excerpts + legacy fragments
+        # Runs the BUILT searchtools.js + shims in jsdom and drives the
+        # exact search-excerpt path (Search.htmlToText) with apostrophe,
+        # parens, colon, and legacy digit anchors, asserting section-scoped
+        # excerpts render without SyntaxError; also asserts the
+        # legacy-fragment redirect rewrites pre-rename numbered hashes.
+        # Catches a Sphinx/theme upgrade changing the hook's signature or
+        # invocation semantics, which the greps above cannot. jsdom is
+        # pinned by the committed .github/scripts/package{,-lock}.json.
+        run: |
+          npm ci --prefix .github/scripts --ignore-scripts
+          node .github/scripts/search_excerpt_smoke.mjs
+
   docs-deps-py39-smoke:
     name: Validate docs deps install on Python 3.9 (floor compat)
     # Skip unrelated label churn: a non-ready-for-ci label add/remove won't run this job.
diff --git a/.gitignore b/.gitignore
index 38c9fefc2..98fe1dc17 100644
--- a/.gitignore
+++ b/.gitignore
@@ -83,6 +83,7 @@ target/
 # Local scripts (not part of package)
 scripts/
 !.claude/scripts/
+!.github/scripts/
 
 # Launch directories (local only)
 launch/
@@ -125,3 +126,9 @@ benchmarks/refresh_2026_07/results/refresh_results_smoke.json
 # the golden JSON each generator produces IS committed (benchmarks/data/*.json).
 # Root-anchored to the generators' own logs so it doesn't hide logs elsewhere.
 /generate_*_golden.log
+
+# Node install tree for the docs search-excerpt smoke test. jsdom is pinned
+# by .github/scripts/package.json + package-lock.json (both COMMITTED - do
+# not ignore manifests repo-wide); `npm ci --prefix .github/scripts` puts
+# node_modules next to them.
+node_modules/
diff --git a/CHANGELOG.md b/CHANGELOG.md
index 2c431a9ba..cfccde141 100644
--- a/CHANGELOG.md
+++ b/CHANGELOG.md
@@ -8,6 +8,18 @@ and this project adheres to [Semantic Versioning](https://semver.org/spec/v2.0.0
 ## [Unreleased]
 
 ### Added
+- **Docs: search-result excerpts fixed and search-as-you-type enabled.** Notebook
+  section anchors containing CSS-special characters (digit-leading `#3.-Fit-Event-Study`,
+  apostrophes, parentheses, colons) crashed Sphinx's search-excerpt renderer
+  (`querySelector` SyntaxError), silently dropping result summaries. A `CSS.escape`
+  shim (`docs/_static/searchtools-css-escape.js`) now makes every anchor a valid
+  selector, numeric prefixes are stripped from 185 tutorial headings across 22
+  notebooks (ordering is conveyed by the grouped tutorials index; prose "Section N"
+  cross-references reworded to heading names, paper-section citations kept verbatim),
+  and the theme's `search_as_you_type` live-filtering overlay is enabled. Page paths
+  are unchanged; the renamed section fragments are kept backward-compatible by a
+  client-side redirect (`docs/_static/legacy-fragment-redirect.js`) that rewrites old
+  `#3.-Fit-Event-Study`-style deep links to their unnumbered targets.
 - **ImputationDiD leave-one-out SE now anchored against Stata `did_imputation, leaveout`
   (no library behavior change).** The Borusyak-Jaravel-Spiess (2024) App. A.9 LOO variance
   (`leave_one_out=True`) has no runnable R reference (R `didimputation` omits LOO), so it
diff --git a/docs/_static/legacy-fragment-redirect.js b/docs/_static/legacy-fragment-redirect.js
new file mode 100644
index 000000000..93be2e93b
--- /dev/null
+++ b/docs/_static/legacy-fragment-redirect.js
@@ -0,0 +1,45 @@
+/* Redirect pre-2026-07 numbered section fragments to their renamed targets.
+ *
+ * Tutorial headings used to carry numeric prefixes ("## 3. Fit Event Study"),
+ * giving nbsphinx anchors like "#3.-Fit-Event-Study". The prefixes were
+ * stripped (the anchors crashed search-excerpt rendering and the numbering
+ * duplicated the grouped tutorials index), which renamed those fragments to
+ * "#Fit-Event-Study". Page paths are unchanged; this shim keeps old deep
+ * links working: when the requested fragment does not exist but the same id
+ * without a leading "N.-" / "N.M.-" prefix does, it replaces the hash (no
+ * extra history entry) so the browser scrolls to the renamed section.
+ *
+ * Repeated heading titles ("Background", "Summary") cannot make this
+ * ambiguous: those duplicates were already unnumbered BEFORE the rename
+ * (no "#N.-Background" fragment ever existed to redirect), and nbsphinx
+ * emits an id only for the first occurrence, so getElementById has exactly
+ * one possible target. The search smoke test pins per-page id uniqueness.
+ */
+(function () {
+  function migrateLegacyFragment() {
+    var hash = window.location.hash;
+    if (!hash || hash.length < 2) return;
+    var id;
+    try {
+      id = decodeURIComponent(hash.slice(1));
+    } catch (err) {
+      return;
+    }
+    if (document.getElementById(id)) return; // fragment still valid
+    var m = id.match(/^\d+(?:\.\d+)*\.-(.+)$/);
+    if (!m) return;
+    var target = document.getElementById(m[1]);
+    if (!target) return;
+    if (window.history && window.history.replaceState) {
+      window.history.replaceState(null, "", "#" + m[1]);
+      if (typeof target.scrollIntoView === "function") target.scrollIntoView();
+    } else {
+      window.location.replace("#" + m[1]);
+    }
+  }
+  if (document.readyState === "loading") {
+    document.addEventListener("DOMContentLoaded", migrateLegacyFragment);
+  } else {
+    migrateLegacyFragment();
+  }
+})();
diff --git a/docs/_static/searchtools-css-escape.js b/docs/_static/searchtools-css-escape.js
new file mode 100644
index 000000000..4fa95975c
--- /dev/null
+++ b/docs/_static/searchtools-css-escape.js
@@ -0,0 +1,46 @@
+/* Make search-result excerpts survive CSS-special anchor ids.
+ *
+ * Sphinx's searchtools.js renders each search-result excerpt by fetching the
+ * target page and running document.querySelector('[role="main"] ' + anchor)
+ * with the raw section anchor (htmlToText). nbsphinx keeps raw heading text
+ * in notebook anchors, so ids like "#Callaway-Sant'Anna-Estimator",
+ * "#Group-Time-Effects-ATT(g,t)" or "#Understanding-Why-TWFE-Fails:-..."
+ * are not valid CSS selectors: querySelector throws SyntaxError and the
+ * excerpt is silently dropped from the results list.
+ *
+ * Wrapping Search.htmlToText and passing the id through CSS.escape() makes
+ * every anchor a valid selector. Already-valid ids escape to themselves, so
+ * this is a no-op for normal pages.
+ *
+ * Load-order note: search.html emits this file BEFORE searchtools.js (custom
+ * js files render with the early script group; searchtools.js comes from the
+ * search page template), so Search does not exist yet at parse time. The
+ * wrap is deferred to DOMContentLoaded, which runs registered-order-first:
+ * this listener is registered before searchtools.js registers its own
+ * DOMContentLoaded query runner, so the wrap is in place before any search
+ * executes.
+ */
+(function () {
+  function wrapHtmlToText() {
+    if (
+      typeof Search === "undefined" ||
+      !Search.htmlToText ||
+      !window.CSS ||
+      !CSS.escape
+    ) {
+      return;
+    }
+    var origHtmlToText = Search.htmlToText.bind(Search);
+    Search.htmlToText = function (htmlString, anchor) {
+      if (anchor && anchor.charAt(0) === "#") {
+        anchor = "#" + CSS.escape(anchor.slice(1));
+      }
+      return origHtmlToText(htmlString, anchor);
+    };
+  }
+  if (document.readyState === "loading") {
+    document.addEventListener("DOMContentLoaded", wrapHtmlToText);
+  } else {
+    wrapHtmlToText();
+  }
+})();
diff --git a/docs/conf.py b/docs/conf.py
index 73408e905..121487285 100644
--- a/docs/conf.py
+++ b/docs/conf.py
@@ -132,6 +132,9 @@
     "navbar_end": ["theme-switcher", "navbar-icon-links"],
     "navigation_depth": 3,
     "show_toc_level": 2,
+    # Live-filtering search overlay. Safe to enable now that search-result
+    # excerpt rendering survives notebook anchors (searchtools-css-escape.js).
+    "search_as_you_type": True,
 }
 
 # -- Options for sphinxext-opengraph -----------------------------------------
@@ -184,6 +187,17 @@
     )
 
 
-# -- Custom CSS --------------------------------------------------------------
+# -- Custom CSS / JS ---------------------------------------------------------
 def setup(app):
     app.add_css_file("custom.css")
+    # CSS.escape()s section anchors so notebook heading ids containing
+    # ' ( ) : etc. don't crash search-result excerpt rendering. NB: on
+    # search.html this file is emitted BEFORE searchtools.js (custom js
+    # renders with the early script group; searchtools comes from the
+    # search page template), so the wrapper defers to DOMContentLoaded -
+    # see the file's header comment before touching load order.
+    app.add_js_file("searchtools-css-escape.js")
+    # Keeps pre-rename numbered section deep links ("#3.-Fit-Event-Study")
+    # working after the 2026-07 heading-number strip: rewrites the hash to
+    # the renamed fragment when the legacy one no longer exists.
+    app.add_js_file("legacy-fragment-redirect.js")
diff --git a/docs/tutorials/01_basic_did.ipynb b/docs/tutorials/01_basic_did.ipynb
index 2bb785838..510d232ee 100644
--- a/docs/tutorials/01_basic_did.ipynb
+++ b/docs/tutorials/01_basic_did.ipynb
@@ -32,7 +32,7 @@
    "cell_type": "markdown",
    "metadata": {},
    "source": [
-    "## 1. Generate Sample Data\n",
+    "## Generate Sample Data\n",
     "\n",
     "The `generate_did_data` function creates synthetic panel data with a known treatment effect, which is useful for learning and testing."
    ]
@@ -73,7 +73,7 @@
    "cell_type": "markdown",
    "metadata": {},
    "source": [
-    "## 2. Basic DiD Estimation\n",
+    "## Basic DiD Estimation\n",
     "\n",
     "The `DifferenceInDifferences` estimator provides an sklearn-like interface with a `fit()` method."
    ]
@@ -133,7 +133,7 @@
    "cell_type": "markdown",
    "metadata": {},
    "source": [
-    "## 3. Using the Formula Interface\n",
+    "## Using the Formula Interface\n",
     "\n",
     "For those familiar with R, `diff-diff` supports a formula interface similar to R's notation."
    ]
@@ -170,7 +170,7 @@
    "cell_type": "markdown",
    "metadata": {},
    "source": [
-    "## 4. Adding Covariates\n",
+    "## Adding Covariates\n",
     "\n",
     "You can include additional control variables to improve precision and reduce bias from observed confounders."
    ]
@@ -215,7 +215,7 @@
    "cell_type": "markdown",
    "metadata": {},
    "source": [
-    "## 5. Fixed Effects\n",
+    "## Fixed Effects\n",
     "\n",
     "Fixed effects control for time-invariant unobserved heterogeneity. `diff-diff` supports two approaches:\n",
     "\n",
@@ -304,7 +304,7 @@
    "cell_type": "markdown",
    "metadata": {},
    "source": [
-    "## 6. Two-Way Fixed Effects (TWFE)\n",
+    "## Two-Way Fixed Effects (TWFE)\n",
     "\n",
     "For panel data, the `TwoWayFixedEffects` estimator automatically includes both unit and time fixed effects using within-transformation."
    ]
@@ -332,7 +332,7 @@
    "cell_type": "markdown",
    "metadata": {},
    "source": [
-    "## 7. Robust Inference\n",
+    "## Robust Inference\n",
     "\n",
     "### Cluster-Robust Standard Errors\n",
     "\n",
@@ -455,7 +455,7 @@
    "cell_type": "markdown",
    "metadata": {},
    "source": [
-    "## 8. Exporting Results\n",
+    "## Exporting Results\n",
     "\n",
     "Results can be exported to various formats for reporting."
    ]
diff --git a/docs/tutorials/02_staggered_did.ipynb b/docs/tutorials/02_staggered_did.ipynb
index 4657608ce..7421fa981 100644
--- a/docs/tutorials/02_staggered_did.ipynb
+++ b/docs/tutorials/02_staggered_did.ipynb
@@ -54,7 +54,7 @@
    "cell_type": "markdown",
    "metadata": {},
    "source": [
-    "## 1. Understanding Staggered Adoption\n",
+    "## Understanding Staggered Adoption\n",
     "\n",
     "In a staggered adoption design, units adopt treatment at different times. We call the period when a unit first receives treatment its **cohort** or **group**."
    ]
@@ -108,7 +108,7 @@
    "cell_type": "markdown",
    "metadata": {},
    "source": [
-    "## 2. The Problem with TWFE in Staggered Settings\n",
+    "## The Problem with TWFE in Staggered Settings\n",
     "\n",
     "Traditional Two-Way Fixed Effects (TWFE) can give biased estimates because:\n",
     "- It uses already-treated units as controls for newly-treated units\n",
@@ -206,7 +206,7 @@
    "cell_type": "markdown",
    "metadata": {},
    "source": [
-    "## 3. Callaway-Sant'Anna Estimator\n",
+    "## Callaway-Sant'Anna Estimator\n",
     "\n",
     "The CS estimator avoids these problems by:\n",
     "1. Computing separate effects for each (group, time) pair: ATT(g,t)\n",
@@ -242,7 +242,7 @@
    "cell_type": "markdown",
    "metadata": {},
    "source": [
-    "## 4. Group-Time Effects ATT(g,t)\n",
+    "## Group-Time Effects ATT(g,t)\n",
     "\n",
     "The CS estimator computes separate effects for each combination of:\n",
     "- **g**: Treatment cohort (when the group was first treated)\n",
@@ -281,7 +281,7 @@
    "cell_type": "markdown",
    "metadata": {},
    "source": [
-    "## 5. Aggregating Effects\n",
+    "## Aggregating Effects\n",
     "\n",
     "We often want to summarize the group-time effects into a single number or event-study style estimates."
    ]
@@ -336,7 +336,7 @@
    "cell_type": "markdown",
    "metadata": {},
    "source": [
-    "## 6. Bootstrap Inference\n",
+    "## Bootstrap Inference\n",
     "\n",
     "With few clusters or when analytical standard errors may be unreliable, the **multiplier bootstrap** provides valid inference. This implements the approach from Callaway & Sant'Anna (2021), perturbing unit-level influence functions.\n",
     "\n",
@@ -412,7 +412,7 @@
    "cell_type": "markdown",
    "metadata": {},
    "source": [
-    "## 7. Visualization\n",
+    "## Visualization\n",
     "\n",
     "Event-study plots are the standard way to visualize DiD results with multiple periods."
    ]
@@ -461,7 +461,7 @@
    "cell_type": "markdown",
    "metadata": {},
    "source": [
-    "## 8. Pre-Treatment Effects and Parallel Trends Testing\n",
+    "## Pre-Treatment Effects and Parallel Trends Testing\n",
     "\n",
     "The Callaway-Sant'Anna estimator can compute **pre-treatment effects** ATT(g,t) for periods before treatment. These should be near zero if parallel trends holds.\n",
     "\n",
@@ -589,8 +589,8 @@
     "- **Important caveat**: A \"passing\" test doesn't prove parallel trends—the test may lack power\n",
     "\n",
     "**When concerned about pre-trends:**\n",
-    "- Add covariates for precision (Section 11)\n",
-    "- Use `control_group=\"not_yet_treated\"` for more data (Section 9)\n",
+    "- Add covariates for precision (see Adding Covariates)\n",
+    "- Use `control_group=\"not_yet_treated\"` for more data (see Different Control Group Options)\n",
     "- Apply Honest DiD sensitivity analysis to bound effects under violations (Tutorial 05)\n",
     "- Assess pre-trends test power using Tutorial 07\n",
     "\n",
@@ -602,7 +602,7 @@
    "cell_type": "markdown",
    "metadata": {},
    "source": [
-    "## 9. Different Control Group Options\n",
+    "## Different Control Group Options\n",
     "\n",
     "The CS estimator supports different control group specifications:\n",
     "- `\"never_treated\"`: Only use units that are never treated\n",
@@ -640,7 +640,7 @@
    "cell_type": "markdown",
    "metadata": {},
    "source": [
-    "## 10. Handling Anticipation Effects\n",
+    "## Handling Anticipation Effects\n",
     "\n",
     "If units start changing behavior before official treatment (anticipation), you can specify the anticipation period."
    ]
@@ -672,7 +672,7 @@
    "cell_type": "markdown",
    "metadata": {},
    "source": [
-    "## 11. Adding Covariates\n",
+    "## Adding Covariates\n",
     "\n",
     "You can include covariates to improve precision through outcome regression or propensity score methods."
    ]
@@ -708,7 +708,7 @@
    "cell_type": "markdown",
    "metadata": {},
    "source": [
-    "## 12. Comparing with MultiPeriodDiD\n",
+    "## Comparing with MultiPeriodDiD\n",
     "\n",
     "For comparison, here's how you would use `MultiPeriodDiD` which estimates period-specific effects. \n",
     "\n",
@@ -768,7 +768,7 @@
    "cell_type": "markdown",
    "metadata": {},
    "source": [
-    "## 13. Sun-Abraham Interaction-Weighted Estimator\n",
+    "## Sun-Abraham Interaction-Weighted Estimator\n",
     "\n",
     "The Sun-Abraham (2021) estimator provides an alternative approach to staggered DiD. While Callaway-Sant'Anna aggregates 2x2 DiD comparisons, Sun-Abraham uses an **interaction-weighted regression** approach:\n",
     "\n",
@@ -835,7 +835,7 @@
    "cell_type": "markdown",
    "metadata": {},
    "source": [
-    "## 14. Comparing CS and SA as a Robustness Check\n",
+    "## Comparing CS and SA as a Robustness Check\n",
     "\n",
     "Running both estimators provides a useful robustness check. When they agree, results are more credible.\n",
     "\n",
diff --git a/docs/tutorials/03_synthetic_did.ipynb b/docs/tutorials/03_synthetic_did.ipynb
index eca4cb2f0..2f68234a0 100644
--- a/docs/tutorials/03_synthetic_did.ipynb
+++ b/docs/tutorials/03_synthetic_did.ipynb
@@ -48,7 +48,7 @@
    "cell_type": "markdown",
    "metadata": {},
    "source": [
-    "## 1. When to Use Synthetic DiD\n",
+    "## When to Use Synthetic DiD\n",
     "\n",
     "Consider SDID when:\n",
     "- You have **few treated units** (1-10 is common)\n",
@@ -166,7 +166,7 @@
    "cell_type": "markdown",
    "metadata": {},
    "source": [
-    "## 2. Basic Synthetic DiD Estimation"
+    "## Basic Synthetic DiD Estimation"
    ]
   },
   {
@@ -197,7 +197,7 @@
    "cell_type": "markdown",
    "metadata": {},
    "source": [
-    "## 3. Comparing SDID with Standard DiD\n",
+    "## Comparing SDID with Standard DiD\n",
     "\n",
     "Let's compare the Synthetic DiD estimate with standard DiD to see the difference."
    ]
@@ -239,7 +239,7 @@
    "cell_type": "markdown",
    "metadata": {},
    "source": [
-    "## 4. Understanding Unit Weights\n",
+    "## Understanding Unit Weights\n",
     "\n",
     "SDID assigns weights to control units based on how well they match the treated units in the pre-treatment period. Units with higher weights are more similar to the treated units."
    ]
@@ -295,7 +295,7 @@
    "cell_type": "markdown",
    "metadata": {},
    "source": [
-    "## 5. Understanding Time Weights\n",
+    "## Understanding Time Weights\n",
     "\n",
     "SDID also computes **time weights** that determine how much each pre-treatment period contributes to the baseline. Periods where treated and control outcomes are more similar get higher weight."
    ]
@@ -335,7 +335,7 @@
    "cell_type": "markdown",
    "metadata": {},
    "source": [
-    "## 6. Pre-treatment Fit\n",
+    "## Pre-treatment Fit\n",
     "\n",
     "A key diagnostic is how well the synthetic control matches the treated units in the pre-treatment period."
    ]
@@ -398,7 +398,7 @@
   {
    "cell_type": "markdown",
    "metadata": {},
-   "source": "## 7. Inference Methods\n\nSDID supports three inference methods:\n\n1. **Placebo** (`variance_method=\"placebo\"`, default): Placebo-based variance using Algorithm 4 from Arkhangelsky et al. (2021). Library default (R's default is bootstrap — we deviate because placebo is unconditionally available on pweight-only survey designs and sidesteps the refit bootstrap slowdown).\n2. **Bootstrap** (`variance_method=\"bootstrap\"`): Paper-faithful pairs bootstrap (Algorithm 2 step 2) — re-estimates ω and λ via Frank-Wolfe on each draw. Matches R's default `synthdid::vcov(method=\"bootstrap\")` behavior. Expect ~5–30× slower per fit than placebo (panel-size dependent).\n3. **Jackknife** (`variance_method=\"jackknife\"`): Algorithm 3 — fixed-weight leave-one-out. Deterministic; no bootstrap replications."
+   "source": "## Inference Methods\n\nSDID supports three inference methods:\n\n1. **Placebo** (`variance_method=\"placebo\"`, default): Placebo-based variance using Algorithm 4 from Arkhangelsky et al. (2021). Library default (R's default is bootstrap — we deviate because placebo is unconditionally available on pweight-only survey designs and sidesteps the refit bootstrap slowdown).\n2. **Bootstrap** (`variance_method=\"bootstrap\"`): Paper-faithful pairs bootstrap (Algorithm 2 step 2) — re-estimates ω and λ via Frank-Wolfe on each draw. Matches R's default `synthdid::vcov(method=\"bootstrap\")` behavior. Expect ~5–30× slower per fit than placebo (panel-size dependent).\n3. **Jackknife** (`variance_method=\"jackknife\"`): Algorithm 3 — fixed-weight leave-one-out. Deterministic; no bootstrap replications."
   },
   {
    "cell_type": "code",
@@ -456,7 +456,7 @@
    "cell_type": "markdown",
    "metadata": {},
    "source": [
-    "## 8. Tuning Regularization\n",
+    "## Tuning Regularization\n",
     "\n",
     "By default, SDID **auto-computes** regularization from the data noise level, matching R's `synthdid` package:\n",
     "\n",
@@ -512,7 +512,7 @@
    "cell_type": "markdown",
    "metadata": {},
    "source": [
-    "## 9. Single Treated Unit Case\n",
+    "## Single Treated Unit Case\n",
     "\n",
     "SDID is particularly useful when you have only **one treated unit** (like the classic synthetic control case)."
    ]
@@ -560,7 +560,7 @@
    "cell_type": "markdown",
    "metadata": {},
    "source": [
-    "## 10. Including Covariates\n",
+    "## Including Covariates\n",
     "\n",
     "You can include covariates to improve the synthetic control match."
    ]
diff --git a/docs/tutorials/04_parallel_trends.ipynb b/docs/tutorials/04_parallel_trends.ipynb
index 4170d4209..0ab85d2f1 100644
--- a/docs/tutorials/04_parallel_trends.ipynb
+++ b/docs/tutorials/04_parallel_trends.ipynb
@@ -53,7 +53,7 @@
    "cell_type": "markdown",
    "metadata": {},
    "source": [
-    "## 1. Create Example Data\n",
+    "## Create Example Data\n",
     "\n",
     "We'll create two datasets:\n",
     "- One where parallel trends **holds**\n",
@@ -105,7 +105,7 @@
    "cell_type": "markdown",
    "metadata": {},
    "source": [
-    "## 2. Visual Inspection\n",
+    "## Visual Inspection\n",
     "\n",
     "The first step is always to **plot the data**. Look for:\n",
     "- Similar slopes in pre-treatment periods\n",
@@ -147,7 +147,7 @@
    "cell_type": "markdown",
    "metadata": {},
    "source": [
-    "## 3. Simple Parallel Trends Test\n",
+    "## Simple Parallel Trends Test\n",
     "\n",
     "The `check_parallel_trends()` function computes and compares the pre-treatment trends."
    ]
@@ -208,7 +208,7 @@
    "cell_type": "markdown",
    "metadata": {},
    "source": [
-    "## 4. Robust Parallel Trends Test (Wasserstein)\n",
+    "## Robust Parallel Trends Test (Wasserstein)\n",
     "\n",
     "The `check_parallel_trends_robust()` function uses the Wasserstein (Earth Mover's) distance to compare the **full distribution** of outcome changes, not just means."
    ]
@@ -300,7 +300,7 @@
    "cell_type": "markdown",
    "metadata": {},
    "source": [
-    "## 5. Equivalence Testing (TOST)\n",
+    "## Equivalence Testing (TOST)\n",
     "\n",
     "Standard hypothesis testing has **low power** to detect parallel trends. A better approach is **equivalence testing** using the Two One-Sided Tests (TOST) procedure.\n",
     "\n",
@@ -362,7 +362,7 @@
    "cell_type": "markdown",
    "metadata": {},
    "source": [
-    "## 6. Placebo Tests\n",
+    "## Placebo Tests\n",
     "\n",
     "Placebo tests check whether we would detect \"effects\" where none should exist. Types of placebo tests:\n",
     "\n",
@@ -500,7 +500,7 @@
    "cell_type": "markdown",
    "metadata": {},
    "source": [
-    "## 7. Comprehensive Diagnostics\n",
+    "## Comprehensive Diagnostics\n",
     "\n",
     "Run all placebo tests at once with `run_all_placebo_tests()`."
    ]
@@ -541,7 +541,7 @@
    "cell_type": "markdown",
    "metadata": {},
    "source": [
-    "## 8. Event Study as a Parallel Trends Check\n",
+    "## Event Study as a Parallel Trends Check\n",
     "\n",
     "An **event study** shows period-by-period effects. Pre-treatment coefficients should be close to zero if parallel trends holds."
    ]
@@ -591,7 +591,7 @@
    "cell_type": "markdown",
    "metadata": {},
    "source": [
-    "## 9. What to Do If Parallel Trends Fails?\n",
+    "## What to Do If Parallel Trends Fails?\n",
     "\n",
     "If parallel trends is violated, consider:\n",
     "\n",
diff --git a/docs/tutorials/05_honest_did.ipynb b/docs/tutorials/05_honest_did.ipynb
index 702ac7d63..d72faebb2 100644
--- a/docs/tutorials/05_honest_did.ipynb
+++ b/docs/tutorials/05_honest_did.ipynb
@@ -55,7 +55,7 @@
    "id": "cell-2",
    "metadata": {},
    "source": [
-    "## 1. Motivation: The Problem with Pre-trend Testing\n",
+    "## Motivation: The Problem with Pre-trend Testing\n",
     "\n",
     "Researchers often test for parallel trends by checking if pre-treatment coefficients are statistically significant. However, this approach has serious problems:\n",
     "\n",
@@ -74,7 +74,7 @@
    "id": "cell-3",
    "metadata": {},
    "source": [
-    "## 2. Generate Example Data\n",
+    "## Generate Example Data\n",
     "\n",
     "We'll create panel data with:\n",
     "- A true treatment effect of 5.0\n",
@@ -146,7 +146,7 @@
    "id": "cell-5",
    "metadata": {},
    "source": [
-    "## 3. Fit Standard Event Study\n",
+    "## Fit Standard Event Study\n",
     "\n",
     "First, let's estimate a standard event study using `MultiPeriodDiD`."
    ]
@@ -197,7 +197,7 @@
    "id": "cell-8",
    "metadata": {},
    "source": [
-    "## 4. Basic Honest DiD: Relative Magnitudes\n",
+    "## Basic Honest DiD: Relative Magnitudes\n",
     "\n",
     "The **relative magnitudes** approach bounds post-treatment violations by M times the maximum observed pre-treatment violation:\n",
     "\n",
@@ -268,7 +268,7 @@
    "id": "cell-12",
    "metadata": {},
    "source": [
-    "## 5. Sensitivity Analysis\n",
+    "## Sensitivity Analysis\n",
     "\n",
     "A key feature of Honest DiD is examining how results change as we allow larger violations. This helps answer: \"How much would parallel trends need to be violated to overturn our conclusions?\""
    ]
@@ -346,7 +346,7 @@
    "id": "cell-17",
    "metadata": {},
    "source": [
-    "## 6. Different Restriction Parameters\n",
+    "## Different Restriction Parameters\n",
     "\n",
     "Let's compare results for different values of M:"
    ]
@@ -375,7 +375,7 @@
    "id": "cell-19",
    "metadata": {},
    "source": [
-    "## 7. Breakdown Value\n",
+    "## Breakdown Value\n",
     "\n",
     "The **breakdown value** is the smallest M where the robust CI first includes zero. It tells us how robust our conclusion is to parallel trends violations."
    ]
@@ -413,7 +413,7 @@
    "id": "cell-21",
    "metadata": {},
    "source": [
-    "## 8. Smoothness Restrictions\n",
+    "## Smoothness Restrictions\n",
     "\n",
     "An alternative approach restricts the **second differences** of the trend violations:\n",
     "\n",
@@ -467,7 +467,7 @@
    "id": "cell-24",
    "metadata": {},
    "source": [
-    "## 9. Using the Convenience Function\n",
+    "## Using the Convenience Function\n",
     "\n",
     "For quick analysis, use `compute_honest_did()`:"
    ]
@@ -494,7 +494,7 @@
    "id": "cell-26",
    "metadata": {},
    "source": [
-    "## 10. Converting Results to DataFrames\n",
+    "## Converting Results to DataFrames\n",
     "\n",
     "Results can be exported for further analysis:"
    ]
diff --git a/docs/tutorials/06_power_analysis.ipynb b/docs/tutorials/06_power_analysis.ipynb
index 31cdae1e6..917bd3e25 100644
--- a/docs/tutorials/06_power_analysis.ipynb
+++ b/docs/tutorials/06_power_analysis.ipynb
@@ -19,7 +19,7 @@
    "id": "cell-2",
    "metadata": {},
    "source": [
-    "## 1. The Power Analysis Problem\n",
+    "## The Power Analysis Problem\n",
     "\n",
     "Before running a DiD study, researchers need to answer key design questions:\n",
     "\n",
@@ -52,7 +52,7 @@
    "id": "cell-4",
    "metadata": {},
    "source": [
-    "## 2. Minimum Detectable Effect (MDE)\n",
+    "## Minimum Detectable Effect (MDE)\n",
     "\n",
     "The MDE is the smallest treatment effect you can detect with your specified power and sample size. It depends on:\n",
     "\n",
@@ -127,7 +127,7 @@
    "id": "cell-9",
    "metadata": {},
    "source": [
-    "## 3. Required Sample Size\n",
+    "## Required Sample Size\n",
     "\n",
     "Given a target effect size to detect, calculate the required sample size."
    ]
@@ -171,7 +171,7 @@
    "id": "cell-12",
    "metadata": {},
    "source": [
-    "## 4. Power Calculation\n",
+    "## Power Calculation\n",
     "\n",
     "Given a specific effect size and sample size, calculate the statistical power."
    ]
@@ -202,7 +202,7 @@
    "id": "cell-14",
    "metadata": {},
    "source": [
-    "## 5. Power Curves\n",
+    "## Power Curves\n",
     "\n",
     "Power curves show how power changes with effect size (or sample size). They're essential for understanding the trade-offs in study design."
    ]
@@ -256,7 +256,7 @@
    "cell_type": "markdown",
    "id": "cell-19",
    "metadata": {},
-   "source": "## 6. Panel Data Considerations\n\nFor panel DiD with multiple time periods, power depends on:\n- Number of pre/post periods (`n_pre`, `n_post`)\n- Within-unit (serial) equicorrelation `rho` — the correlation of a unit's outcomes across time\n\nMore periods improve precision. **Within-unit correlation makes the DiD *easier* to detect** (it *lowers* the MDE): the difference-in-differences cancels the shared within-unit component, so a higher `rho` leaves less residual noise. This is the within-unit equicorrelated case of Burlig, Preonas & Woerman (2020), Eq. 2 — `Var(ATT) = σ²(1/n_T + 1/n_C)(1/m + 1/r)(1 − rho)` — and is the *opposite* of the classic Moulton \"design effect\" intuition (which inflates the variance of a cluster *mean*, not a *difference*)."
+   "source": "## Panel Data Considerations\n\nFor panel DiD with multiple time periods, power depends on:\n- Number of pre/post periods (`n_pre`, `n_post`)\n- Within-unit (serial) equicorrelation `rho` — the correlation of a unit's outcomes across time\n\nMore periods improve precision. **Within-unit correlation makes the DiD *easier* to detect** (it *lowers* the MDE): the difference-in-differences cancels the shared within-unit component, so a higher `rho` leaves less residual noise. This is the within-unit equicorrelated case of Burlig, Preonas & Woerman (2020), Eq. 2 — `Var(ATT) = σ²(1/n_T + 1/n_C)(1/m + 1/r)(1 − rho)` — and is the *opposite* of the classic Moulton \"design effect\" intuition (which inflates the variance of a cluster *mean*, not a *difference*)."
   },
   {
    "cell_type": "code",
@@ -297,7 +297,7 @@
    "id": "cell-22",
    "metadata": {},
    "source": [
-    "## 7. Simulation-Based Power Analysis\n",
+    "## Simulation-Based Power Analysis\n",
     "\n",
     "For complex designs (staggered adoption, synthetic DiD, etc.), analytical formulas may not exist. Use Monte Carlo simulation instead.\n",
     "\n",
@@ -397,7 +397,7 @@
   {
    "cell_type": "markdown",
    "id": "cjpvh2ze7lh",
-   "source": "## 8. Power Analysis for Any Estimator\n\nThe simulation-based approach works with **all 12 supported estimators** — not just basic DiD. An internal registry automatically selects the appropriate data-generating process (DGP) and fit signature for each registered estimator. Just swap in the estimator object and everything else is handled. See the support table below for the full list, and Section 11 for using custom DGPs with unsupported estimators.\n\n### Staggered Adoption Estimators",
+   "source": "## Power Analysis for Any Estimator\n\nThe simulation-based approach works with **all 12 supported estimators** — not just basic DiD. An internal registry automatically selects the appropriate data-generating process (DGP) and fit signature for each registered estimator. Just swap in the estimator object and everything else is handled. See the support table below for the full list, and the Custom Data Generators section for using custom DGPs with unsupported estimators.\n\n### Staggered Adoption Estimators",
    "metadata": {}
   },
   {
@@ -439,7 +439,7 @@
   {
    "cell_type": "markdown",
    "id": "6kb8ovmue4m",
-   "source": "### Supported Estimators\n\nThe following 12 estimators are supported by the simulation power analysis registry. Each is automatically paired with the correct data-generating process:\n\n| DGP Family | Estimators | Min N |\n|---|---|---|\n| **Basic DiD** (`generate_did_data`) | DifferenceInDifferences, TwoWayFixedEffects, MultiPeriodDiD | 20 |\n| **Staggered** (`generate_staggered_data`) | CallawaySantAnna, SunAbraham, ImputationDiD, TwoStageDiD, StackedDiD, EfficientDiD | 40 |\n| **Factor Model** (`generate_factor_data`) | TROP, SyntheticDiD | 30 |\n| **Triple Difference** (`generate_ddd_data` / `generate_ddd_panel_data`) | TripleDifference | 16* |\n\n\\* DDD routes by `n_periods`: for `n_periods ≤ 2` the cross-sectional `generate_ddd_data` rounds effective N to `max(2, n_units // 8) * 8` (minimum 16) and `simulate_sample_size()` starts its bisection at the registry floor of 64 (an explicit `n_range` can still go down to the 16-unit realizability floor); for `n_periods > 2` the panel `generate_ddd_panel_data` maps `n_units` directly (use `TripleDifference(cluster=\"unit\")`).\n\n> **Note:** `ContinuousDiD` is not in the registry because continuous/dose-response treatments require a different DGP structure. `BaconDecomposition` and `HonestDiD` are diagnostic/sensitivity tools rather than treatment effect estimators. For unsupported estimators, you can pass a custom `data_generator` and `result_extractor` (see Section 11).",
+   "source": "### Supported Estimators\n\nThe following 12 estimators are supported by the simulation power analysis registry. Each is automatically paired with the correct data-generating process:\n\n| DGP Family | Estimators | Min N |\n|---|---|---|\n| **Basic DiD** (`generate_did_data`) | DifferenceInDifferences, TwoWayFixedEffects, MultiPeriodDiD | 20 |\n| **Staggered** (`generate_staggered_data`) | CallawaySantAnna, SunAbraham, ImputationDiD, TwoStageDiD, StackedDiD, EfficientDiD | 40 |\n| **Factor Model** (`generate_factor_data`) | TROP, SyntheticDiD | 30 |\n| **Triple Difference** (`generate_ddd_data` / `generate_ddd_panel_data`) | TripleDifference | 16* |\n\n\\* DDD routes by `n_periods`: for `n_periods ≤ 2` the cross-sectional `generate_ddd_data` rounds effective N to `max(2, n_units // 8) * 8` (minimum 16) and `simulate_sample_size()` starts its bisection at the registry floor of 64 (an explicit `n_range` can still go down to the 16-unit realizability floor); for `n_periods > 2` the panel `generate_ddd_panel_data` maps `n_units` directly (use `TripleDifference(cluster=\"unit\")`).\n\n> **Note:** `ContinuousDiD` is not in the registry because continuous/dose-response treatments require a different DGP structure. `BaconDecomposition` and `HonestDiD` are diagnostic/sensitivity tools rather than treatment effect estimators. For unsupported estimators, you can pass a custom `data_generator` and `result_extractor` (see Custom Data Generators).",
    "metadata": {}
   },
   {
@@ -459,7 +459,7 @@
   {
    "cell_type": "markdown",
    "id": "kqw6y4du5u",
-   "source": "## 9. Finding MDE via Simulation\n\nThe analytical `PowerAnalysis.mde()` works for basic DiD, but for complex estimators there is no closed-form formula. `simulate_mde()` uses **bisection search** to find the minimum detectable effect: it repeatedly calls `simulate_power()` at different effect sizes, narrowing the bracket until it finds the smallest effect that achieves the target power.",
+   "source": "## Finding MDE via Simulation\n\nThe analytical `PowerAnalysis.mde()` works for basic DiD, but for complex estimators there is no closed-form formula. `simulate_mde()` uses **bisection search** to find the minimum detectable effect: it repeatedly calls `simulate_power()` at different effect sizes, narrowing the bracket until it finds the smallest effect that achieves the target power.",
    "metadata": {}
   },
   {
@@ -507,7 +507,7 @@
   {
    "cell_type": "markdown",
    "id": "qrfhvtizg58",
-   "source": "## 10. Finding Required Sample Size via Simulation\n\n`simulate_sample_size()` uses bisection search over `n_units` to find the smallest sample size that achieves the target power for a given effect size.",
+   "source": "## Finding Required Sample Size via Simulation\n\n`simulate_sample_size()` uses bisection search over `n_units` to find the smallest sample size that achieves the target power for a given effect size.",
    "metadata": {}
   },
   {
@@ -555,7 +555,7 @@
   {
    "cell_type": "markdown",
    "id": "usp5iwyacop",
-   "source": "## 11. Custom Data Generators\n\nThe default DGPs cover common designs, but you can customize them in two ways:\n1. **Tweak the default DGP** with `data_generator_kwargs` (e.g., add multiple treatment cohorts)\n2. **Supply a fully custom DGP** with `data_generator`\n\n### Tweaking the Default DGP\n\nPass additional keyword arguments to the registry's DGP via `data_generator_kwargs`. For example, the default staggered DGP generates a single treatment cohort — here we create a multi-cohort design:\n\n> **Note:** Some keys are *protected* and cannot be overridden via `data_generator_kwargs` because they are controlled by the simulation function itself: `treatment_effect`, `noise_sd`, `n_units`, `n_periods`, `treatment_fraction`, `treatment_period`, `n_pre`, `n_post`.",
+   "source": "## Custom Data Generators\n\nThe default DGPs cover common designs, but you can customize them in two ways:\n1. **Tweak the default DGP** with `data_generator_kwargs` (e.g., add multiple treatment cohorts)\n2. **Supply a fully custom DGP** with `data_generator`\n\n### Tweaking the Default DGP\n\nPass additional keyword arguments to the registry's DGP via `data_generator_kwargs`. For example, the default staggered DGP generates a single treatment cohort — here we create a multi-cohort design:\n\n> **Note:** Some keys are *protected* and cannot be overridden via `data_generator_kwargs` because they are controlled by the simulation function itself: `treatment_effect`, `noise_sd`, `n_units`, `n_periods`, `treatment_fraction`, `treatment_period`, `n_pre`, `n_post`.",
    "metadata": {}
   },
   {
@@ -584,7 +584,7 @@
    "cell_type": "markdown",
    "id": "cell-28",
    "metadata": {},
-   "source": "## 12. Convenience Functions\n\nFor quick calculations, use the convenience functions:"
+   "source": "## Convenience Functions\n\nFor quick calculations, use the convenience functions:"
   },
   {
    "cell_type": "code",
@@ -610,7 +610,7 @@
    "cell_type": "markdown",
    "id": "cell-30",
    "metadata": {},
-   "source": "## 13. Practical Recommendations\n\n### Estimating Sigma (Residual SD)\n\nThe residual standard deviation is crucial for power calculations. Options:\n\n1. **Pilot data**: Fit a model on historical data and get residual SD\n2. **Literature**: Find similar studies and use their reported SDs\n3. **Domain knowledge**: Expert judgment about outcome variability\n4. **Sensitivity analysis**: Calculate power for a range of sigma values"
+   "source": "## Practical Recommendations\n\n### Estimating Sigma (Residual SD)\n\nThe residual standard deviation is crucial for power calculations. Options:\n\n1. **Pilot data**: Fit a model on historical data and get residual SD\n2. **Literature**: Find similar studies and use their reported SDs\n3. **Domain knowledge**: Expert judgment about outcome variability\n4. **Sensitivity analysis**: Calculate power for a range of sigma values"
   },
   {
    "cell_type": "code",
diff --git a/docs/tutorials/07_pretrends_power.ipynb b/docs/tutorials/07_pretrends_power.ipynb
index 6654afc67..0bd1639c9 100644
--- a/docs/tutorials/07_pretrends_power.ipynb
+++ b/docs/tutorials/07_pretrends_power.ipynb
@@ -55,7 +55,7 @@
    "id": "cell-2",
    "metadata": {},
    "source": [
-    "## 1. Motivation: The Problem with Pre-Trends Tests\n",
+    "## Motivation: The Problem with Pre-Trends Tests\n",
     "\n",
     "Standard practice in DiD analysis is to test for parallel trends by checking if pre-treatment coefficients are jointly zero. However, this approach has a fundamental problem:\n",
     "\n",
@@ -79,7 +79,7 @@
    "id": "cell-3",
    "metadata": {},
    "source": [
-    "## 2. Generate Example Data\n",
+    "## Generate Example Data\n",
     "\n",
     "We'll create panel data suitable for event study analysis with multiple pre-treatment periods."
    ]
@@ -120,7 +120,7 @@
    "id": "cell-5",
    "metadata": {},
    "source": [
-    "## 3. Fit Event Study\n",
+    "## Fit Event Study\n",
     "\n",
     "First, we estimate a standard event study to get the pre-period coefficients and their variance-covariance matrix."
    ]
@@ -167,7 +167,7 @@
    "id": "cell-9",
    "metadata": {},
    "source": [
-    "## 4. Basic Pre-Trends Power Analysis\n",
+    "## Basic Pre-Trends Power Analysis\n",
     "\n",
     "The `PreTrendsPower` class computes the power of the pre-trends test to detect violations of different magnitudes."
    ]
@@ -224,7 +224,7 @@
    "id": "cell-13",
    "metadata": {},
    "source": [
-    "## 5. Power at Specific Violation Magnitudes\n",
+    "## Power at Specific Violation Magnitudes\n",
     "\n",
     "You can compute the power to detect a specific violation magnitude:"
    ]
@@ -253,7 +253,7 @@
    "id": "cell-15",
    "metadata": {},
    "source": [
-    "## 6. Power Curves\n",
+    "## Power Curves\n",
     "\n",
     "A **power curve** shows how the power to detect violations changes with violation magnitude. This is the most useful visualization for understanding your test's informativeness."
    ]
@@ -319,7 +319,7 @@
    "id": "cell-19",
    "metadata": {},
    "source": [
-    "## 7. Different Violation Types\n",
+    "## Different Violation Types\n",
     "\n",
     "Pre-trends violations can take different forms. The `violation_type` parameter specifies the pattern:\n",
     "\n",
@@ -419,7 +419,7 @@
    "id": "cell-24",
    "metadata": {},
    "source": [
-    "## 8. Integration with Honest DiD\n",
+    "## Integration with Honest DiD\n",
     "\n",
     "Pre-trends power analysis connects naturally with **Honest DiD** (Rambachan & Roth 2023). The workflow:\n",
     "\n",
@@ -483,7 +483,7 @@
    "id": "cell-27",
    "metadata": {},
    "source": [
-    "## 9. Convenience Functions\n",
+    "## Convenience Functions\n",
     "\n",
     "For quick calculations, use the convenience functions:"
    ]
@@ -509,7 +509,7 @@
    "id": "cell-29",
    "metadata": {},
    "source": [
-    "## 10. Working with Real Event Studies\n",
+    "## Working with Real Event Studies\n",
     "\n",
     "In practice, you'll apply pre-trends power analysis to your actual event study estimates. Here's the typical workflow:"
    ]
@@ -527,7 +527,7 @@
    "id": "cell-31",
    "metadata": {},
    "source": [
-    "## 11. Exporting Results\n",
+    "## Exporting Results\n",
     "\n",
     "Results can be exported to DataFrames for further analysis or reporting:"
    ]
diff --git a/docs/tutorials/09_real_world_examples.ipynb b/docs/tutorials/09_real_world_examples.ipynb
index 8c5532171..93766223e 100644
--- a/docs/tutorials/09_real_world_examples.ipynb
+++ b/docs/tutorials/09_real_world_examples.ipynb
@@ -72,7 +72,7 @@
    "source": [
     "---\n",
     "\n",
-    "## 1. Card & Krueger (1994): Minimum Wage and Employment\n",
+    "## Card & Krueger (1994): Minimum Wage and Employment\n",
     "\n",
     "### Background\n",
     "\n",
@@ -313,7 +313,7 @@
    "source": [
     "---\n",
     "\n",
-    "## 2. Castle Doctrine Laws: Staggered Adoption\n",
+    "## Castle Doctrine Laws: Staggered Adoption\n",
     "\n",
     "### Background\n",
     "\n",
@@ -606,7 +606,7 @@
    "source": [
     "---\n",
     "\n",
-    "## 3. Unilateral Divorce Laws: Long Panel with Staggered Adoption\n",
+    "## Unilateral Divorce Laws: Long Panel with Staggered Adoption\n",
     "\n",
     "### Background\n",
     "\n",
diff --git a/docs/tutorials/10_trop.ipynb b/docs/tutorials/10_trop.ipynb
index 873e9cd86..d7e62f177 100644
--- a/docs/tutorials/10_trop.ipynb
+++ b/docs/tutorials/10_trop.ipynb
@@ -29,7 +29,7 @@
    "cell_type": "markdown",
    "metadata": {},
    "source": [
-    "## 1. When to Use TROP\n",
+    "## When to Use TROP\n",
     "\n",
     "Consider TROP when:\n",
     "- You suspect **factor structure** in the data (e.g., economic cycles, regional shocks)\n",
@@ -124,7 +124,7 @@
    "cell_type": "markdown",
    "metadata": {},
    "source": [
-    "## 2. Basic TROP Estimation\n",
+    "## Basic TROP Estimation\n",
     "\n",
     "TROP uses leave-one-out cross-validation (LOOCV) to select three tuning parameters:\n",
     "- **\u03bb_time**: Time weight decay (higher = focus on periods near treatment)\n",
@@ -190,7 +190,7 @@
    "cell_type": "markdown",
    "metadata": {},
    "source": [
-    "## 3. Understanding the Tuning Parameters\n",
+    "## Understanding the Tuning Parameters\n",
     "\n",
     "The three tuning parameters control different aspects of the estimation:\n",
     "\n",
@@ -253,7 +253,7 @@
    "cell_type": "markdown",
    "metadata": {},
    "source": [
-    "## 4. Examining the Factor Structure\n",
+    "## Examining the Factor Structure\n",
     "\n",
     "TROP estimates a low-rank factor matrix L that captures interactive fixed effects. We can examine this structure."
    ]
@@ -309,7 +309,7 @@
    "cell_type": "markdown",
    "metadata": {},
    "source": [
-    "## 5. Examining Unit and Time Effects\n",
+    "## Examining Unit and Time Effects\n",
     "\n",
     "TROP also estimates traditional unit and time fixed effects (\u03b1_i and \u03b2_t)."
    ]
@@ -373,7 +373,7 @@
    "cell_type": "markdown",
    "metadata": {},
    "source": [
-    "## 6. Comparing TROP vs SDID\n",
+    "## Comparing TROP vs SDID\n",
     "\n",
     "Let's compare TROP with Synthetic DiD to see the benefit of factor adjustment when the DGP has factor structure."
    ]
@@ -438,7 +438,7 @@
    "cell_type": "markdown",
    "metadata": {},
    "source": [
-    "## 7. Monte Carlo Comparison\n",
+    "## Monte Carlo Comparison\n",
     "\n",
     "Let's run a small Monte Carlo simulation to compare TROP and SDID under the factor DGP."
    ]
@@ -552,7 +552,7 @@
    "cell_type": "markdown",
    "metadata": {},
    "source": [
-    "## 8. Using the Convenience Function\n",
+    "## Using the Convenience Function\n",
     "\n",
     "For quick estimation, you can use the `trop()` convenience function."
    ]
@@ -587,7 +587,7 @@
   {
    "cell_type": "markdown",
    "metadata": {},
-   "source": "## 9. Variance Estimation\n\nTROP uses **unit-level stratified block bootstrap** for variance estimation, as specified in Algorithm 3 of Athey et al. (2025). Control and treated units are sampled separately to preserve the treatment ratio. The number of bootstrap replications is controlled by the `n_bootstrap` parameter (default: 200)."
+   "source": "## Variance Estimation\n\nTROP uses **unit-level stratified block bootstrap** for variance estimation, as specified in Algorithm 3 of Athey et al. (2025). Control and treated units are sampled separately to preserve the treatment ratio. The number of bootstrap replications is controlled by the `n_bootstrap` parameter (default: 200)."
   },
   {
    "cell_type": "code",
@@ -606,946 +606,24 @@
   {
    "cell_type": "markdown",
    "source": [
-    "#",
-    "#",
-    " ",
-    "1",
-    "0",
-    ".",
-    " ",
-    "E",
-    "s",
-    "t",
-    "i",
-    "m",
-    "a",
-    "t",
-    "i",
-    "o",
-    "n",
-    " ",
-    "M",
-    "e",
-    "t",
-    "h",
-    "o",
-    "d",
-    "s",
-    ":",
-    " ",
-    "L",
-    "o",
-    "c",
-    "a",
-    "l",
-    " ",
-    "v",
-    "s",
-    " ",
-    "G",
-    "l",
-    "o",
-    "b",
-    "a",
-    "l",
+    "## Estimation Methods: Local vs Global\n",
     "\n",
+    "TROP supports two estimation methods via the `method` parameter:\n",
     "\n",
-    "T",
-    "R",
-    "O",
-    "P",
-    " ",
-    "s",
-    "u",
-    "p",
-    "p",
-    "o",
-    "r",
-    "t",
-    "s",
-    " ",
-    "t",
-    "w",
-    "o",
-    " ",
-    "e",
-    "s",
-    "t",
-    "i",
-    "m",
-    "a",
-    "t",
-    "i",
-    "o",
-    "n",
-    " ",
-    "m",
-    "e",
-    "t",
-    "h",
-    "o",
-    "d",
-    "s",
-    " ",
-    "v",
-    "i",
-    "a",
-    " ",
-    "t",
-    "h",
-    "e",
-    " ",
-    "`",
-    "m",
-    "e",
-    "t",
-    "h",
-    "o",
-    "d",
-    "`",
-    " ",
-    "p",
-    "a",
-    "r",
-    "a",
-    "m",
-    "e",
-    "t",
-    "e",
-    "r",
-    ":",
+    "**Local Method** (`method='local'`, default):\n",
+    "- Follows Algorithm 2 from the paper\n",
+    "- Computes observation-specific weights for each treated observation\n",
+    "- Fits a model per treated observation, then averages the individual effects\n",
+    "- More flexible, allows for heterogeneous treatment effects\n",
+    "- Computationally intensive (N_treated optimizations)\n",
     "\n",
+    "**Global Method** (`method='global'`):\n",
+    "- Fits a single model on control data using (1-W) masked weights (per paper Eq. 2)\n",
+    "- Extracts per-observation treatment effects as post-hoc residuals: \u03c4_it = Y_it - \u03bc - \u03b1_i - \u03b2_t - L_it\n",
+    "- ATT = mean(\u03c4_it) over treated observations\n",
+    "- Faster (single optimization) with global weights\n",
     "\n",
-    "*",
-    "*",
-    "L",
-    "o",
-    "c",
-    "a",
-    "l",
-    " ",
-    "M",
-    "e",
-    "t",
-    "h",
-    "o",
-    "d",
-    "*",
-    "*",
-    " ",
-    "(",
-    "`",
-    "m",
-    "e",
-    "t",
-    "h",
-    "o",
-    "d",
-    "=",
-    "'",
-    "l",
-    "o",
-    "c",
-    "a",
-    "l",
-    "'",
-    "`",
-    ",",
-    " ",
-    "d",
-    "e",
-    "f",
-    "a",
-    "u",
-    "l",
-    "t",
-    ")",
-    ":",
-    "\n",
-    "-",
-    " ",
-    "F",
-    "o",
-    "l",
-    "l",
-    "o",
-    "w",
-    "s",
-    " ",
-    "A",
-    "l",
-    "g",
-    "o",
-    "r",
-    "i",
-    "t",
-    "h",
-    "m",
-    " ",
-    "2",
-    " ",
-    "f",
-    "r",
-    "o",
-    "m",
-    " ",
-    "t",
-    "h",
-    "e",
-    " ",
-    "p",
-    "a",
-    "p",
-    "e",
-    "r",
-    "\n",
-    "-",
-    " ",
-    "C",
-    "o",
-    "m",
-    "p",
-    "u",
-    "t",
-    "e",
-    "s",
-    " ",
-    "o",
-    "b",
-    "s",
-    "e",
-    "r",
-    "v",
-    "a",
-    "t",
-    "i",
-    "o",
-    "n",
-    "-",
-    "s",
-    "p",
-    "e",
-    "c",
-    "i",
-    "f",
-    "i",
-    "c",
-    " ",
-    "w",
-    "e",
-    "i",
-    "g",
-    "h",
-    "t",
-    "s",
-    " ",
-    "f",
-    "o",
-    "r",
-    " ",
-    "e",
-    "a",
-    "c",
-    "h",
-    " ",
-    "t",
-    "r",
-    "e",
-    "a",
-    "t",
-    "e",
-    "d",
-    " ",
-    "o",
-    "b",
-    "s",
-    "e",
-    "r",
-    "v",
-    "a",
-    "t",
-    "i",
-    "o",
-    "n",
-    "\n",
-    "-",
-    " ",
-    "F",
-    "i",
-    "t",
-    "s",
-    " ",
-    "a",
-    " ",
-    "m",
-    "o",
-    "d",
-    "e",
-    "l",
-    " ",
-    "p",
-    "e",
-    "r",
-    " ",
-    "t",
-    "r",
-    "e",
-    "a",
-    "t",
-    "e",
-    "d",
-    " ",
-    "o",
-    "b",
-    "s",
-    "e",
-    "r",
-    "v",
-    "a",
-    "t",
-    "i",
-    "o",
-    "n",
-    ",",
-    " ",
-    "t",
-    "h",
-    "e",
-    "n",
-    " ",
-    "a",
-    "v",
-    "e",
-    "r",
-    "a",
-    "g",
-    "e",
-    "s",
-    " ",
-    "t",
-    "h",
-    "e",
-    " ",
-    "i",
-    "n",
-    "d",
-    "i",
-    "v",
-    "i",
-    "d",
-    "u",
-    "a",
-    "l",
-    " ",
-    "e",
-    "f",
-    "f",
-    "e",
-    "c",
-    "t",
-    "s",
-    "\n",
-    "-",
-    " ",
-    "M",
-    "o",
-    "r",
-    "e",
-    " ",
-    "f",
-    "l",
-    "e",
-    "x",
-    "i",
-    "b",
-    "l",
-    "e",
-    ",",
-    " ",
-    "a",
-    "l",
-    "l",
-    "o",
-    "w",
-    "s",
-    " ",
-    "f",
-    "o",
-    "r",
-    " ",
-    "h",
-    "e",
-    "t",
-    "e",
-    "r",
-    "o",
-    "g",
-    "e",
-    "n",
-    "e",
-    "o",
-    "u",
-    "s",
-    " ",
-    "t",
-    "r",
-    "e",
-    "a",
-    "t",
-    "m",
-    "e",
-    "n",
-    "t",
-    " ",
-    "e",
-    "f",
-    "f",
-    "e",
-    "c",
-    "t",
-    "s",
-    "\n",
-    "-",
-    " ",
-    "C",
-    "o",
-    "m",
-    "p",
-    "u",
-    "t",
-    "a",
-    "t",
-    "i",
-    "o",
-    "n",
-    "a",
-    "l",
-    "l",
-    "y",
-    " ",
-    "i",
-    "n",
-    "t",
-    "e",
-    "n",
-    "s",
-    "i",
-    "v",
-    "e",
-    " ",
-    "(",
-    "N",
-    "_",
-    "t",
-    "r",
-    "e",
-    "a",
-    "t",
-    "e",
-    "d",
-    " ",
-    "o",
-    "p",
-    "t",
-    "i",
-    "m",
-    "i",
-    "z",
-    "a",
-    "t",
-    "i",
-    "o",
-    "n",
-    "s",
-    ")",
-    "\n",
-    "\n",
-    "*",
-    "*",
-    "G",
-    "l",
-    "o",
-    "b",
-    "a",
-    "l",
-    " ",
-    "M",
-    "e",
-    "t",
-    "h",
-    "o",
-    "d",
-    "*",
-    "*",
-    " ",
-    "(",
-    "`",
-    "m",
-    "e",
-    "t",
-    "h",
-    "o",
-    "d",
-    "=",
-    "'",
-    "g",
-    "l",
-    "o",
-    "b",
-    "a",
-    "l",
-    "'",
-    "`",
-    ")",
-    ":",
-    "\n",
-    "-",
-    " ",
-    "F",
-    "i",
-    "t",
-    "s",
-    " ",
-    "a",
-    " ",
-    "s",
-    "i",
-    "n",
-    "g",
-    "l",
-    "e",
-    " ",
-    "m",
-    "o",
-    "d",
-    "e",
-    "l",
-    " ",
-    "o",
-    "n",
-    " ",
-    "c",
-    "o",
-    "n",
-    "t",
-    "r",
-    "o",
-    "l",
-    " ",
-    "d",
-    "a",
-    "t",
-    "a",
-    " ",
-    "u",
-    "s",
-    "i",
-    "n",
-    "g",
-    " ",
-    "(",
-    "1",
-    "-",
-    "W",
-    ")",
-    " ",
-    "m",
-    "a",
-    "s",
-    "k",
-    "e",
-    "d",
-    " ",
-    "w",
-    "e",
-    "i",
-    "g",
-    "h",
-    "t",
-    "s",
-    " ",
-    "(",
-    "p",
-    "e",
-    "r",
-    " ",
-    "p",
-    "a",
-    "p",
-    "e",
-    "r",
-    " ",
-    "E",
-    "q",
-    ".",
-    " ",
-    "2",
-    ")",
-    "\n",
-    "-",
-    " ",
-    "E",
-    "x",
-    "t",
-    "r",
-    "a",
-    "c",
-    "t",
-    "s",
-    " ",
-    "p",
-    "e",
-    "r",
-    "-",
-    "o",
-    "b",
-    "s",
-    "e",
-    "r",
-    "v",
-    "a",
-    "t",
-    "i",
-    "o",
-    "n",
-    " ",
-    "t",
-    "r",
-    "e",
-    "a",
-    "t",
-    "m",
-    "e",
-    "n",
-    "t",
-    " ",
-    "e",
-    "f",
-    "f",
-    "e",
-    "c",
-    "t",
-    "s",
-    " ",
-    "a",
-    "s",
-    " ",
-    "p",
-    "o",
-    "s",
-    "t",
-    "-",
-    "h",
-    "o",
-    "c",
-    " ",
-    "r",
-    "e",
-    "s",
-    "i",
-    "d",
-    "u",
-    "a",
-    "l",
-    "s",
-    ":",
-    " ",
-    "\u03c4",
-    "_",
-    "i",
-    "t",
-    " ",
-    "=",
-    " ",
-    "Y",
-    "_",
-    "i",
-    "t",
-    " ",
-    "-",
-    " ",
-    "\u03bc",
-    " ",
-    "-",
-    " ",
-    "\u03b1",
-    "_",
-    "i",
-    " ",
-    "-",
-    " ",
-    "\u03b2",
-    "_",
-    "t",
-    " ",
-    "-",
-    " ",
-    "L",
-    "_",
-    "i",
-    "t",
-    "\n",
-    "-",
-    " ",
-    "A",
-    "T",
-    "T",
-    " ",
-    "=",
-    " ",
-    "m",
-    "e",
-    "a",
-    "n",
-    "(",
-    "\u03c4",
-    "_",
-    "i",
-    "t",
-    ")",
-    " ",
-    "o",
-    "v",
-    "e",
-    "r",
-    " ",
-    "t",
-    "r",
-    "e",
-    "a",
-    "t",
-    "e",
-    "d",
-    " ",
-    "o",
-    "b",
-    "s",
-    "e",
-    "r",
-    "v",
-    "a",
-    "t",
-    "i",
-    "o",
-    "n",
-    "s",
-    "\n",
-    "-",
-    " ",
-    "F",
-    "a",
-    "s",
-    "t",
-    "e",
-    "r",
-    " ",
-    "(",
-    "s",
-    "i",
-    "n",
-    "g",
-    "l",
-    "e",
-    " ",
-    "o",
-    "p",
-    "t",
-    "i",
-    "m",
-    "i",
-    "z",
-    "a",
-    "t",
-    "i",
-    "o",
-    "n",
-    ")",
-    " ",
-    "w",
-    "i",
-    "t",
-    "h",
-    " ",
-    "g",
-    "l",
-    "o",
-    "b",
-    "a",
-    "l",
-    " ",
-    "w",
-    "e",
-    "i",
-    "g",
-    "h",
-    "t",
-    "s",
-    "\n",
-    "\n",
-    "N",
-    "o",
-    "t",
-    "e",
-    ":",
-    " ",
-    "`",
-    "m",
-    "e",
-    "t",
-    "h",
-    "o",
-    "d",
-    "=",
-    "'",
-    "t",
-    "w",
-    "o",
-    "s",
-    "t",
-    "e",
-    "p",
-    "'",
-    "`",
-    " ",
-    "i",
-    "s",
-    " ",
-    "a",
-    " ",
-    "d",
-    "e",
-    "p",
-    "r",
-    "e",
-    "c",
-    "a",
-    "t",
-    "e",
-    "d",
-    " ",
-    "a",
-    "l",
-    "i",
-    "a",
-    "s",
-    " ",
-    "f",
-    "o",
-    "r",
-    " ",
-    "`",
-    "m",
-    "e",
-    "t",
-    "h",
-    "o",
-    "d",
-    "=",
-    "'",
-    "l",
-    "o",
-    "c",
-    "a",
-    "l",
-    "'",
-    "`",
-    ",",
-    " ",
-    "a",
-    "n",
-    "d",
-    " ",
-    "`",
-    "m",
-    "e",
-    "t",
-    "h",
-    "o",
-    "d",
-    "=",
-    "'",
-    "j",
-    "o",
-    "i",
-    "n",
-    "t",
-    "'",
-    "`",
-    " ",
-    "i",
-    "s",
-    " ",
-    "a",
-    " ",
-    "d",
-    "e",
-    "p",
-    "r",
-    "e",
-    "c",
-    "a",
-    "t",
-    "e",
-    "d",
-    " ",
-    "a",
-    "l",
-    "i",
-    "a",
-    "s",
-    " ",
-    "f",
-    "o",
-    "r",
-    " ",
-    "`",
-    "m",
-    "e",
-    "t",
-    "h",
-    "o",
-    "d",
-    "=",
-    "'",
-    "g",
-    "l",
-    "o",
-    "b",
-    "a",
-    "l",
-    "'",
-    "`",
-    ".",
-    " ",
-    "B",
-    "o",
-    "t",
-    "h",
-    " ",
-    "w",
-    "i",
-    "l",
-    "l",
-    " ",
-    "b",
-    "e",
-    " ",
-    "r",
-    "e",
-    "m",
-    "o",
-    "v",
-    "e",
-    "d",
-    " ",
-    "i",
-    "n",
-    " ",
-    "v",
-    "3",
-    ".",
-    "0",
-    "."
+    "Note: `method='twostep'` is a deprecated alias for `method='local'`, and `method='joint'` is a deprecated alias for `method='global'`. Both will be removed in v3.0."
    ],
    "metadata": {}
   },
@@ -1553,7 +631,7 @@
    "cell_type": "markdown",
    "metadata": {},
    "source": [
-    "## 10. Results Export\n",
+    "## Results Export\n",
     "\n",
     "TROP results can be easily exported to different formats."
    ]
@@ -1576,9 +654,7 @@
   },
   {
    "cell_type": "markdown",
-   "execution_count": null,
    "metadata": {},
-   "outputs": [],
    "source": "## Summary\n\nKey takeaways for TROP:\n\n1. **Best use cases**: Factor confounding, unobserved time-varying confounders with interactive effects\n2. **Factor estimation**: Nuclear norm regularization with LOOCV for tuning\n3. **Three tuning parameters**: \u03bb_time, \u03bb_unit, \u03bb_nn selected automatically via LOOCV\n4. **Unit weights**: Exponential distance-based weighting of control units, where distance is computed as RMS outcome difference on control periods excluding the target period\n5. **Time weights**: Exponential decay weighting of pre-treatment periods\n6. **Weights**: Importance weights controlling relative contribution of observations (higher = more relevant)\n7. **Estimation methods**:\n   - `method='local'` (default): Per-observation estimation, allows heterogeneous effects\n   - `method='global'`: Single model with (1-W) masking, post-hoc heterogeneous effects, faster\n\n**When to use TROP vs SDID**:\n- Use **SDID** when parallel trends is plausible and factors are not a concern\n- Use **TROP** when you suspect factor confounding (regional shocks, economic cycles, latent factors)\n- Running both provides a useful robustness check\n\n**When to use local vs global method**:\n- Use **local** (default) for maximum flexibility with per-observation weights\n- Use **global** for faster estimation with global weights\n\n**Reference**:\n- Athey, S., Imbens, G. W., Qu, Z., & Viviano, D. (2025). Triply Robust Panel Estimators. *Working Paper*. https://arxiv.org/abs/2508.21536"
   },
   {
diff --git a/docs/tutorials/14_continuous_did.ipynb b/docs/tutorials/14_continuous_did.ipynb
index d3840c633..73008923c 100644
--- a/docs/tutorials/14_continuous_did.ipynb
+++ b/docs/tutorials/14_continuous_did.ipynb
@@ -62,7 +62,7 @@
    "id": "c1df897c",
    "metadata": {},
    "source": [
-    "## 1. Why Continuous Treatment Needs Special Methods\n",
+    "## Why Continuous Treatment Needs Special Methods\n",
     "\n",
     "Consider a job training program. Workers receive different amounts of training, and we want to know: **does more training lead to higher earnings?** Three problems arise with naive approaches:\n",
     "\n",
@@ -103,7 +103,7 @@
    "id": "b00d92bc",
    "metadata": {},
    "source": [
-    "## 2. Data Setup\n",
+    "## Data Setup\n",
     "\n",
     "We generate a balanced panel of workers. Some are assigned to a training cohort (treated in period 2), while 30% are never treated. Each trained worker receives a dose (hours) drawn from a log-normal distribution. The true treatment effect is **ATT(d) = 1 + 2d** — each hour of training adds \\\\$2 to earnings, plus a base effect of \\\\$1 for any training at all."
    ]
@@ -175,7 +175,7 @@
    "id": "9c685ba3",
    "metadata": {},
    "source": [
-    "## 3. Basic Estimation\n",
+    "## Basic Estimation\n",
     "\n",
     "The `ContinuousDiD` estimator follows an sklearn-like API. Call `fit()` with column names for outcome, unit, time, first treatment period, and dose. Setting `aggregate=\"dose\"` computes the dose-response curve and global summary parameters."
    ]
@@ -229,7 +229,7 @@
     "- **ATT_glob** is the average total effect of training (at the observed dose levels) compared to no training. It answers: \"On average, how much more do trained workers earn?\" The API reports this as `overall_att`. Under standard PT, this identifies the local average effect (ATT_loc); under strong PT, it additionally equals the global average (ATT_glob).\n",
     "- **ACRT_glob** is the average marginal return to an additional hour of training. It answers: \"What is one more hour of training worth, on average?\" This requires strong PT for causal interpretation.\n",
     "\n",
-    "Analytical standard errors are the default (fast, no resampling). For bootstrap-based inference, set `n_bootstrap` (see Section 6)."
+    "Analytical standard errors are the default (fast, no resampling). For bootstrap-based inference, set `n_bootstrap` (see Advanced Features)."
    ]
   },
   {
@@ -237,7 +237,7 @@
    "id": "8214ff79",
    "metadata": {},
    "source": [
-    "## 4. Dose-Response Curves\n",
+    "## Dose-Response Curves\n",
     "\n",
     "The key advantage of continuous DiD is the **dose-response curve** — how the treatment effect varies across dose levels. The estimator evaluates ATT(d) and ACRT(d) on a grid of dose values (by default, percentiles from P10 to P99 of the observed dose distribution).\n",
     "\n",
@@ -312,7 +312,7 @@
    "id": "c88d540c",
    "metadata": {},
    "source": [
-    "## 5. Event Study Diagnostics\n",
+    "## Event Study Diagnostics\n",
     "\n",
     "An event study aggregation binarizes treatment (treated vs. untreated) and shows effects by relative time period. This is useful for **pre-trends diagnostics**: pre-treatment coefficients should be near zero if parallel trends holds.\n",
     "\n",
@@ -389,7 +389,7 @@
    "id": "7b22b542",
    "metadata": {},
    "source": [
-    "## 6. Advanced Features\n",
+    "## Advanced Features\n",
     "\n",
     "### B-spline configuration\n",
     "\n",
@@ -564,7 +564,7 @@
    "id": "a5230e69",
    "metadata": {},
    "source": [
-    "## 7. Comparison to Binary DiD\n",
+    "## Comparison to Binary DiD\n",
     "\n",
     "What if we ignore dose entirely and just run a standard binary Callaway-Sant'Anna estimator? Both approaches should give a similar **binarized ATT** (treated vs. untreated), but the binary approach discards all dose information — no dose-response curve, no marginal effects.\n",
     "\n",
@@ -643,7 +643,7 @@
    "id": "b5af7533",
    "metadata": {},
    "source": [
-    "## 8. Results Export and Group-Time Effects\n",
+    "## Results Export and Group-Time Effects\n",
     "\n",
     "Results can be exported to DataFrames at different levels of aggregation using `to_dataframe()`:\n",
     "\n",
@@ -677,7 +677,7 @@
    "id": "315c7664",
    "metadata": {},
    "source": [
-    "## 9. Summary\n",
+    "## Summary\n",
     "\n",
     "### Key takeaways\n",
     "\n",
diff --git a/docs/tutorials/16_survey_did.ipynb b/docs/tutorials/16_survey_did.ipynb
index caf7ff27f..216dd927c 100644
--- a/docs/tutorials/16_survey_did.ipynb
+++ b/docs/tutorials/16_survey_did.ipynb
@@ -67,7 +67,7 @@
    "id": "cell-02-9151c763",
    "metadata": {},
    "source": [
-    "## 1. Why Survey Design Matters for DiD\n",
+    "## Why Survey Design Matters for DiD\n",
     "\n",
     "Our health survey, like most large-scale surveys, uses a **stratified multi-stage** design. Ignoring this structure leads to three problems:\n",
     "\n",
@@ -289,7 +289,7 @@
     "- **Clustering** (dominant): respondents in the same PSU share local conditions, so their outcomes are correlated (ICC = 0.84 in this DGP). Ignoring this makes standard errors far too small.\n",
     "- **Unequal weights**: oversampled groups get higher weights, adding variance. The Kish design effect from weight dispersion alone is ~1.4.\n",
     "\n",
-    "Note: this SE ratio compares two different estimators (survey-weighted vs. unweighted), so it is not a formal design effect (DEFF). The library's `compute_deff()` computes the proper DEFF against an SRS baseline for the same weighted estimator -- Section 7 covers that API."
+    "Note: this SE ratio compares two different estimators (survey-weighted vs. unweighted), so it is not a formal design effect (DEFF). The library's `compute_deff()` computes the proper DEFF against an SRS baseline for the same weighted estimator -- the DEFF Diagnostics section covers that API."
    ]
   },
   {
@@ -425,7 +425,7 @@
    "id": "cell-08-f642e821",
    "metadata": {},
    "source": [
-    "## 2. The Survey Design Object\n",
+    "## The Survey Design Object\n",
     "\n",
     "Our health survey has four design elements that map directly to `SurveyDesign` parameters:\n",
     "\n",
@@ -479,7 +479,7 @@
    "id": "cell-11-2d141c54",
    "metadata": {},
    "source": [
-    "## 3. Basic DiD with Survey Design\n",
+    "## Basic DiD with Survey Design\n",
     "\n",
     "Let's start simple: a 2x2 comparison for respondents in states that adopted the program at period 3, vs. respondents in states that never adopted."
    ]
@@ -603,7 +603,7 @@
    "id": "cell-17-a24a4c22",
    "metadata": {},
    "source": [
-    "## 4. Staggered DiD with Survey Design\n",
+    "## Staggered DiD with Survey Design\n",
     "\n",
     "In practice, states adopted the preventive care program at different times. This is a **staggered adoption** design. `CallawaySantAnna` handles both the staggered timing and the survey design.\n",
     "\n",
@@ -769,7 +769,7 @@
    "id": "cell-24-fb38f8fc",
    "metadata": {},
    "source": [
-    "## 5. Replicate Weights\n",
+    "## Replicate Weights\n",
     "\n",
     "Some surveys (MEPS, ACS PUMS) publish **replicate weights** instead of strata/PSU/FPC columns. The design information is encoded in the replicates themselves -- each replicate represents a perturbed version of the full sample.\n",
     "\n",
@@ -857,7 +857,7 @@
    "id": "cell-27-08164c2d",
    "metadata": {},
    "source": [
-    "## 6. Subpopulation Analysis\n",
+    "## Subpopulation Analysis\n",
     "\n",
     "Suppose we want to estimate the treatment effect only for respondents in **metro areas** (strata 0 and 1). A common mistake is to simply subset the data and ignore the survey design:\n",
     "\n",
@@ -920,7 +920,7 @@
    "id": "cell-29-a08d6b09",
    "metadata": {},
    "source": [
-    "## 7. DEFF Diagnostics\n",
+    "## DEFF Diagnostics\n",
     "\n",
     "The **design effect (DEFF)** quantifies how much the survey design inflates variance compared to a simple random sample (SRS) of the same size:\n",
     "\n",
@@ -1014,7 +1014,7 @@
    "id": "cell-32-ad3eea75",
    "metadata": {},
    "source": [
-    "## 8. Repeated Cross-Sections with Survey Design\n",
+    "## Repeated Cross-Sections with Survey Design\n",
     "\n",
     "Many health surveys are **repeated cross-sections** -- different respondents are sampled each period. Examples: BRFSS (monthly), CPS (monthly), ACS 1-year estimates.\n",
     "\n",
@@ -1087,14 +1087,14 @@
    "cell_type": "markdown",
    "id": "cell-35-f1ef376c",
    "metadata": {},
-   "source": "## 9. Which Estimators Support Survey Design?\n\n`diff-diff` supports survey design across all estimators, though the level of support varies:\n\n| Estimator | Weights | Strata/PSU/FPC (TSL) | Replicate Weights | Survey-Aware Bootstrap |\n|-----------|---------|---------------------|-------------------|------------------------|\n| **DifferenceInDifferences** | Full | Full | -- | -- |\n| **TwoWayFixedEffects** | Full | Full | -- | -- |\n| **MultiPeriodDiD** | Full | Full | -- | -- |\n| **CallawaySantAnna** | pweight only | Full | Full | Multiplier at PSU |\n| **TripleDifference** | pweight only | Full | Full (analytical) | -- |\n| **StaggeredTripleDifference** | pweight only | Full | Full | Multiplier at PSU |\n| **SunAbraham** | Full | Full | -- | Rao-Wu rescaled |\n| **StackedDiD** | pweight only | Full (pweight only) | -- | -- |\n| **ImputationDiD** | pweight only | Partial (no FPC) | -- | Multiplier at PSU |\n| **TwoStageDiD** | pweight only | Partial (no FPC) | -- | Multiplier at PSU |\n| **ContinuousDiD** | Full | Full | Full (analytical) | Multiplier at PSU |\n| **EfficientDiD** | Full | Full | Full (analytical) | Multiplier at PSU |\n| **SyntheticDiD** | pweight only | Full (all three variance methods) | -- | Hybrid pairs-bootstrap + Rao-Wu (bootstrap); stratified permutation (placebo); PSU-LOO (jackknife) |\n| **TROP** | pweight only | -- | -- | Rao-Wu rescaled |\n| **BaconDecomposition** | Diagnostic | Diagnostic | -- | -- |\n\n**Legend:**\n- **Full**: All weight types (pweight/fweight/aweight) + strata/PSU/FPC + Taylor Series Linearization variance\n- **Full (pweight only)**: Full TSL support with strata/PSU/FPC, but only accepts `pweight` weight type (`fweight`/`aweight` rejected because Q-weight composition changes their semantics)\n- **Partial (no FPC)**: Weights + strata (for df) + PSU (for clustering); FPC raises `NotImplementedError`\n- **pweight only** (Weights column): Only `pweight` accepted; `fweight`/`aweight` raise an error\n- **pweight only** (TSL column): Sampling weights for point estimates; no strata/PSU/FPC design elements\n- **Full (all three variance methods)** (SyntheticDiD TSL column): Strata/PSU/FPC supported on all three `variance_method` choices — `bootstrap` via weighted Frank-Wolfe + Rao-Wu, `placebo` via stratified permutation + weighted FW, `jackknife` via PSU-level LOO with stratum aggregation. Replicate-weight designs remain rejected (pre-existing limitation).\n- **Diagnostic**: Weighted descriptive statistics only (no inference)\n- **--**: Not supported\n\n**Note on SyntheticDiD:** all three variance methods now support full strata/PSU/FPC designs.\n\n- **Bootstrap** (PR #355) composes per-draw Rao-Wu rescaled weights with a weighted Frank-Wolfe variant of `_sc_weight_fw`. Each draw solves `min ||A·diag(rw)·ω - b||² + ζ²·Σ rw_i ω_i²` and composes `ω_eff = rw·ω/Σ(rw·ω)` for the SDID estimator. Pweight-only fits use the constant per-control survey weight as `rw`; full designs use Rao-Wu rescaling per draw.\n- **Placebo** uses a stratified permutation allocator: pseudo-treated indices are drawn from controls *within each stratum* containing actual treated units; weighted FW re-estimates ω and λ per draw with per-control survey weights flowing into both loss and regularization. SE follows Arkhangelsky Algorithm 4. The allocator requires at least `n_treated_h` controls per treated-containing stratum; fit-time guards raise targeted `ValueError` on infeasible configurations.\n- **Jackknife** uses PSU-level leave-one-out with stratum aggregation: `SE² = Σ_h (1-f_h)·(n_h-1)/n_h·Σ_{j∈h}(τ̂_{(h,j)} - τ̄_h)²` (Rust & Rao 1996). FPC folded via `(1-f_h)`; strata with fewer than 2 PSUs are silently skipped. Known anti-conservatism with few PSUs per stratum — for tight SE calibration in that regime, prefer `variance_method=\"bootstrap\"`.\n\nSee `docs/methodology/REGISTRY.md` §SyntheticDiD `Note (survey + bootstrap / placebo / jackknife composition)` for the full objectives, allocator asymmetry rationale (placebo ignores PSU axis, jackknife respects it), and validation details.\n\n**Note:** `EfficientDiD` supports `covariates` and `survey_design` simultaneously. The doubly-robust (DR) path threads survey weights through WLS outcome regression, weighted sieve propensity ratios, and survey-weighted kernel smoothing.\n\nFor full details, see `docs/survey-roadmap.md`."
+   "source": "## Which Estimators Support Survey Design?\n\n`diff-diff` supports survey design across all estimators, though the level of support varies:\n\n| Estimator | Weights | Strata/PSU/FPC (TSL) | Replicate Weights | Survey-Aware Bootstrap |\n|-----------|---------|---------------------|-------------------|------------------------|\n| **DifferenceInDifferences** | Full | Full | -- | -- |\n| **TwoWayFixedEffects** | Full | Full | -- | -- |\n| **MultiPeriodDiD** | Full | Full | -- | -- |\n| **CallawaySantAnna** | pweight only | Full | Full | Multiplier at PSU |\n| **TripleDifference** | pweight only | Full | Full (analytical) | -- |\n| **StaggeredTripleDifference** | pweight only | Full | Full | Multiplier at PSU |\n| **SunAbraham** | Full | Full | -- | Rao-Wu rescaled |\n| **StackedDiD** | pweight only | Full (pweight only) | -- | -- |\n| **ImputationDiD** | pweight only | Partial (no FPC) | -- | Multiplier at PSU |\n| **TwoStageDiD** | pweight only | Partial (no FPC) | -- | Multiplier at PSU |\n| **ContinuousDiD** | Full | Full | Full (analytical) | Multiplier at PSU |\n| **EfficientDiD** | Full | Full | Full (analytical) | Multiplier at PSU |\n| **SyntheticDiD** | pweight only | Full (all three variance methods) | -- | Hybrid pairs-bootstrap + Rao-Wu (bootstrap); stratified permutation (placebo); PSU-LOO (jackknife) |\n| **TROP** | pweight only | -- | -- | Rao-Wu rescaled |\n| **BaconDecomposition** | Diagnostic | Diagnostic | -- | -- |\n\n**Legend:**\n- **Full**: All weight types (pweight/fweight/aweight) + strata/PSU/FPC + Taylor Series Linearization variance\n- **Full (pweight only)**: Full TSL support with strata/PSU/FPC, but only accepts `pweight` weight type (`fweight`/`aweight` rejected because Q-weight composition changes their semantics)\n- **Partial (no FPC)**: Weights + strata (for df) + PSU (for clustering); FPC raises `NotImplementedError`\n- **pweight only** (Weights column): Only `pweight` accepted; `fweight`/`aweight` raise an error\n- **pweight only** (TSL column): Sampling weights for point estimates; no strata/PSU/FPC design elements\n- **Full (all three variance methods)** (SyntheticDiD TSL column): Strata/PSU/FPC supported on all three `variance_method` choices — `bootstrap` via weighted Frank-Wolfe + Rao-Wu, `placebo` via stratified permutation + weighted FW, `jackknife` via PSU-level LOO with stratum aggregation. Replicate-weight designs remain rejected (pre-existing limitation).\n- **Diagnostic**: Weighted descriptive statistics only (no inference)\n- **--**: Not supported\n\n**Note on SyntheticDiD:** all three variance methods now support full strata/PSU/FPC designs.\n\n- **Bootstrap** (PR #355) composes per-draw Rao-Wu rescaled weights with a weighted Frank-Wolfe variant of `_sc_weight_fw`. Each draw solves `min ||A·diag(rw)·ω - b||² + ζ²·Σ rw_i ω_i²` and composes `ω_eff = rw·ω/Σ(rw·ω)` for the SDID estimator. Pweight-only fits use the constant per-control survey weight as `rw`; full designs use Rao-Wu rescaling per draw.\n- **Placebo** uses a stratified permutation allocator: pseudo-treated indices are drawn from controls *within each stratum* containing actual treated units; weighted FW re-estimates ω and λ per draw with per-control survey weights flowing into both loss and regularization. SE follows Arkhangelsky Algorithm 4. The allocator requires at least `n_treated_h` controls per treated-containing stratum; fit-time guards raise targeted `ValueError` on infeasible configurations.\n- **Jackknife** uses PSU-level leave-one-out with stratum aggregation: `SE² = Σ_h (1-f_h)·(n_h-1)/n_h·Σ_{j∈h}(τ̂_{(h,j)} - τ̄_h)²` (Rust & Rao 1996). FPC folded via `(1-f_h)`; strata with fewer than 2 PSUs are silently skipped. Known anti-conservatism with few PSUs per stratum — for tight SE calibration in that regime, prefer `variance_method=\"bootstrap\"`.\n\nSee `docs/methodology/REGISTRY.md` §SyntheticDiD `Note (survey + bootstrap / placebo / jackknife composition)` for the full objectives, allocator asymmetry rationale (placebo ignores PSU axis, jackknife respects it), and validation details.\n\n**Note:** `EfficientDiD` supports `covariates` and `survey_design` simultaneously. The doubly-robust (DR) path threads survey weights through WLS outcome regression, weighted sieve propensity ratios, and survey-weighted kernel smoothing.\n\nFor full details, see `docs/survey-roadmap.md`."
   },
   {
    "cell_type": "markdown",
    "id": "cell-36-adb24679",
    "metadata": {},
    "source": [
-    "## 10. Real-World Example: ACA Young Adult Coverage (NHANES)\n",
+    "## Real-World Example: ACA Young Adult Coverage (NHANES)\n",
     "\n",
     "The examples above used synthetic data to illustrate survey design concepts. Now let's apply\n",
     "the same methods to **real CDC survey data** from the National Health and Nutrition Examination\n",
diff --git a/docs/tutorials/17_brand_awareness_survey.ipynb b/docs/tutorials/17_brand_awareness_survey.ipynb
index 372d46350..adf3dc1ae 100644
--- a/docs/tutorials/17_brand_awareness_survey.ipynb
+++ b/docs/tutorials/17_brand_awareness_survey.ipynb
@@ -10,7 +10,7 @@
    "cell_type": "markdown",
    "metadata": {},
    "source": [
-    "## 1. Setup"
+    "## Setup"
    ],
    "id": "63c132f7"
   },
@@ -25,7 +25,7 @@
   {
    "cell_type": "markdown",
    "metadata": {},
-   "source": "## 2. Data Preparation\n\nWe'll generate synthetic brand tracking data that mirrors a real survey:\n200 respondents across 8 waves, sampled from 5 geographic regions with\ncluster sampling and demographic weighting. The campaign launches at wave 5\nfor respondents in certain markets.",
+   "source": "## Data Preparation\n\nWe'll generate synthetic brand tracking data that mirrors a real survey:\n200 respondents across 8 waves, sampled from 5 geographic regions with\ncluster sampling and demographic weighting. The campaign launches at wave 5\nfor respondents in certain markets.",
    "id": "69d0010f"
   },
   {
@@ -48,7 +48,7 @@
    "cell_type": "markdown",
    "metadata": {},
    "source": [
-    "## 3. Visual Inspection\n",
+    "## Visual Inspection\n",
     "\n",
     "Before running any analysis, plot awareness over time for campaign vs. control markets.\n",
     "The key question: were the two groups trending similarly *before* the campaign launched?"
@@ -77,7 +77,7 @@
    "cell_type": "markdown",
    "metadata": {},
    "source": [
-    "## 4. Naive DiD (Ignoring Survey Design)\n",
+    "## Naive DiD (Ignoring Survey Design)\n",
     "\n",
     "First, run a standard DiD analysis that treats every survey response equally."
    ],
@@ -105,7 +105,7 @@
    "cell_type": "markdown",
    "metadata": {},
    "source": [
-    "## 5. Survey-Aware DiD\n",
+    "## Survey-Aware DiD\n",
     "\n",
     "Brand tracking surveys rarely use simple random sampling. Respondents are sampled in\n",
     "geographic clusters with demographic quotas and weighting. The `SurveyDesign` object\n",
@@ -157,7 +157,7 @@
    "cell_type": "markdown",
    "metadata": {},
    "source": [
-    "## 6. Multiple Brand Metrics\n",
+    "## Multiple Brand Metrics\n",
     "\n",
     "Brand campaigns don't just move awareness — they should also move consideration and\n",
     "purchase intent. Let's measure the lift across the full brand funnel."
@@ -216,7 +216,7 @@
    "cell_type": "markdown",
    "metadata": {},
    "source": [
-    "## 7. Is This Result Trustworthy?\n",
+    "## Is This Result Trustworthy?\n",
     "\n",
     "Two diagnostic checks help validate the result."
    ],
@@ -225,7 +225,7 @@
   {
    "cell_type": "markdown",
    "metadata": {},
-   "source": "### Parallel Trends Check\n\nDiD assumes campaign and control groups would have continued trending the same way if\nthe campaign hadn't run. `check_parallel_trends()` is a quick informal check that\ncompares pre-campaign slopes — it does not account for survey design, so treat it as\na sanity check rather than a formal test. The formal robustness assessment comes from\nHonestDiD in Section 8.",
+   "source": "### Parallel Trends Check\n\nDiD assumes campaign and control groups would have continued trending the same way if\nthe campaign hadn't run. `check_parallel_trends()` is a quick informal check that\ncompares pre-campaign slopes — it does not account for survey design, so treat it as\na sanity check rather than a formal test. The formal robustness assessment comes from\nHonestDiD in the staggered-rollout extension below.",
    "id": "96bbef84"
   },
   {
@@ -258,7 +258,7 @@
   {
    "cell_type": "markdown",
    "metadata": {},
-   "source": "The informal trend check is consistent with parallel trends, and the survey-aware\nplacebo test finds no effect where none should exist. Together these are supportive\nevidence, though neither formally proves the parallel trends assumption — it is always\nan untestable assumption about what *would have* happened.\n\nFor event study designs (Section 8 below), HonestDiD sensitivity analysis provides\na formal assessment of how robust the result is to violations of this assumption.",
+   "source": "The informal trend check is consistent with parallel trends, and the survey-aware\nplacebo test finds no effect where none should exist. Together these are supportive\nevidence, though neither formally proves the parallel trends assumption — it is always\nan untestable assumption about what *would have* happened.\n\nFor event study designs (the staggered-rollout extension below), HonestDiD sensitivity analysis provides\na formal assessment of how robust the result is to violations of this assumption.",
    "id": "5a36cef4"
   },
   {
@@ -290,7 +290,7 @@
    "cell_type": "markdown",
    "metadata": {},
    "source": [
-    "## 8. Extension: Staggered Campaign Rollout\n",
+    "## Extension: Staggered Campaign Rollout\n",
     "\n",
     "Many campaigns don't launch in all markets at once — they roll out in waves.\n",
     "Some markets go live in month 2, others in month 4. When this happens, basic\n",
@@ -367,7 +367,7 @@
    "cell_type": "markdown",
    "metadata": {},
    "source": [
-    "## 9. Communicating Results to Leadership\n",
+    "## Communicating Results to Leadership\n",
     "\n",
     "Here's how to write up the finding for stakeholders:"
    ],
diff --git a/docs/tutorials/18_geo_experiments.ipynb b/docs/tutorials/18_geo_experiments.ipynb
index 129b0a03f..9fb5a9c15 100644
--- a/docs/tutorials/18_geo_experiments.ipynb
+++ b/docs/tutorials/18_geo_experiments.ipynb
@@ -15,7 +15,7 @@
    "id": "t18-cell-002",
    "metadata": {},
    "source": [
-    "## 1. The Geo-Experiment Problem"
+    "## The Geo-Experiment Problem"
    ]
   },
   {
@@ -73,7 +73,7 @@
    "id": "t18-cell-007",
    "metadata": {},
    "source": [
-    "## 2. The Data"
+    "## The Data"
    ]
   },
   {
@@ -425,7 +425,7 @@
    "id": "t18-cell-014",
    "metadata": {},
    "source": [
-    "## 3. SyntheticDiD - The Idea and the Fit"
+    "## SyntheticDiD - The Idea and the Fit"
    ]
   },
   {
@@ -536,7 +536,7 @@
    "id": "t18-cell-019",
    "metadata": {},
    "source": [
-    "## 4. Diagnostics - Who Is the Synthetic Control?"
+    "## Diagnostics - Who Is the Synthetic Control?"
    ]
   },
   {
@@ -813,7 +813,7 @@
    ],
    "source": [
     "# Drift guards: tolerance-based asserts that lock the numbers quoted in the\n",
-    "# Section 3 narrative and the Section 6 stakeholder template. nbmake will fail\n",
+    "# SyntheticDiD fit narrative and the leadership-communication template. nbmake will fail\n",
     "# if generate_factor_data() or SyntheticDiD output drifts outside these ranges,\n",
     "# forcing the markdown to be updated before this notebook can pass CI.\n",
     "assert 305 <= results.att <= 320, f\"ATT drifted to {results.att:.2f} (expected ~312)\"\n",
@@ -832,7 +832,7 @@
    "id": "t18-cell-027",
    "metadata": {},
    "source": [
-    "## 5. Inference and Trustworthiness"
+    "## Inference and Trustworthiness"
    ]
   },
   {
@@ -1022,7 +1022,7 @@
    "id": "t18-cell-031",
    "metadata": {},
    "source": [
-    "**Why no parallel-trends test?** The Baker et al. workflow includes an explicit parallel-trends check (Step 3), but `practitioner_next_steps` skips it for `SyntheticDiD`. The reason isn't that SDiD has no identifying assumptions — it's that the relevant diagnostic is different. Basic DiD requires the *unweighted* control units to be on a parallel trend with the treated, an assumption that's notoriously fragile in geo-experiments where markets respond to different macro forces. SDiD instead reweights both units and time periods so the synthetic control's pre-trend matches the treated pre-trend by construction, then estimates the effect on those weighted comparisons. Identification still rests on assumptions (parallel trends in the *weighted* averages, no anticipation, design plausibility), but the empirical diagnostic shifts from \"are the raw control units parallel?\" to \"is the synthetic match tight enough?\" — which is exactly the pre-treatment fit RMSE we inspected in Section 4. This is what Arkhangelsky et al. (2021) call making the DiD strategy \"more plausible\" by automating the kind of pre-period adjustment that practitioners already do informally with covariates and period selection."
+    "**Why no parallel-trends test?** The Baker et al. workflow includes an explicit parallel-trends check (Step 3), but `practitioner_next_steps` skips it for `SyntheticDiD`. The reason isn't that SDiD has no identifying assumptions — it's that the relevant diagnostic is different. Basic DiD requires the *unweighted* control units to be on a parallel trend with the treated, an assumption that's notoriously fragile in geo-experiments where markets respond to different macro forces. SDiD instead reweights both units and time periods so the synthetic control's pre-trend matches the treated pre-trend by construction, then estimates the effect on those weighted comparisons. Identification still rests on assumptions (parallel trends in the *weighted* averages, no anticipation, design plausibility), but the empirical diagnostic shifts from \"are the raw control units parallel?\" to \"is the synthetic match tight enough?\" — which is exactly the pre-treatment fit RMSE we inspected in the diagnostics section. This is what Arkhangelsky et al. (2021) call making the DiD strategy \"more plausible\" by automating the kind of pre-period adjustment that practitioners already do informally with covariates and period selection."
    ]
   },
   {
@@ -1030,7 +1030,7 @@
    "id": "t18-cell-032",
    "metadata": {},
    "source": [
-    "## 6. Communicating Results to Leadership"
+    "## Communicating Results to Leadership"
    ]
   },
   {
@@ -1038,13 +1038,13 @@
    "id": "t18-cell-033",
    "metadata": {},
    "source": [
-    "A stakeholder-ready summary of the synthetic walkthrough (Sections 3-5 results):\n",
+    "A stakeholder-ready summary of the synthetic walkthrough (fit, diagnostics, and inference results):\n",
     "\n",
     "> **Headline.** The pilot campaign lifted weekly conversions in the 5 test markets by approximately **312 per market per week** (95% CI: 298 to 326), or about a **21% lift** on a baseline of ~1,500 conversions per market per week.\n",
     ">\n",
     "> **Sample size and design.** 5 pilot markets, 75 control markets. 12 weeks of weekly data: 6 weeks pre-launch, 6 weeks post-launch. Outcome: weekly conversions per market. Method: Synthetic Difference-in-Differences (Arkhangelsky et al. 2021), the canonical generalization of synthetic control to multi-treated panel settings. The 75 control markets serve as the donor pool that SDiD reweights to construct a counterfactual specific to the 5 pilot markets.\n",
     ">\n",
-    "> **Validity evidence.** The synthetic control's pre-treatment fit RMSE is well below the standard deviation of treated pre-period outcomes (the library would warn otherwise), which means the weighted blend of donor markets tracks the treated pre-trend closely. The placebo standard error matches the published Arkhangelsky et al. (2021) method, and we cross-checked with paper-faithful refit bootstrap inference (Section 5) — both methods agree on the point estimate (311.85) and on the result's significance, with bootstrap producing a narrower CI than placebo on this small panel (5 treated × 6 pre-periods, factor-model heterogeneity). The estimate is statistically significant under both inference methods, and the placebo 95% CI cleanly covers the true treatment effect on the synthetic data we used to demonstrate the workflow.\n",
+    "> **Validity evidence.** The synthetic control's pre-treatment fit RMSE is well below the standard deviation of treated pre-period outcomes (the library would warn otherwise), which means the weighted blend of donor markets tracks the treated pre-trend closely. The placebo standard error matches the published Arkhangelsky et al. (2021) method, and we cross-checked with paper-faithful refit bootstrap inference (see Inference and Trustworthiness) — both methods agree on the point estimate (311.85) and on the result's significance, with bootstrap producing a narrower CI than placebo on this small panel (5 treated × 6 pre-periods, factor-model heterogeneity). The estimate is statistically significant under both inference methods, and the placebo 95% CI cleanly covers the true treatment effect on the synthetic data we used to demonstrate the workflow.\n",
     ">\n",
     "> **What \"312 conversions per market per week\" means in business terms.** Across 5 pilot markets and 6 weeks, that's roughly 9,400 incremental conversions attributable to the campaign in this small pilot. Translate to your own revenue-per-conversion to compare against the pilot's campaign spend, then use the per-market lift estimate to project what a broader rollout would deliver.\n",
     ">\n",
diff --git a/docs/tutorials/19_dcdh_marketing_pulse.ipynb b/docs/tutorials/19_dcdh_marketing_pulse.ipynb
index 0e6cd652b..9f16d725c 100644
--- a/docs/tutorials/19_dcdh_marketing_pulse.ipynb
+++ b/docs/tutorials/19_dcdh_marketing_pulse.ipynb
@@ -10,7 +10,7 @@
    "cell_type": "markdown",
    "id": "t19-cell-002",
    "metadata": {},
-   "source": "## 1. The Marketing Pulse Problem\n\nYour team runs paid-promo pulses across 60 markets. Some markets ran the promo at the start of the quarter and turned it off as the campaign budget rolled to the next geo (leavers); others started untreated and switched the promo on at some point during the quarter (joiners). Leadership wants the average lift on weekly checkout sessions while the promo was on.\n\n**Why dCDH.** This panel has *reversible* (non-absorbing) treatment in the dCDH sense: across the panel, the promo turns on in some markets and off in others - both directions appear in the same dataset. Every other modern staggered-DiD estimator in diff-diff (Callaway-Sant'Anna, Sun-Abraham, Wooldridge ETWFE, ImputationDiD, TwoStageDiD, EfficientDiD) assumes treatment is absorbing: once treated, always treated. They simply don't apply to a panel that contains leavers. dCDH does, following [de Chaisemartin & D'Haultfoeuille (2020)](https://www.aeaweb.org/articles?id=10.1257/aer.20181169) and the [dynamic companion paper](https://www.nber.org/papers/w29873).\n\n**Scope of this tutorial.** Each market in our panel switches *at most once* during the quarter (the dCDH paper's Assumption 5, which the default analytical SE path requires). So a market is either a stable-untreated unit, a joiner that turns the promo on exactly once, a leaver that turns it off exactly once, or a stable-treated unit. dCDH does support multi-switch within-market paths (e.g., on-off-on cycles) via `drop_larger_lower=False` plus `by_path=k` for per-path effects, but that's a separate scope - see the extensions section at the end. Implementation details and any documented deviations from R's `did_multiplegt_dyn` reference live in [`docs/methodology/REGISTRY.md`](https://github.com/igerber/diff-diff/blob/main/docs/methodology/REGISTRY.md)."
+   "source": "## The Marketing Pulse Problem\n\nYour team runs paid-promo pulses across 60 markets. Some markets ran the promo at the start of the quarter and turned it off as the campaign budget rolled to the next geo (leavers); others started untreated and switched the promo on at some point during the quarter (joiners). Leadership wants the average lift on weekly checkout sessions while the promo was on.\n\n**Why dCDH.** This panel has *reversible* (non-absorbing) treatment in the dCDH sense: across the panel, the promo turns on in some markets and off in others - both directions appear in the same dataset. Every other modern staggered-DiD estimator in diff-diff (Callaway-Sant'Anna, Sun-Abraham, Wooldridge ETWFE, ImputationDiD, TwoStageDiD, EfficientDiD) assumes treatment is absorbing: once treated, always treated. They simply don't apply to a panel that contains leavers. dCDH does, following [de Chaisemartin & D'Haultfoeuille (2020)](https://www.aeaweb.org/articles?id=10.1257/aer.20181169) and the [dynamic companion paper](https://www.nber.org/papers/w29873).\n\n**Scope of this tutorial.** Each market in our panel switches *at most once* during the quarter (the dCDH paper's Assumption 5, which the default analytical SE path requires). So a market is either a stable-untreated unit, a joiner that turns the promo on exactly once, a leaver that turns it off exactly once, or a stable-treated unit. dCDH does support multi-switch within-market paths (e.g., on-off-on cycles) via `drop_larger_lower=False` plus `by_path=k` for per-path effects, but that's a separate scope - see the extensions section at the end. Implementation details and any documented deviations from R's `did_multiplegt_dyn` reference live in [`docs/methodology/REGISTRY.md`](https://github.com/igerber/diff-diff/blob/main/docs/methodology/REGISTRY.md)."
   },
   {
    "cell_type": "code",
@@ -34,7 +34,7 @@
    "cell_type": "markdown",
    "id": "t19-cell-004",
    "metadata": {},
-   "source": "## 2. The Data\n\nWe'll simulate a panel that mirrors a marketing pulse campaign:\n\n- **60 markets**, each observed for **8 weeks**\n- Some markets started the quarter with the promo on and switched it off (leavers); others started untreated and switched the promo on (joiners). Each market switches exactly once during the panel - the A5 single-switch contract (see [`docs/methodology/REGISTRY.md`](https://github.com/igerber/diff-diff/blob/main/docs/methodology/REGISTRY.md)) the analytical SE is derived under.\n- Outcome: weekly checkout sessions per market, baseline ~110\n- True treatment effect: **+12 sessions per market-week** when the promo is on, with mild cell-level heterogeneity around that average."
+   "source": "## The Data\n\nWe'll simulate a panel that mirrors a marketing pulse campaign:\n\n- **60 markets**, each observed for **8 weeks**\n- Some markets started the quarter with the promo on and switched it off (leavers); others started untreated and switched the promo on (joiners). Each market switches exactly once during the panel - the A5 single-switch contract (see [`docs/methodology/REGISTRY.md`](https://github.com/igerber/diff-diff/blob/main/docs/methodology/REGISTRY.md)) the analytical SE is derived under.\n- Outcome: weekly checkout sessions per market, baseline ~110\n- True treatment effect: **+12 sessions per market-week** when the promo is on, with mild cell-level heterogeneity around that average."
   },
   {
    "cell_type": "code",
@@ -120,7 +120,7 @@
    "cell_type": "markdown",
    "id": "t19-cell-008",
    "metadata": {},
-   "source": "## 3. Fitting dCDH\n\n`DID_M` is the headline dCDH estimator: the average across periods of two pieces:\n\n- **`DID_+`** (joiners): markets switching `0 → 1` between consecutive periods, compared to *contemporaneously untreated* control cells.\n- **`DID_-`** (leavers): markets switching `1 → 0`, compared to *contemporaneously treated* control cells.\n\nBoth pieces use only cells whose treatment status was stable across the two periods being compared. No *switching* cell is used as a control; stable-untreated cells serve as controls for joiners, and stable-treated cells serve as controls for leavers. The library reports `DID_+`, `DID_-`, and their average `DID_M` separately, so you can see if the two halves agree."
+   "source": "## Fitting dCDH\n\n`DID_M` is the headline dCDH estimator: the average across periods of two pieces:\n\n- **`DID_+`** (joiners): markets switching `0 → 1` between consecutive periods, compared to *contemporaneously untreated* control cells.\n- **`DID_-`** (leavers): markets switching `1 → 0`, compared to *contemporaneously treated* control cells.\n\nBoth pieces use only cells whose treatment status was stable across the two periods being compared. No *switching* cell is used as a control; stable-untreated cells serve as controls for joiners, and stable-treated cells serve as controls for leavers. The library reports `DID_+`, `DID_-`, and their average `DID_M` separately, so you can see if the two halves agree."
   },
   {
    "cell_type": "markdown",
@@ -155,7 +155,7 @@
    "source": [
     "**Reading the headline.** dCDH estimates the lift at **about 12.1 sessions per market-week** while the promo was on (95% CI: 11.3 to 12.8), recovering the true effect of 12.0 within sampling uncertainty. The CI half-width is about 0.7 sessions, which translates to a ~6% margin of error around a roughly 11% lift on a baseline of ~110 weekly sessions.\n",
     "\n",
-    "(We passed `placebo=False` on this fit because Phase 1's single-lag placebo SE is `NaN` by design - the per-period aggregation path doesn't have an analytical influence-function derivation. We get valid placebo CIs from the multi-horizon path in Section 4 below, which has a proper IF.)"
+    "(We passed `placebo=False` on this fit because Phase 1's single-lag placebo SE is `NaN` by design - the per-period aggregation path doesn't have an analytical influence-function derivation. We get valid placebo CIs from the multi-horizon event-study path below, which has a proper IF.)"
    ]
   },
   {
@@ -183,7 +183,7 @@
    "id": "t19-cell-014",
    "metadata": {},
    "source": [
-    "## 4. Multi-Horizon Event Study with Bootstrap\n",
+    "## Multi-Horizon Event Study with Bootstrap\n",
     "\n",
     "DID_M collapses the dynamic effect to one number - the average lift across all switching cells. Setting `L_max=L` instead computes `DID_l` for each horizon `l = 1..L` after each switch, plus `DID^pl_l` placebos at horizons `l = -L..-1`. This tells you whether the on-impact lift is sustained or fades, and whether the pre-treatment placebos sit on zero.\n",
     "\n",
@@ -293,14 +293,14 @@
     "- **On-impact effect** at l = 1 is about **+12.4 sessions** with a 95% bootstrap CI of roughly [11.4, 13.3], covering the true effect of 12.\n",
     "- **Sustained effect** at l = 2 is **+12.6 sessions** with CI [11.5, 13.6]. The lift didn't fade in the second week post-switch.\n",
     "\n",
-    "Bootstrap CIs reflect the cohort-recentered influence-function variance with the finite-sample stability the multiplier bootstrap provides. Both horizons agree closely with each other AND with the headline `DID_M` from Section 3 - a built-in consistency check across the per-period and per-group aggregation paths."
+    "Bootstrap CIs reflect the cohort-recentered influence-function variance with the finite-sample stability the multiplier bootstrap provides. Both horizons agree closely with each other AND with the headline `DID_M` from the initial dCDH fit - a built-in consistency check across the per-period and per-group aggregation paths."
    ]
   },
   {
    "cell_type": "markdown",
    "id": "t19-cell-019",
    "metadata": {},
-   "source": "## 5. Communicating Results to Leadership\n\nA stakeholder-ready summary of the analysis above:\n\n> **Headline.** The pulse campaign lifted weekly checkout sessions by approximately **12 sessions per market per week** while the promo was on (95% CI: 11.3 to 12.8). On a baseline of about 110 weekly sessions per market, that's roughly an **11% lift**. *[Source: `results.overall_att` from Section 3.]*\n>\n> **Sample size and design.** 60 markets observed for 8 weeks (480 market-weeks). Of those, 38 markets started untreated and switched the promo on at some point during the quarter (joiners), and 22 markets started with the promo on and switched it off (leavers). Method: dCDH (de Chaisemartin & D'Haultfoeuille 2020) - diff-diff's most general estimator built for treatment that can switch on AND off in the same panel. *[Source: switcher counts and panel shape from Section 2.]*\n>\n> **Validity evidence.** Two checks supported the result. (a) The joiners-vs-leavers split agreed: joiners produced a +12.1 lift, leavers a +11.9 lift, well within sampling uncertainty of each other and of the headline. (b) The multi-horizon placebos at l = -2 and l = -1 both sat on zero with bootstrap CIs comfortably covering it - parallel pre-trends look credible. *[Sources: joiners/leavers from Section 3, multi-horizon placebos from Section 4.]*\n>\n> **What \"+12 sessions per market per week\" means in business terms.** Across 60 markets and the weeks each one had the promo on, that's the per-market-week lift attributable to the campaign. Translate to your own revenue-per-session to compare against campaign spend, then use the per-market lift estimate to project what scaling the promo to additional markets would deliver.\n>\n> **Practical significance caveat.** The 11% lift is statistically significant (bootstrap p < 0.01 at both post-treatment horizons), and the on-impact effect persists at the second horizon - the pulse worked while it was on. Whether 11% justifies the campaign cost is a business judgment, not a statistical one. *[Sources: dynamic horizons from Section 4.]*"
+   "source": "## Communicating Results to Leadership\n\nA stakeholder-ready summary of the analysis above:\n\n> **Headline.** The pulse campaign lifted weekly checkout sessions by approximately **12 sessions per market per week** while the promo was on (95% CI: 11.3 to 12.8). On a baseline of about 110 weekly sessions per market, that's roughly an **11% lift**. *[Source: `results.overall_att` from Fitting dCDH.]*\n>\n> **Sample size and design.** 60 markets observed for 8 weeks (480 market-weeks). Of those, 38 markets started untreated and switched the promo on at some point during the quarter (joiners), and 22 markets started with the promo on and switched it off (leavers). Method: dCDH (de Chaisemartin & D'Haultfoeuille 2020) - diff-diff's most general estimator built for treatment that can switch on AND off in the same panel. *[Source: switcher counts and panel shape from The Data.]*\n>\n> **Validity evidence.** Two checks supported the result. (a) The joiners-vs-leavers split agreed: joiners produced a +12.1 lift, leavers a +11.9 lift, well within sampling uncertainty of each other and of the headline. (b) The multi-horizon placebos at l = -2 and l = -1 both sat on zero with bootstrap CIs comfortably covering it - parallel pre-trends look credible. *[Sources: joiners/leavers from Fitting dCDH, multi-horizon placebos from the event study.]*\n>\n> **What \"+12 sessions per market per week\" means in business terms.** Across 60 markets and the weeks each one had the promo on, that's the per-market-week lift attributable to the campaign. Translate to your own revenue-per-session to compare against campaign spend, then use the per-market lift estimate to project what scaling the promo to additional markets would deliver.\n>\n> **Practical significance caveat.** The 11% lift is statistically significant (bootstrap p < 0.01 at both post-treatment horizons), and the on-impact effect persists at the second horizon - the pulse worked while it was on. Whether 11% justifies the campaign cost is a business judgment, not a statistical one. *[Sources: dynamic horizons from the event study.]*"
   },
   {
    "cell_type": "markdown",
@@ -314,7 +314,7 @@
    "cell_type": "markdown",
    "id": "t19-cell-021",
    "metadata": {},
-   "source": "## 6. Extensions and Where to Go Next\n\nThis tutorial covered the core dCDH workflow on a reversible panel: `DID_M` with the joiners/leavers split, plus the `L_max` multi-horizon event study with multiplier bootstrap. The library also supports several extensions we did not demonstrate here:\n\n- **Per-trajectory disaggregation** (`by_path=k`): when joiners and leavers each follow a few common treatment paths (e.g., on-off-on vs on-on-off), `by_path=k` reports the event study separately for the top-k most common observed paths. Useful for pulse campaigns where the schedule varies across markets.\n- **Group-specific linear trends** (`trends_linear=True`): allows each market to have its own pre-treatment slope, absorbing differential trends.\n- **State-set-specific trends** (`trends_nonparam=...`): allows non-parametric trends shared within state-set strata.\n- **HonestDiD sensitivity analysis** (`honest_did=True`): Rambachan-Roth (2023) bounds on the post-treatment effects under controlled parallel-trends violations, computed on the placebo event-study surface.\n- **Survey-design support** (`survey_design=...`): Taylor-series linearization with sampling weights, strata, PSU, and FPC.\n\nSee [`docs/api/chaisemartin_dhaultfoeuille.rst`](../api/chaisemartin_dhaultfoeuille.html) for the full parameter reference and [`docs/methodology/REGISTRY.md`](https://github.com/igerber/diff-diff/blob/main/docs/methodology/REGISTRY.md) for the methodology contract on each surface."
+   "source": "## Extensions and Where to Go Next\n\nThis tutorial covered the core dCDH workflow on a reversible panel: `DID_M` with the joiners/leavers split, plus the `L_max` multi-horizon event study with multiplier bootstrap. The library also supports several extensions we did not demonstrate here:\n\n- **Per-trajectory disaggregation** (`by_path=k`): when joiners and leavers each follow a few common treatment paths (e.g., on-off-on vs on-on-off), `by_path=k` reports the event study separately for the top-k most common observed paths. Useful for pulse campaigns where the schedule varies across markets.\n- **Group-specific linear trends** (`trends_linear=True`): allows each market to have its own pre-treatment slope, absorbing differential trends.\n- **State-set-specific trends** (`trends_nonparam=...`): allows non-parametric trends shared within state-set strata.\n- **HonestDiD sensitivity analysis** (`honest_did=True`): Rambachan-Roth (2023) bounds on the post-treatment effects under controlled parallel-trends violations, computed on the placebo event-study surface.\n- **Survey-design support** (`survey_design=...`): Taylor-series linearization with sampling weights, strata, PSU, and FPC.\n\nSee [`docs/api/chaisemartin_dhaultfoeuille.rst`](../api/chaisemartin_dhaultfoeuille.html) for the full parameter reference and [`docs/methodology/REGISTRY.md`](https://github.com/igerber/diff-diff/blob/main/docs/methodology/REGISTRY.md) for the methodology contract on each surface."
   },
   {
    "cell_type": "markdown",
diff --git a/docs/tutorials/20_had_brand_campaign.ipynb b/docs/tutorials/20_had_brand_campaign.ipynb
index 262992f3a..9c9706e66 100644
--- a/docs/tutorials/20_had_brand_campaign.ipynb
+++ b/docs/tutorials/20_had_brand_campaign.ipynb
@@ -15,7 +15,7 @@
    "id": "t20-cell-002",
    "metadata": {},
    "source": [
-    "## 1. The Measurement Problem\n",
+    "## The Measurement Problem\n",
     "\n",
     "Your team launched a Q4 brand campaign with a national TV blast that hit every DMA on the same date. On top of that, regional teams allocated incremental ad budget on a per-DMA basis. The lightest-touch markets put in about $5K; the heaviest committed up to $50K. Every DMA got something - the regional teams all participated - but intensity varied widely. Leadership wants the per-dollar lift on weekly site visits attributable to the regional add-on spend."
    ]
@@ -61,7 +61,7 @@
    "id": "t20-cell-005",
    "metadata": {},
    "source": [
-    "## 2. The Data\n",
+    "## The Data\n",
     "\n",
     "60 DMAs over 8 weeks. Weeks 1-4 are the pre-campaign baseline; the campaign launches at week 5 with the national TV blast hitting every DMA. Regional teams' add-on spend varies from $5K to $50K per DMA - every DMA participates, none sits at $0. The true per-$1K lift on weekly site visits is locked at the seed; the tutorial recovers it."
    ]
@@ -182,7 +182,7 @@
    "id": "t20-cell-009",
    "metadata": {},
    "source": [
-    "## 3. Fitting HAD: The Headline Per-Dollar Lift\n",
+    "## Fitting HAD: The Headline Per-Dollar Lift\n",
     "\n",
     "HAD targets the **Weighted Average Slope** (WAS): roughly, the average per-dollar marginal effect across the dose distribution. Because all DMAs received at least some regional spend (none at exactly zero), HAD anchors the local-linear fit at the *boundary* of the dose distribution - the lightest-touch DMA's spend, called `d_lower`. The estimand is then `WAS_d_lower`: the average per-dollar marginal effect of regional spend *above* `d_lower`.\n",
     "\n",
@@ -263,7 +263,7 @@
    "id": "t20-cell-014",
    "metadata": {},
    "source": [
-    "## 4. Multi-Week Event Study\n",
+    "## Multi-Week Event Study\n",
     "\n",
     "The headline WAS_d_lower collapses 4 post-launch weeks into a single number. To see whether the per-dollar lift was immediate, building, or fading across the campaign, fit HAD's event-study mode on the full multi-week panel. It returns a per-week WAS_d_lower for each post-launch week (weeks since launch e=0, 1, 2, 3) plus pre-launch placebo horizons (e=-2, -3, -4) that should sit on zero if pre-trends are parallel.\n",
     "\n",
@@ -356,19 +356,19 @@
    "id": "t20-cell-019",
    "metadata": {},
    "source": [
-    "## 5. Communicating the Result to Leadership\n",
+    "## Communicating the Result to Leadership\n",
     "\n",
     "When you bring this back to a non-technical audience, lead with the headline number and ground each follow-on claim in a specific piece of evidence. The template below mirrors the structure used in the brand-awareness and geo-experiment tutorials in this library.\n",
     "\n",
-    "> **Headline.** Across 60 DMAs, the regional add-on ad spend delivered approximately **100 incremental weekly visits per additional $1,000** of regional spend above the lightest-touch DMA's $5K floor (95% CI: 98.6 to 101.4). On a baseline of about 5,000 weekly visits per DMA, the heaviest-spending DMAs ($50K) saw roughly 4,500 extra weekly visits during the campaign window. *[Source: `result.att` and `result.conf_int` from Section 3.]*\n",
+    "> **Headline.** Across 60 DMAs, the regional add-on ad spend delivered approximately **100 incremental weekly visits per additional $1,000** of regional spend above the lightest-touch DMA's $5K floor (95% CI: 98.6 to 101.4). On a baseline of about 5,000 weekly visits per DMA, the heaviest-spending DMAs ($50K) saw roughly 4,500 extra weekly visits during the campaign window. *[Source: `result.att` and `result.conf_int` from the headline HAD fit.]*\n",
     ">\n",
-    "> **Sample size and design.** 60 DMAs observed for 8 weeks (480 DMA-weeks). Every DMA received the national TV blast on the same date; regional add-on spend ranged from a $5K floor to $50K (median ~$25K). Method: `HeterogeneousAdoptionDiD` (HAD), built for panels where treatment varies in intensity across units and no never-treated unit is available; comparison comes from the dose variation across markets. *[Source: panel shape and dose distribution from Section 2.]*\n",
+    "> **Sample size and design.** 60 DMAs observed for 8 weeks (480 DMA-weeks). Every DMA received the national TV blast on the same date; regional add-on spend ranged from a $5K floor to $50K (median ~$25K). Method: `HeterogeneousAdoptionDiD` (HAD), built for panels where treatment varies in intensity across units and no never-treated unit is available; comparison comes from the dose variation across markets. *[Source: panel shape and dose distribution from The Data.]*\n",
     ">\n",
-    "> **Validity evidence.** Two visual checks. (a) The per-week placebo horizons before launch (weeks -4, -3, -2) sit at essentially zero, with 95% CIs comfortably bracketing zero - DMAs were not already trending differently by spend tier. (This is a visual parallel-trends check, not a formal pretest; HAD also ships a composite pretest workflow we did not run here.) (b) Per-week post-launch effects are stable across all four campaign weeks at ~100 visits per $1K. *[Source: event-study horizons from Section 4.]*\n",
+    "> **Validity evidence.** Two visual checks. (a) The per-week placebo horizons before launch (weeks -4, -3, -2) sit at essentially zero, with 95% CIs comfortably bracketing zero - DMAs were not already trending differently by spend tier. (This is a visual parallel-trends check, not a formal pretest; HAD also ships a composite pretest workflow we did not run here.) (b) Per-week post-launch effects are stable across all four campaign weeks at ~100 visits per $1K. *[Source: event-study horizons from the Multi-Week Event Study.]*\n",
     ">\n",
     "> **Business interpretation.** The per-$1K lift applies to spend *above* the $5K floor (HAD's local boundary). Subtracting the floor from each DMA's actual spend and multiplying gives the per-DMA visit lift estimate: a DMA at $30K saw roughly (30 - 5) x 100 = 2,500 extra weekly visits during the campaign; a DMA at $10K saw roughly (10 - 5) x 100 = 500. Translate to your own revenue-per-visit to compare against regional spend.\n",
     ">\n",
-    "> **Practical significance.** The per-dollar lift was stable across all four post-launch weeks - the campaign delivered its return on impact and held it. Whether the absolute per-DMA dollar lift justifies the regional spend is a business judgment, not a statistical one. Identification of WAS_d_lower additionally requires Assumption 6 (local linearity of the dose-response near the boundary), which is non-testable and should be argued from domain knowledge. *[Source: per-week dynamics from Section 4.]*"
+    "> **Practical significance.** The per-dollar lift was stable across all four post-launch weeks - the campaign delivered its return on impact and held it. Whether the absolute per-DMA dollar lift justifies the regional spend is a business judgment, not a statistical one. Identification of WAS_d_lower additionally requires Assumption 6 (local linearity of the dose-response near the boundary), which is non-testable and should be argued from domain knowledge. *[Source: per-week dynamics from the Multi-Week Event Study.]*"
    ]
   },
   {
@@ -384,12 +384,12 @@
    "id": "t20-cell-021",
    "metadata": {},
    "source": [
-    "## 6. Extensions\n",
+    "## Extensions\n",
     "\n",
     "This tutorial covered HAD's headline workflow: the overall WAS_d_lower fit and the multi-week event study. The library also supports several extensions we did not demonstrate here.\n",
     "\n",
     "- **Population-weighted (survey-aware) inference**: when some markets or regions carry more weight than others - e.g., DMAs weighted by population - HAD accepts a `SurveyDesign` object via the `survey_design=` kwarg on the same `fit()` interface. [Tutorial 22](22_had_survey_design.ipynb) walks the BRFSS-shape survey-design path end-to-end including the pretest workflow.\n",
-    "- **Composite pretest workflow**: HAD ships a `did_had_pretest_workflow` that combines the QUG support-infimum test (`H0: d_lower = 0`, which adjudicates between the `continuous_at_zero` and `continuous_near_d_lower` design paths) with linearity tests (Stute and Yatchew-HR). On the two-period (`aggregate='overall'`) path this workflow checks QUG and linearity only; the parallel-trends step is closed by the multi-period (`aggregate='event_study'`) joint variants (`stute_joint_pretest`, `joint_pretrends_test`, `joint_homogeneity_test`). The visual placebo check we used in Section 4 is a parallel-trends sanity check, not a substitute for the formal joint pretests; see [Tutorial 21](21_had_pretest_workflow.ipynb) for an end-to-end pretest walkthrough.\n",
+    "- **Composite pretest workflow**: HAD ships a `did_had_pretest_workflow` that combines the QUG support-infimum test (`H0: d_lower = 0`, which adjudicates between the `continuous_at_zero` and `continuous_near_d_lower` design paths) with linearity tests (Stute and Yatchew-HR). On the two-period (`aggregate='overall'`) path this workflow checks QUG and linearity only; the parallel-trends step is closed by the multi-period (`aggregate='event_study'`) joint variants (`stute_joint_pretest`, `joint_pretrends_test`, `joint_homogeneity_test`). The visual placebo check we used in the Multi-Week Event Study is a parallel-trends sanity check, not a substitute for the formal joint pretests; see [Tutorial 21](21_had_pretest_workflow.ipynb) for an end-to-end pretest walkthrough.\n",
     "- **`continuous_at_zero` design path**: if the lightest-touch DMA had no regional add-on (spend exactly $0), HAD switches to the Design 1' identification path with target `WAS` instead of `WAS_d_lower`. The auto-detection picks it up.\n",
     "- **Mass-point design path**: if a meaningful chunk of DMAs sit at exactly the same minimum spend (rather than spread continuously near the boundary), HAD switches to a 2SLS estimator with matching identification logic. Auto-detected as well.\n",
     "\n",
diff --git a/docs/tutorials/21_had_pretest_workflow.ipynb b/docs/tutorials/21_had_pretest_workflow.ipynb
index 3985d956f..24ca51fc3 100644
--- a/docs/tutorials/21_had_pretest_workflow.ipynb
+++ b/docs/tutorials/21_had_pretest_workflow.ipynb
@@ -17,7 +17,7 @@
    "id": "c45952e0",
    "metadata": {},
    "source": [
-    "## 1. The Pre-test Battery\n",
+    "## The Pre-test Battery\n",
     "\n",
     "de Chaisemartin et al. (2026) Section 4.2 lays out a four-step pre-test workflow for HAD identification:\n",
     "\n",
@@ -28,7 +28,7 @@
     "\n",
     "The library bundles the testable steps into one entry point: `did_had_pretest_workflow`. It dispatches to a two-period implementation (steps 1 + 3 only - step 2 needs at least two pre-periods) or a multi-period implementation (steps 1 + 2 + 3 jointly). The Yatchew-HR test from Step 3 is also exposed standalone with two null modes; we exercise both in the side panel.\n",
     "\n",
-    "**Non-testable identification caveat (separate from the four-step workflow).** Identification of the WAS estimand under Design 1' (`continuous_at_zero`, target = `WAS`) requires **Assumption 3** (uniform continuity of `d -> Y_2(d)` at zero, holds if the dose-response is Lipschitz; not testable). The Design 1 paths (`continuous_near_d_lower` / `mass_point`, target = `WAS_d_lower`) instead need **Assumption 5** (sign identification) or **Assumption 6** (`WAS_d_lower` point identification) - that is the caveat T20's tutorial flagged because T20's panel was Design 1. T21's panel resolves to Design 1' (see Section 2 + Section 3), so the relevant non-testable caveat here is Assumption 3, NOT Assumptions 5/6. The library reflects this: it emits a UserWarning about Assumption 5/6 on Design 1 fits and does not emit it on `continuous_at_zero` (Design 1') fits.\n"
+    "**Non-testable identification caveat (separate from the four-step workflow).** Identification of the WAS estimand under Design 1' (`continuous_at_zero`, target = `WAS`) requires **Assumption 3** (uniform continuity of `d -> Y_2(d)` at zero, holds if the dose-response is Lipschitz; not testable). The Design 1 paths (`continuous_near_d_lower` / `mass_point`, target = `WAS_d_lower`) instead need **Assumption 5** (sign identification) or **Assumption 6** (`WAS_d_lower` point identification) - that is the caveat T20's tutorial flagged because T20's panel was Design 1. T21's panel resolves to Design 1' (see the panel section + Step 1), so the relevant non-testable caveat here is Assumption 3, NOT Assumptions 5/6. The library reflects this: it emits a UserWarning about Assumption 5/6 on Design 1 fits and does not emit it on `continuous_at_zero` (Design 1') fits.\n"
    ]
   },
   {
@@ -36,7 +36,7 @@
    "id": "e75c0ee7",
    "metadata": {},
    "source": [
-    "## 2. The Panel\n",
+    "## The Panel\n",
     "\n",
     "We use a panel close in shape to T20's brand campaign (60 DMAs over 8 weeks, regional add-on spend on top of a national TV blast at week 5, true per-$1K lift = 100 weekly visits). The one difference: regional spend in this tutorial is drawn from `Uniform[$0.01K, $50K]` instead of T20's `Uniform[$5K, $50K]`. The true support of the dose distribution is therefore strictly positive (down to about $10), but very near zero - some markets barely participated in the regional add-on. Two independent things follow from that small `D_(1)`. (a) The QUG test in Step 1 will fail to reject `H0: d_lower = 0`, which means the data are **statistically consistent with** the `continuous_at_zero` (Design 1') identification path even though the true simulation lower bound is positive. (b) Independently, HAD's `design=\"auto\"` detection - which uses a separate min/median heuristic, NOT the QUG p-value (`continuous_at_zero` fires when `d.min() < 0.01 * median(|d|)`) - also lands on `continuous_at_zero` here, because `D_(1) / median(D)` is below 0.01 on this panel. Both checks point to the same identification path on this panel, but they are independent rules; the workflow's `_detect_design` does not consume the pre-test outcomes. The point of this tutorial is not to assert that the data is Design 1' from the DGP up; the point is to read what the workflow concludes from the data and what it leaves open.\n"
    ]
@@ -121,7 +121,7 @@
    "id": "f101faae",
    "metadata": {},
    "source": [
-    "## 3. Step 1: The Overall Workflow (Two-Period Path)\n",
+    "## Step 1: The Overall Workflow (Two-Period Path)\n",
     "\n",
     "T20's headline used a two-period collapse of the panel - average pre-launch outcome per DMA against average post-launch outcome per DMA. That's also the natural input shape for HAD's two-period (`aggregate=\"overall\"`) pre-test workflow, which runs **paper Step 1 (QUG) + paper Step 3 (linearity, via Stute and Yatchew-HR)**. Step 2 (parallel pre-trends) is not implemented on this path - a single pre-period structurally can't support a pre-trends test - and the workflow's verdict says so explicitly.\n",
     "\n",
@@ -195,7 +195,7 @@
    "source": [
     "**Reading the overall verdict.** Three things to note.\n",
     "\n",
-    "- **Step 1 (QUG) fails to reject:** the test statistic `T = D_(1) / (D_(2) - D_(1)) ~ 3.86` lands well below its critical value (`1/alpha - 1 = 19` at alpha = 0.05); the data are statistically consistent with `d_lower = 0`. (Failing to reject is non-rejection, not proof - the true support could still be slightly above zero in finite samples; here it is, by construction of the DGP. QUG's outcome supports interpreting the data as Design 1', but the QUG test is independent of HAD's `design=\"auto\"` selector - which uses the min/median heuristic described in Section 2 to reach the same `continuous_at_zero` decision on this panel.)\n",
+    "- **Step 1 (QUG) fails to reject:** the test statistic `T = D_(1) / (D_(2) - D_(1)) ~ 3.86` lands well below its critical value (`1/alpha - 1 = 19` at alpha = 0.05); the data are statistically consistent with `d_lower = 0`. (Failing to reject is non-rejection, not proof - the true support could still be slightly above zero in finite samples; here it is, by construction of the DGP. QUG's outcome supports interpreting the data as Design 1', but the QUG test is independent of HAD's `design=\"auto\"` selector - which uses the min/median heuristic described in the panel section to reach the same `continuous_at_zero` decision on this panel.)\n",
     "- **Step 3 (linearity) fails to reject** on both Stute (CvM) and Yatchew-HR. The diagnostics do not flag heterogeneity bias on the dose dimension, so reading the WAS as an average per-dose marginal effect is supported by these tests (subject to finite-sample power).\n",
     "- **Step 2 (Assumption 7 pre-trends) is not run on this path.** The verdict says so verbatim: `\"Assumption 7 pre-trends test NOT run (paper step 2 deferred to Phase 3 follow-up)\"`. With a single pre-period (the avg over weeks 1-4), there is nothing to compare against - we need at least two pre-periods to run a parallel-trends test on the dose dimension. The structural fields back this up: `pretrends_joint` and `homogeneity_joint` on the report are both `None` (the joint-Stute output containers don't get populated on the two-period path).\n",
     "\n",
@@ -284,7 +284,7 @@
    "id": "86c21280",
    "metadata": {},
    "source": [
-    "## 4. Step 2: Upgrade to the Event-Study Workflow\n",
+    "## Step 2: Upgrade to the Event-Study Workflow\n",
     "\n",
     "The two-period workflow ran Steps 1 and 3 but did not run Step 2 (parallel pre-trends). Our panel actually has 8 weeks - that is enough pre-periods to add the joint Stute pre-trends diagnostic (paper Section 4.2 step 2 + Hlavka-Huskova 2020 / Delgado-Manteiga 2001 dependence-preserving Mammen multiplier bootstrap).\n",
     "\n",
@@ -351,7 +351,7 @@
    "source": [
     "**Reading the event-study verdict.** Now the verdict reads `\"QUG, joint pre-trends, and joint linearity diagnostics fail-to-reject (TWFE admissible under Section 4 assumptions)\"`. The `\"deferred\"` caveat from the overall path is gone because the joint pre-trends and joint homogeneity diagnostics now ran. The structural fields confirm: `pretrends_joint` and `homogeneity_joint` are both populated.\n",
     "\n",
-    "A note on the verdict's \"TWFE admissible\" language. This is the workflow's classifier output when none of the three testable diagnostics rejects at the configured `alpha = 0.05` (paper Step 4 decision rule). That is non-rejection evidence under the diagnostics' finite-sample power and specification, not proof that the identifying assumptions hold. The non-testable Design 1' identification caveat (Assumption 3 / boundary regularity at zero, see Section 1) sits alongside this and is not covered by any of the three diagnostics.\n",
+    "A note on the verdict's \"TWFE admissible\" language. This is the workflow's classifier output when none of the three testable diagnostics rejects at the configured `alpha = 0.05` (paper Step 4 decision rule). That is non-rejection evidence under the diagnostics' finite-sample power and specification, not proof that the identifying assumptions hold. The non-testable Design 1' identification caveat (Assumption 3 / boundary regularity at zero, see The Pre-test Battery) sits alongside this and is not covered by any of the three diagnostics.\n",
     "\n",
     "The joint pre-trends test runs over `n_horizons = 3` (pre-periods 1, 2, 3, with week 4 reserved as the base period). The joint homogeneity test runs over `n_horizons = 4` (post-periods 5, 6, 7, 8). Let's inspect the per-horizon detail.\n"
    ]
@@ -453,7 +453,7 @@
    "id": "63068545",
    "metadata": {},
    "source": [
-    "## 5. Side Panel: Yatchew-HR Null Modes\n",
+    "## Side Panel: Yatchew-HR Null Modes\n",
     "\n",
     "The Yatchew-HR test exposes two `null=` modes (the second was added in 2026-04 for parity with the R `YatchewTest` package).\n",
     "\n",
@@ -552,9 +552,9 @@
    "id": "b16b856f",
    "metadata": {},
    "source": [
-    "## 6. Communicating the Diagnostics to Leadership\n",
+    "## Communicating the Diagnostics to Leadership\n",
     "\n",
-    "Pre-test results travel awkwardly to non-technical audiences. The template below structures the diagnostics around what each test does and does not rule out - mirroring the headline-and-evidence pattern from T20 Section 5.\n",
+    "Pre-test results travel awkwardly to non-technical audiences. The template below structures the diagnostics around what each test does and does not rule out - mirroring the headline-and-evidence pattern from T20's leadership-communication section.\n",
     "\n",
     "> **The HAD pre-test diagnostics on the brand-campaign panel do not flag a violation of the testable identifying assumptions.**\n",
     ">\n",
@@ -572,7 +572,7 @@
    "id": "eb1d3712",
    "metadata": {},
    "source": [
-    "## 7. Extensions\n",
+    "## Extensions\n",
     "\n",
     "This tutorial covered the composite pre-test workflow on a single panel where QUG fail-to-reject and HAD's `design=\"auto\"` heuristic both pointed independently to the `continuous_at_zero` (Design 1') identification path. A few directions we did not exercise here:\n",
     "\n",
@@ -594,7 +594,7 @@
    "id": "dd1dce6f",
    "metadata": {},
    "source": [
-    "## 8. Summary Checklist\n",
+    "## Summary Checklist\n",
     "\n",
     "- HAD's pre-test workflow `did_had_pretest_workflow` bundles paper Section 4.2 Steps 1 (QUG support infimum), 2 (joint Stute pre-trends - event-study path only), and 3 (Stute / Yatchew-HR linearity, joint variant on event-study path).\n",
     "- The two-period (`aggregate=\"overall\"`) path runs Steps 1 + 3 only - it cannot run Step 2 because a single pre-period structurally has nothing to test against. The verdict says so verbatim: \"Assumption 7 pre-trends test NOT run\".\n",
diff --git a/docs/tutorials/22_had_survey_design.ipynb b/docs/tutorials/22_had_survey_design.ipynb
index f3ea02a6a..4203b2702 100644
--- a/docs/tutorials/22_had_survey_design.ipynb
+++ b/docs/tutorials/22_had_survey_design.ipynb
@@ -91,7 +91,7 @@
    "id": "239a35f5",
    "metadata": {},
    "source": [
-    "## 1. Survey design adds two problems for HAD\n",
+    "## Survey design adds two problems for HAD\n",
     "\n",
     "When the panel comes out of a complex survey instead of a simple random\n",
     "sample, HAD's headline numbers acquire two new failure modes:\n",
@@ -114,7 +114,7 @@
     "neighborhood at the boundary** rather than from a regression coefficient\n",
     "on the full panel - so survey-aware SE inflation manifests differently\n",
     "than the rule-of-thumb you would expect from CallawaySantAnna or\n",
-    "LinearRegression. We come back to this in section 4.\n",
+    "LinearRegression. We come back to this in the SE-inflation section.\n",
     "\n",
     "The Phase 4.5 C0 deferral is the second user-facing caveat: the HAD\n",
     "QUG step (paper Section 4 step 1; tests `H_0: d_lower = 0` via the\n",
@@ -123,8 +123,8 @@
     "differentiable functionals of the empirical CDF, so no off-the-shelf\n",
     "survey-adjusted EVT fits the slot. The pretest workflow surfaces this\n",
     "deferral in the verdict text and on `report.qug` (set to `None`); the\n",
-    "linearity-conditional remainder of the verdict still runs. Section 6\n",
-    "walks the verdict text in detail.\n"
+    "linearity-conditional remainder of the verdict still runs. The pretest-workflow\n",
+    "section walks the verdict text in detail.\n"
    ]
   },
   {
@@ -132,7 +132,7 @@
    "id": "f74f76d9",
    "metadata": {},
    "source": [
-    "## 2. The panel: BRFSS-shape state rollout\n",
+    "## The panel: BRFSS-shape state rollout\n",
     "\n",
     "The DGP shape is identical to T20 (60 states x 8 weeks; rollout at\n",
     "w=5; dose ~ Uniform[$5K, $50K] of supplemental campaign spend per\n",
@@ -293,7 +293,7 @@
    "id": "326875a6",
    "metadata": {},
    "source": [
-    "## 3. Naive vs survey-aware headline fit\n",
+    "## Naive vs survey-aware headline fit\n",
     "\n",
     "T20's headline path collapses to two periods (pre-mean vs post-mean\n",
     "per state) and fits HAD with `design=\"auto\"` — the heuristic lands\n",
@@ -343,7 +343,7 @@
     "    )\n",
     "    # NB: we deliberately do NOT filter the\n",
     "    # `continuous_near_d_lower ... Assumption 5 or 6` warning here\n",
-    "    # so it fires once on the headline fit (the §3 interpretation\n",
+    "    # so it fires once on the headline fit (the headline-fit interpretation\n",
     "    # cell discusses it in prose). Subsequent fit cells suppress it\n",
     "    # as redundant.\n",
     "    naive = HAD(design=\"auto\").fit(\n",
@@ -410,7 +410,7 @@
     "**Assumption 6** for point identification of `WAS_d_lower` (or\n",
     "**Assumption 5** for sign identification only) — both are about\n",
     "local linearity of the dose-response near `d_lower` and are **not\n",
-    "testable from data**. The §6 linearity diagnostics (Stute,\n",
+    "testable from data**. The pretest-workflow linearity diagnostics (Stute,\n",
     "Yatchew, joint pretrends/homogeneity) are necessary but not\n",
     "sufficient. Assumption 6 itself is justified from domain\n",
     "knowledge (is the marginal effect of the next $1K of supplemental\n",
@@ -418,7 +418,7 @@
     "a `UserWarning` on every Design 1 fit; the headline cell above\n",
     "lets it surface, subsequent cells filter it as redundant. This is\n",
     "the load-bearing methodology caveat alongside the QUG-under-survey\n",
-    "deferral (§6)."
+    "deferral (the pretest-workflow section)."
    ]
   },
   {
@@ -426,7 +426,7 @@
    "id": "637dd580",
    "metadata": {},
    "source": [
-    "## 4. Why the SE inflation is modest for HAD\n",
+    "## Why the SE inflation is modest for HAD\n",
     "\n",
     "**The intuition.** `WAS_d_lower` is the average slope above d_lower,\n",
     "but its leading-order variance reads off a local-linear boundary\n",
@@ -500,7 +500,7 @@
     "})\n",
     "print(ratios.to_string(index=False))\n",
     "print()\n",
-    "print(f\"overall SE ratio (from section 3) = {survey.se / naive.se:.3f}\")\n"
+    "print(f\"overall SE ratio (from the headline fit) = {survey.se / naive.se:.3f}\")\n"
    ]
   },
   {
@@ -531,7 +531,7 @@
    "id": "fc9b05ca",
    "metadata": {},
    "source": [
-    "## 5. Event-study under the survey design\n",
+    "## Event-study under the survey design\n",
     "\n",
     "Refit with `aggregate=\"event_study\"` and `cband=True` to get\n",
     "per-horizon ATT estimates plus a sup-t confidence band that adjusts\n",
@@ -627,7 +627,7 @@
     "the treatment-effect contribution (cross-unit variation in\n",
     "`slope * D_{g,F}`), driving larger local-linear residual variance\n",
     "than the pre-period placebos see (placebos use the same `D_{g,F}`\n",
-    "regressor, but `ΔY_{g,t}` is noise-only — see section 4).\n"
+    "regressor, but `ΔY_{g,t}` is noise-only — see the SE-inflation section).\n"
    ]
   },
   {
@@ -635,7 +635,7 @@
    "id": "95252100",
    "metadata": {},
    "source": [
-    "## 6. Pretest workflow under the survey design\n",
+    "## Pretest workflow under the survey design\n",
     "\n",
     "`did_had_pretest_workflow` runs the three-step diagnostic battery\n",
     "(QUG; Stute or joint Stute; Yatchew) and composes a single verdict\n",
@@ -768,7 +768,7 @@
    "id": "a40a113e",
    "metadata": {},
    "source": [
-    "## 7. Communicating the design-based verdict to leadership\n",
+    "## Communicating the design-based verdict to leadership\n",
     "\n",
     "> **For the executive.** The campaign moved screening uptake by\n",
     "> **about 100 per 1,000 adults per $1K of supplemental spend**, with\n",
@@ -797,7 +797,7 @@
     "> because HAD uses a local-linear boundary fit at `d_lower` to\n",
     "> estimate the boundary-limit term in the `WAS_d_lower` formula\n",
     "> (variance is dominated by the few states near the boundary; see\n",
-    "> §4).\n"
+    "> the SE-inflation section).\n"
    ]
   },
   {
@@ -805,7 +805,7 @@
    "id": "d79ad79a",
    "metadata": {},
    "source": [
-    "## 8. Extensions + summary checklist\n",
+    "## Extensions + summary checklist\n",
     "\n",
     "**Extensions to keep on the radar:**\n",
     "\n",
@@ -831,12 +831,12 @@
     "\n",
     "**What you can ship to leadership:**\n",
     "\n",
-    "- Single design-aware ATT with a survey-design CI (section 3, 5).\n",
+    "- Single design-aware ATT with a survey-design CI (headline-fit and event-study sections).\n",
     "- Verdict text that flags the C0 deferral verbatim - do not summarize\n",
-    "  it away (section 6).\n",
+    "  it away (the pretest-workflow section).\n",
     "- Per-horizon ATT and sup-t cband when you want a rollout chart\n",
-    "  (section 5).\n",
-    "- A side note on the modest SE inflation magnitude (section 4) so the\n",
+    "  (the event-study section).\n",
+    "- A side note on the modest SE inflation magnitude (the SE-inflation section) so the\n",
     "  audience does not over-correct against the simpler analysis.\n",
     "- The workflow's `report.summary()` block as the methodologist-facing\n",
     "  artifact; `report.verdict` as the executive-facing artifact.\n"
diff --git a/docs/tutorials/23_spillover_tva.ipynb b/docs/tutorials/23_spillover_tva.ipynb
index be44cbb6d..8128dfe4e 100644
--- a/docs/tutorials/23_spillover_tva.ipynb
+++ b/docs/tutorials/23_spillover_tva.ipynb
@@ -50,7 +50,7 @@
    "id": "a16e76a4",
    "metadata": {},
    "source": [
-    "## 2. The synthetic panel\n",
+    "## The synthetic panel\n",
     "\n",
     "We build a 4-period, 200-unit panel laid out as three geographic bands:\n",
     "\n",
@@ -183,7 +183,7 @@
    "id": "8be4e7c9",
    "metadata": {},
    "source": [
-    "## 3. The naive headline \u2014 multi-period TWFE on the full sample\n",
+    "## The naive headline \u2014 multi-period TWFE on the full sample\n",
     "\n",
     "A practitioner reaching for a default DiD specification fits a\n",
     "multi-period TWFE on the full panel, treating ALL non-treated-cluster\n",
@@ -244,7 +244,7 @@
    "id": "cb77d87a",
    "metadata": {},
    "source": [
-    "## 4. Choosing the spillover bandwidth\n",
+    "## Choosing the spillover bandwidth\n",
     "\n",
     "`SpilloverDiD` is parameterized by `rings=[r_0, r_1, ..., r_K]` where\n",
     "the $j$-th ring covers distance band $[r_{j-1}, r_j]$ km. The\n",
@@ -333,10 +333,10 @@
    "id": "ce1cb17e",
    "metadata": {},
    "source": [
-    "## 5. Fit `SpilloverDiD` and interpret\n",
+    "## Fit `SpilloverDiD` and interpret\n",
     "\n",
     "With `d_bar = 100`, the headline fit recovers both the direct effect\n",
-    "and the spillover. We compare side-by-side with the naive TWFE from \u00a73.\n"
+    "and the spillover. We compare side-by-side with the naive TWFE from the naive-headline section.\n"
    ]
   },
   {
@@ -348,7 +348,7 @@
    "source": [
     "est = SpilloverDiD(rings=[0.0, D_BAR_KM], conley_coords=(\"lat\", \"lon\"))\n",
     "with warnings.catch_warnings():\n",
-    "    # See \u00a74 for the rationale on this narrow matmul-RuntimeWarning filter.\n",
+    "    # See the bandwidth section for the rationale on this narrow matmul-RuntimeWarning filter.\n",
     "    warnings.filterwarnings(\n",
     "        \"ignore\", category=RuntimeWarning, message=\".*encountered in matmul\"\n",
     "    )\n",
@@ -390,9 +390,9 @@
    "id": "2e005043",
    "metadata": {},
    "source": [
-    "## 6. Robust inference with Conley spatial-HAC\n",
+    "## Robust inference with Conley spatial-HAC\n",
     "\n",
-    "The HC1 standard error in \u00a75 treats each observation's score as\n",
+    "The HC1 standard error in the `SpilloverDiD` fit section treats each observation's score as\n",
     "independent of every other. On spatially-correlated panels \u2014 and this\n",
     "is the canonical case for place-based interventions \u2014 that\n",
     "independence assumption is violated. Conley (1999) provides a\n",
@@ -402,7 +402,7 @@
     "`conley_lag_cutoff`.\n",
     "\n",
     "Per Butts (2021) \u00a73.1, the Conley bandwidth is typically chosen to\n",
-    "match the spillover horizon `d_bar`. Below we compare the \u00a75 HC1 SE\n",
+    "match the spillover horizon `d_bar`. Below we compare the fit's HC1 SE\n",
     "against Conley with `conley_cutoff_km = d_bar = 100`, both with\n",
     "`conley_lag_cutoff = 0` (spatial only) and `conley_lag_cutoff = 1` (add\n",
     "within-unit serial correlation across consecutive periods).\n"
@@ -422,7 +422,7 @@
     "        kwargs[\"conley_lag_cutoff\"] = lag\n",
     "    est = SpilloverDiD(**kwargs)\n",
     "    with warnings.catch_warnings():\n",
-    "        # See \u00a74 for the rationale on this narrow matmul-RuntimeWarning filter.\n",
+    "        # See the bandwidth section for the rationale on this narrow matmul-RuntimeWarning filter.\n",
     "        warnings.filterwarnings(\n",
     "            \"ignore\", category=RuntimeWarning, message=\".*encountered in matmul\"\n",
     "        )\n",
@@ -503,7 +503,7 @@
    "id": "5cc89608",
    "metadata": {},
    "source": [
-    "## 7. Practitioner takeaways and where to go next\n",
+    "## Practitioner takeaways and where to go next\n",
     "\n",
     "**When to reach for `SpilloverDiD`.** Whenever your treatment has a\n",
     "plausible geographic externality \u2014 labor-market spillovers,\n",
@@ -517,7 +517,7 @@
     "\n",
     "**How to choose `d_bar`.** Start from theory: what's the plausible\n",
     "spatial range of the spillover mechanism? Triangulate with a\n",
-    "sensitivity grid (\u00a74 above). The estimate stabilizes once `d_bar`\n",
+    "sensitivity grid (the bandwidth section above). The estimate stabilizes once `d_bar`\n",
     "covers the true horizon; further widening is benign unless it sweeps\n",
     "clean controls into the spillover bin.\n",
     "\n",
diff --git a/docs/tutorials/24_staggered_vs_collapsed_power.ipynb b/docs/tutorials/24_staggered_vs_collapsed_power.ipynb
index 9e500a3b4..f79fc93f8 100644
--- a/docs/tutorials/24_staggered_vs_collapsed_power.ipynb
+++ b/docs/tutorials/24_staggered_vs_collapsed_power.ipynb
@@ -275,7 +275,7 @@
    "id": "99001403",
    "metadata": {},
    "source": [
-    "## 1. \"Simplifying\" silently changes the question\n",
+    "## \"Simplifying\" silently changes the question\n",
     "\n",
     "When you collapse the whole post-window into one \"after\" cell, the states that *haven't\n",
     "gone live yet* are still counted as treated. Their flat, pre-treatment outcomes get\n",
@@ -520,7 +520,7 @@
    "id": "c6236105",
    "metadata": {},
    "source": [
-    "## 2. So does CS cost you power? (the headline)\n",
+    "## So does CS cost you power? (the headline)\n",
     "\n",
     "Yes — for *pure detection* the collapsed 2×2 is more powerful, because pooling the whole\n",
     "post-window into one cell gives it a tighter standard error than CS's group-time-then-\n",
diff --git a/docs/tutorials/25_synthetic_control_policy.ipynb b/docs/tutorials/25_synthetic_control_policy.ipynb
index 6c66a3dc0..2a6657472 100644
--- a/docs/tutorials/25_synthetic_control_policy.ipynb
+++ b/docs/tutorials/25_synthetic_control_policy.ipynb
@@ -37,7 +37,7 @@
    "id": "6fcc14b4",
    "metadata": {},
    "source": [
-    "## 1. The setup\n",
+    "## The setup\n",
     "\n",
     "We use `generate_synthetic_control_data`, which builds a factor-model panel with **one\n",
     "treated unit (id `0`)** and a donor pool. The treated unit's latent factor loadings and\n",
@@ -49,7 +49,7 @@
     "it rolls out.\n",
     "\n",
     "We use a long pre-period (60 quarters) and a short post-period (5 quarters): synthetic\n",
-    "control wants a long, well-matched pre-window, and — as we'll see in §5 — the conformal\n",
+    "control wants a long, well-matched pre-window, and — as we'll see in Philosophy B — the conformal\n",
     "average-effect interval needs enough pre-period \"blocks\" to resolve."
    ]
   },
@@ -240,7 +240,7 @@
    "id": "eec7c3db",
    "metadata": {},
    "source": [
-    "## 2. Fit the synthetic control\n",
+    "## Fit the synthetic control\n",
     "\n",
     "`SyntheticControl` chooses **donor weights** (nonnegative, summing to one) so the synthetic\n",
     "state matches the treated state on the predictors, and **predictor weights** `V` (via a\n",
@@ -462,7 +462,7 @@
    "id": "50e1271f",
    "metadata": {},
    "source": [
-    "## 3. The effect path\n",
+    "## The effect path\n",
     "\n",
     "The **gap** (treated − synthetic) is flat and near zero before the policy and opens up\n",
     "after it. Overlaying the known `true_effect` shows the estimate recovers the ramp."
@@ -524,7 +524,7 @@
    "id": "6c8019a7",
    "metadata": {},
    "source": [
-    "## 4. Philosophy A — compare across regions\n",
+    "## Philosophy A — compare across regions\n",
     "\n",
     "If the policy did nothing, the treated state's post/pre RMSPE ratio should look like what we\n",
     "get by pretending *each donor* was treated. The **in-space placebo** (ADH 2010 §2.4) refits\n",
@@ -783,7 +783,7 @@
    "id": "5cb79698",
    "metadata": {},
    "source": [
-    "## 5. Philosophy B — compare across time (conformal)\n",
+    "## Philosophy B — compare across time (conformal)\n",
     "\n",
     "Philosophy A asks whether the treated state is unusual *among units*. Conformal inference\n",
     "(CWZ 2021) asks a different question: given **this** state's own time series, is the\n",
@@ -979,7 +979,7 @@
    "id": "d7f4fd90",
    "metadata": {},
    "source": [
-    "## 6. The two philosophies, side by side\n",
+    "## The two philosophies, side by side\n",
     "\n",
     "| | Philosophy A (across regions) | Philosophy B (across time) |\n",
     "|---|---|---|\n",
@@ -1000,7 +1000,7 @@
    "id": "32a59c77",
    "metadata": {},
    "source": [
-    "## 7. Practitioner takeaways\n",
+    "## Practitioner takeaways\n",
     "\n",
     "- **Reach for synthetic control** when a single (or a few) units are treated and no untreated\n",
     "  unit is a credible stand-alone control, but a *weighted blend* of donors can reproduce the\n",
diff --git a/docs/tutorials/26_composition_drift_calibration.ipynb b/docs/tutorials/26_composition_drift_calibration.ipynb
index caa3f6f1a..3824e3558 100644
--- a/docs/tutorials/26_composition_drift_calibration.ipynb
+++ b/docs/tutorials/26_composition_drift_calibration.ipynb
@@ -10,7 +10,7 @@
    "cell_type": "markdown",
    "id": "cell-01",
    "metadata": {},
-   "source": "## 1. Two regimes of non-response drift\n\n\"Do the survey weights matter?\" is really two questions - one per estimand:\n\n| Non-response drift | Descriptive estimand (a prevalence trend) | Causal DiD estimand (an ATT) |\n|---|---|---|\n| **Common** - same in treated and control units | **Biased** - raking to population margins repairs it | **Robust** - composition changes difference out ([balance's tutorial](https://github.com/facebookresearch/balance/blob/main/tutorials/balance_diff_diff_brfss.ipynb)) |\n| **Differential** - correlated with treatment x time | Biased | **Biased** - the drift *is* treatment x post shaped, so the DiD reads it as treatment effect (**this tutorial**) |\n\nThe dangerous cell is the bottom-right one, because nothing in the standard DiD toolkit flags it: the fit converges, standard errors look fine, and - as we will see - the pre-treatment event-study coefficients stay clean.\n"
+   "source": "## Two regimes of non-response drift\n\n\"Do the survey weights matter?\" is really two questions - one per estimand:\n\n| Non-response drift | Descriptive estimand (a prevalence trend) | Causal DiD estimand (an ATT) |\n|---|---|---|\n| **Common** - same in treated and control units | **Biased** - raking to population margins repairs it | **Robust** - composition changes difference out ([balance's tutorial](https://github.com/facebookresearch/balance/blob/main/tutorials/balance_diff_diff_brfss.ipynb)) |\n| **Differential** - correlated with treatment x time | Biased | **Biased** - the drift *is* treatment x post shaped, so the DiD reads it as treatment effect (**this tutorial**) |\n\nThe dangerous cell is the bottom-right one, because nothing in the standard DiD toolkit flags it: the fit converges, standard errors look fine, and - as we will see - the pre-treatment event-study coefficients stay clean.\n"
   },
   {
    "cell_type": "code",
@@ -24,7 +24,7 @@
    "cell_type": "markdown",
    "id": "cell-03",
    "metadata": {},
-   "source": "## 2. The data: a BRFSS-shaped survey around staggered smoking bans\n\nWe simulate respondent-level microdata mirroring the public-use [BRFSS](https://www.cdc.gov/brfss/annual_data/annual_2024.html) file: 50 states x 7 years (2018-2024), with 1,200 adults invited per state-year, of whom roughly 580-860 respond (~265,000 respondent rows in total). Comprehensive indoor-smoking bans take effect in 10 states in 2020 and 10 more in 2022; 30 states never adopt. The outcome is **current smoking** (binary `smoker`). Each respondent carries raking demographics (`age_band`, `educ_cat`), a complex design (`stratum`, `psu`, `fpc`), and a demographic-blind `design_weight` standing in for BRFSS `_LLCPWT`. A separate 20,000-row \"ACS\" frame supplies the population margins (and per-state adult population counts) we will rake to.\n\nThe generator is built so every headline claim is checkable against ground truth:\n\n- **No arm-specific trends by construction.** Smoking prevalence is additive and linear in education, age, and time; demographic margins are identical in every state and year. Trends differ *by demographic* (smoking falls fastest among the college-educated) but identically across treatment arms, so population parallel trends hold **in expectation**; mean-zero PSU-year shocks (kept in deliberately - they create the design effects the diagnostics section reads) add seed-specific noise but no systematic drift.\n- **The planted treatment effect is -3.0pp** in every treated state-year. One honest wrinkle: probabilities are floored at 1%, and that floor binds for ~2% of treated-post person-years (elderly college-educated respondents in low-smoking states), so the **realized population ATT is -2.98pp**. The generator computes it from the clipped potential outcomes and returns it as `truth[\"realized_att_pp\"]` - that is the truth line we hold every estimate against below.\n- **Non-response depends only on the raking observables** (`age_band`, `educ_cat`) - never on smoking itself. Response rates decline over time everywhere, concentrated among younger and less-educated adults (the real BRFSS pattern). That is the *common* drift.\n- **The differential twist** (`differential=True`): after a state's ban takes effect, response among `hs_or_less` adults falls an extra 7pp per event year - enforcement publicity raises refusals in the low-SES communities where smoking is concentrated, and the implementation consumes the follow-up-call budget. Scenario A (`differential=False`) and scenario B share the identical invited population and outcomes; they differ **only in who answers the phone**.\n\nBecause non-response is driven entirely by variables inside the raking margins (missing-at-random given the margins), calibration *can* fix everything - the interesting question is what kind of calibration. To run this on real data, replace this one cell with `pyreadstat.read_xport(...)` calls per BRFSS year.\n"
+   "source": "## The data: a BRFSS-shaped survey around staggered smoking bans\n\nWe simulate respondent-level microdata mirroring the public-use [BRFSS](https://www.cdc.gov/brfss/annual_data/annual_2024.html) file: 50 states x 7 years (2018-2024), with 1,200 adults invited per state-year, of whom roughly 580-860 respond (~265,000 respondent rows in total). Comprehensive indoor-smoking bans take effect in 10 states in 2020 and 10 more in 2022; 30 states never adopt. The outcome is **current smoking** (binary `smoker`). Each respondent carries raking demographics (`age_band`, `educ_cat`), a complex design (`stratum`, `psu`, `fpc`), and a demographic-blind `design_weight` standing in for BRFSS `_LLCPWT`. A separate 20,000-row \"ACS\" frame supplies the population margins (and per-state adult population counts) we will rake to.\n\nThe generator is built so every headline claim is checkable against ground truth:\n\n- **No arm-specific trends by construction.** Smoking prevalence is additive and linear in education, age, and time; demographic margins are identical in every state and year. Trends differ *by demographic* (smoking falls fastest among the college-educated) but identically across treatment arms, so population parallel trends hold **in expectation**; mean-zero PSU-year shocks (kept in deliberately - they create the design effects the diagnostics section reads) add seed-specific noise but no systematic drift.\n- **The planted treatment effect is -3.0pp** in every treated state-year. One honest wrinkle: probabilities are floored at 1%, and that floor binds for ~2% of treated-post person-years (elderly college-educated respondents in low-smoking states), so the **realized population ATT is -2.98pp**. The generator computes it from the clipped potential outcomes and returns it as `truth[\"realized_att_pp\"]` - that is the truth line we hold every estimate against below.\n- **Non-response depends only on the raking observables** (`age_band`, `educ_cat`) - never on smoking itself. Response rates decline over time everywhere, concentrated among younger and less-educated adults (the real BRFSS pattern). That is the *common* drift.\n- **The differential twist** (`differential=True`): after a state's ban takes effect, response among `hs_or_less` adults falls an extra 7pp per event year - enforcement publicity raises refusals in the low-SES communities where smoking is concentrated, and the implementation consumes the follow-up-call budget. Scenario A (`differential=False`) and scenario B share the identical invited population and outcomes; they differ **only in who answers the phone**.\n\nBecause non-response is driven entirely by variables inside the raking margins (missing-at-random given the margins), calibration *can* fix everything - the interesting question is what kind of calibration. To run this on real data, replace this one cell with `pyreadstat.read_xport(...)` calls per BRFSS year.\n"
   },
   {
    "cell_type": "code",
@@ -46,7 +46,7 @@
    "cell_type": "markdown",
    "id": "cell-06",
    "metadata": {},
-   "source": "## 3. The native seam: `SurveyDesign` + `aggregate_survey` + `CallawaySantAnna`\n\ndiff-diff's survey path consumes calibrated weights as *just a column*. The handoff from microdata to a modern staggered estimator is three calls:\n\n1. **`SurveyDesign(weights=..., strata=..., psu=..., fpc=...)`** - declare the complex design on the microdata (column *names*, no copies).\n2. **`aggregate_survey(micro, by=[\"state\", \"year\"], outcomes=\"smoker\", ...)`** - collapse respondents to a state-year panel of design-weighted prevalences with full Taylor-linearized precision tracking, returning the panel *and* a pre-configured second-stage `SurveyDesign` (population weights + state-level clustering, taken from the first `by` column).\n3. **`CallawaySantAnna(...).fit(panel, ..., survey_design=second_stage)`** - the staggered DiD with design-based variance.\n\nOne convention worth noticing now: we named the design columns `stratum` / `psu` / `fpc`. Those are exactly the defaults in `balance.interop.conventions.DEFAULT_DESIGN_COLUMNS`, so when we reach the balance adapter in section 8 it will wire the same design automatically.\n\nWe wrap the three calls in a helper because we are about to fit the same model under four different weight columns - that swap being a one-argument change is the point of the design.\n"
+   "source": "## The native seam: `SurveyDesign` + `aggregate_survey` + `CallawaySantAnna`\n\ndiff-diff's survey path consumes calibrated weights as *just a column*. The handoff from microdata to a modern staggered estimator is three calls:\n\n1. **`SurveyDesign(weights=..., strata=..., psu=..., fpc=...)`** - declare the complex design on the microdata (column *names*, no copies).\n2. **`aggregate_survey(micro, by=[\"state\", \"year\"], outcomes=\"smoker\", ...)`** - collapse respondents to a state-year panel of design-weighted prevalences with full Taylor-linearized precision tracking, returning the panel *and* a pre-configured second-stage `SurveyDesign` (population weights + state-level clustering, taken from the first `by` column).\n3. **`CallawaySantAnna(...).fit(panel, ..., survey_design=second_stage)`** - the staggered DiD with design-based variance.\n\nOne convention worth noticing now: we named the design columns `stratum` / `psu` / `fpc`. Those are exactly the defaults in `balance.interop.conventions.DEFAULT_DESIGN_COLUMNS`, so when we reach the balance adapter below it will wire the same design automatically.\n\nWe wrap the three calls in a helper because we are about to fit the same model under four different weight columns - that swap being a one-argument change is the point of the design.\n"
   },
   {
    "cell_type": "code",
@@ -60,7 +60,7 @@
    "cell_type": "markdown",
    "id": "cell-08",
    "metadata": {},
-   "source": "## 4. Scenario A - common drift: the descriptive trend breaks, the DiD does not\n\nFirst the regime the balance tutorial covers. Response rates decline over time, concentrated among younger and less-educated adults, but *identically* in treated and control states. The design-weighted national prevalence drifts away from the truth (the sample is increasingly educated, and educated adults smoke less - so the naive trend overstates the decline). Yet the DiD contrast is untouched, because both arms mis-measure levels the same way.\n"
+   "source": "## Scenario A - common drift: the descriptive trend breaks, the DiD does not\n\nFirst the regime the balance tutorial covers. Response rates decline over time, concentrated among younger and less-educated adults, but *identically* in treated and control states. The design-weighted national prevalence drifts away from the truth (the sample is increasingly educated, and educated adults smoke less - so the naive trend overstates the decline). Yet the DiD contrast is untouched, because both arms mis-measure levels the same way.\n"
   },
   {
    "cell_type": "code",
@@ -80,7 +80,7 @@
    "cell_type": "markdown",
    "id": "cell-11",
    "metadata": {},
-   "source": "## 5. Scenario B - differential drift: the causal estimand breaks\n\nSame states, same bans, same outcomes, same common drift - plus one mechanism: once a ban takes effect, response among `hs_or_less` adults in that state falls an extra 7pp per event year. The composition damage is easy to see if you know to look at it:\n"
+   "source": "## Scenario B - differential drift: the causal estimand breaks\n\nSame states, same bans, same outcomes, same common drift - plus one mechanism: once a ban takes effect, response among `hs_or_less` adults in that state falls an extra 7pp per event year. The composition damage is easy to see if you know to look at it:\n"
   },
   {
    "cell_type": "code",
@@ -114,7 +114,7 @@
    "cell_type": "markdown",
    "id": "cell-16",
    "metadata": {},
-   "source": "## 6. Fix attempt 1: per-wave national raking (the natural first move)\n\nThe textbook response - and the balance quickstart recipe - is to rake each survey wave to population margins: reweight so the sample's `age_band` x `educ_cat` distribution matches the ACS frame, year by year. We wrap balance's `Sample -> set_target -> adjust(method=\"rake\")` in a helper that (a) rakes within each cell of a chosen `granularity`, and (b) rescales each cell's raked weights to its population **count** - in real practice raking targets are population counts, so weight totals carry the state scale rather than balance's internal per-call normalization.\n\nOne mechanical detail: balance casts ids to `str`, so we align the returned weights back to the original rows by string id and assert nothing went missing.\n"
+   "source": "## Fix attempt 1: per-wave national raking (the natural first move)\n\nThe textbook response - and the balance quickstart recipe - is to rake each survey wave to population margins: reweight so the sample's `age_band` x `educ_cat` distribution matches the ACS frame, year by year. We wrap balance's `Sample -> set_target -> adjust(method=\"rake\")` in a helper that (a) rakes within each cell of a chosen `granularity`, and (b) rescales each cell's raked weights to its population **count** - in real practice raking targets are population counts, so weight totals carry the state scale rather than balance's internal per-call normalization.\n\nOne mechanical detail: balance casts ids to `str`, so we align the returned weights back to the original rows by string id and assert nothing went missing.\n"
   },
   {
    "cell_type": "code",
@@ -134,7 +134,7 @@
    "cell_type": "markdown",
    "id": "cell-19",
    "metadata": {},
-   "source": "## 7. Fix attempt 2: rake each state-year to the margins\n\nSame helper, `granularity=[\"state\", \"year\"]` - 350 small rakes (~750 respondents each, a couple of seconds total), each rescaled to its state's adult-population count from the ACS frame. State totals are then constant across years by construction, so the second-stage population weights recover the true state scale, uncontaminated by response-rate levels.\n"
+   "source": "## Fix attempt 2: rake each state-year to the margins\n\nSame helper, `granularity=[\"state\", \"year\"]` - 350 small rakes (~750 respondents each, a couple of seconds total), each rescaled to its state's adult-population count from the ACS frame. State totals are then constant across years by construction, so the second-stage population weights recover the true state scale, uncontaminated by response-rate levels.\n"
   },
   {
    "cell_type": "code",
@@ -154,7 +154,7 @@
    "cell_type": "markdown",
    "id": "cell-22",
    "metadata": {},
-   "source": "## 8. The adapter: `balance.interop.diff_diff`\n\nEverything above used the native seam - calibrated weights as a column. balance 0.21 ships an adapter that packages the same handoff when your object in hand is a balance `Sample`:\n\n- **`bd.to_panel_for_did(sample, by=, outcomes=)`** builds the first-stage `SurveyDesign` from the Sample's *active* weight column (auto-wiring our `stratum`/`psu`/`fpc` convention columns), strips balance's bookkeeping columns, and calls `diff_diff.aggregate_survey` - returning the same `(panel, second_stage_design)` pair.\n- **`bd.fit_did(sample, estimator=, ...)`** resolves any diff-diff estimator by name (or short alias: `\"CS\"`, `\"DiD\"`, `\"BJS\"`, `\"HAD\"`), splits your kwargs between `__init__` and `fit()` by signature, forwards `survey_design=`, and attaches the source Sample to the result as `_balance_adjustment` for provenance.\n\nBoth paths must agree exactly - and we assert it:\n"
+   "source": "## The adapter: `balance.interop.diff_diff`\n\nEverything above used the native seam - calibrated weights as a column. balance 0.21 ships an adapter that packages the same handoff when your object in hand is a balance `Sample`:\n\n- **`bd.to_panel_for_did(sample, by=, outcomes=)`** builds the first-stage `SurveyDesign` from the Sample's *active* weight column (auto-wiring our `stratum`/`psu`/`fpc` convention columns), strips balance's bookkeeping columns, and calls `diff_diff.aggregate_survey` - returning the same `(panel, second_stage_design)` pair.\n- **`bd.fit_did(sample, estimator=, ...)`** resolves any diff-diff estimator by name (or short alias: `\"CS\"`, `\"DiD\"`, `\"BJS\"`, `\"HAD\"`), splits your kwargs between `__init__` and `fit()` by signature, forwards `survey_design=`, and attaches the source Sample to the result as `_balance_adjustment` for provenance.\n\nBoth paths must agree exactly - and we assert it:\n"
   },
   {
    "cell_type": "code",
@@ -174,7 +174,7 @@
    "cell_type": "markdown",
    "id": "cell-25",
    "metadata": {},
-   "source": "## 9. Does the fix depend on the estimator?\n\nA tempting escape: \"maybe a more robust estimator handles it.\" It cannot - the composition artifact lives in the *data*, in the treatment x post cell, upstream of any identification strategy. `fit_did`'s one-string estimator swap makes the sweep trivial:\n"
+   "source": "## Does the fix depend on the estimator?\n\nA tempting escape: \"maybe a more robust estimator handles it.\" It cannot - the composition artifact lives in the *data*, in the treatment x post cell, upstream of any identification strategy. `fit_did`'s one-string estimator swap makes the sweep trivial:\n"
   },
   {
    "cell_type": "code",
@@ -194,7 +194,7 @@
    "cell_type": "markdown",
    "id": "cell-28",
    "metadata": {},
-   "source": "## 10. Diagnostics across the seam\n\nEach package owns half the diagnostic picture. diff-diff's `survey_metadata` describes the *second stage* - how much precision the population weighting costs across states. balance's diagnostics describe each *raking cell* - how hard the calibration had to work, which is itself an early-warning signal: the design effect of the rake blows up exactly where composition was most distorted. `bd.as_balance_diagnostic` joins the two into one flat dict.\n"
+   "source": "## Diagnostics across the seam\n\nEach package owns half the diagnostic picture. diff-diff's `survey_metadata` describes the *second stage* - how much precision the population weighting costs across states. balance's diagnostics describe each *raking cell* - how hard the calibration had to work, which is itself an early-warning signal: the design effect of the rake blows up exactly where composition was most distorted. `bd.as_balance_diagnostic` joins the two into one flat dict.\n"
   },
   {
    "cell_type": "code",
@@ -208,13 +208,13 @@
    "cell_type": "markdown",
    "id": "cell-30",
    "metadata": {},
-   "source": "Read the contrast: in a never-treated state the 2024 rake barely works (design effect near 1), while in a 2020-cohort state it pays a large design effect to rebuild the missing low-education mass - and post-adjustment imbalance (`asmd_max_post`) is ~0 in both, confirming the margins were hit. **Rising per-cell raking design effects concentrated in treated-post cells are the operational fingerprint of differential drift.**\n\nThe practical monitoring rule this tutorial suggests: alongside your pre-trend plot, always plot **design-weighted demographic shares by cohort over time** (the section 5 plot) and the **per-cell raking design effects**. Both are one groupby away, and they catch what the pre-trend test structurally cannot.\n"
+   "source": "Read the contrast: in a never-treated state the 2024 rake barely works (design effect near 1), while in a 2020-cohort state it pays a large design effect to rebuild the missing low-education mass - and post-adjustment imbalance (`asmd_max_post`) is ~0 in both, confirming the margins were hit. **Rising per-cell raking design effects concentrated in treated-post cells are the operational fingerprint of differential drift.**\n\nThe practical monitoring rule this tutorial suggests: alongside your pre-trend plot, always plot **design-weighted demographic shares by cohort over time** (the Scenario B plot) and the **per-cell raking design effects**. Both are one groupby away, and they catch what the pre-trend test structurally cannot.\n"
   },
   {
    "cell_type": "markdown",
    "id": "cell-31",
    "metadata": {},
-   "source": "## 11. When do you need calibration? A checklist\n\n| Your estimand | Drift type | What you need |\n|---|---|---|\n| Descriptive (levels, trends) | any | Calibrated weights, per wave ([balance tutorial](https://github.com/facebookresearch/balance/blob/main/tutorials/balance_diff_diff_brfss.ipynb)) |\n| Causal DiD | common across arms | Design weights suffice for the point estimate; calibration still improves descriptives and SEs |\n| Causal DiD | **correlated with treatment timing** | **Calibration at the granularity of your comparison units** (this tutorial) |\n\nThree caveats that decide whether raking is *enough*:\n\n- **Margins must come from a policy-unaffected source.** We raked to ACS-style demographics. If the policy itself changed the margins you rake to, calibration bakes the effect away.\n- **Raking fixes drift on observables inside your margins.** Here non-response depended only on `age_band` x `educ_cat` - missing-at-random given the margins - so raking was exact. If response depends on the *outcome itself within* demographic cells (heavier smokers refusing regardless of education), no reweighting on demographics can fix it. That regime is the subject of Sant'Anna & Xu (2023), \"Difference-in-Differences with Compositional Changes\" ([arXiv:2304.13925](https://arxiv.org/abs/2304.13925)) - a model-based route diff-diff tracks on its roadmap.\n- **Drift can pre-date adoption.** Our mechanism switched on at the effective date, which is why the pre-trends stayed clean. As an exercise, rerun with `simulate_brfss_smoking(differential=True, drift_start_offset=-2)` - drift beginning with the legislative campaign two years early - and watch the design-weight pre-treatment coefficients light up. When your event study *does* flag pre-trends in survey data, composition drift belongs on the suspect list right next to genuine trend violations.\n"
+   "source": "## When do you need calibration? A checklist\n\n| Your estimand | Drift type | What you need |\n|---|---|---|\n| Descriptive (levels, trends) | any | Calibrated weights, per wave ([balance tutorial](https://github.com/facebookresearch/balance/blob/main/tutorials/balance_diff_diff_brfss.ipynb)) |\n| Causal DiD | common across arms | Design weights suffice for the point estimate; calibration still improves descriptives and SEs |\n| Causal DiD | **correlated with treatment timing** | **Calibration at the granularity of your comparison units** (this tutorial) |\n\nThree caveats that decide whether raking is *enough*:\n\n- **Margins must come from a policy-unaffected source.** We raked to ACS-style demographics. If the policy itself changed the margins you rake to, calibration bakes the effect away.\n- **Raking fixes drift on observables inside your margins.** Here non-response depended only on `age_band` x `educ_cat` - missing-at-random given the margins - so raking was exact. If response depends on the *outcome itself within* demographic cells (heavier smokers refusing regardless of education), no reweighting on demographics can fix it. That regime is the subject of Sant'Anna & Xu (2023), \"Difference-in-Differences with Compositional Changes\" ([arXiv:2304.13925](https://arxiv.org/abs/2304.13925)) - a model-based route diff-diff tracks on its roadmap.\n- **Drift can pre-date adoption.** Our mechanism switched on at the effective date, which is why the pre-trends stayed clean. As an exercise, rerun with `simulate_brfss_smoking(differential=True, drift_start_offset=-2)` - drift beginning with the legislative campaign two years early - and watch the design-weight pre-treatment coefficients light up. When your event study *does* flag pre-trends in survey data, composition drift belongs on the suspect list right next to genuine trend violations.\n"
   },
   {
    "cell_type": "markdown",
diff --git a/docs/tutorials/27_cic_distributional_effects.ipynb b/docs/tutorials/27_cic_distributional_effects.ipynb
index 827f3394f..3ab75f5fc 100644
--- a/docs/tutorials/27_cic_distributional_effects.ipynb
+++ b/docs/tutorials/27_cic_distributional_effects.ipynb
@@ -34,7 +34,7 @@
    "id": "9854eb19",
    "metadata": {},
    "source": [
-    "## 1. When is CiC the right tool?\n",
+    "## When is CiC the right tool?\n",
     "\n",
     "The decision comes down to the *design* and the *question*:\n",
     "\n",
@@ -47,7 +47,7 @@
     "- **Staggered adoption across many periods** - CiC does not apply (it is 2x2 only);\n",
     "  use `CallawaySantAnna` or another heterogeneity-robust estimator for mean effects.\n",
     "- **QDiD** is the quantile-by-quantile DiD comparison estimator. Athey & Imbens (2006,\n",
-    "  p. 447) recommend CiC over it - Section 6 makes their reason concrete on our data.\n",
+    "  p. 447) recommend CiC over it - the levels-vs-logs section makes their reason concrete on our data.\n",
     "\n",
     "Our setting is a genuine 2x2 with **repeated cross-sections**: a fresh sample of\n",
     "customers is drawn each period (typical for transaction panels). CiC also supports\n",
@@ -82,7 +82,7 @@
    "id": "a6189b31",
    "metadata": {},
    "source": [
-    "## 2. The data\n",
+    "## The data\n",
     "\n",
     "We simulate the program so that the truth is known by construction - every claim\n",
     "below can be checked against it. Three ingredients, each realistic on its own:\n",
@@ -221,7 +221,7 @@
    "id": "c7d7cee3",
    "metadata": {},
    "source": [
-    "## 3. The mean answer\n",
+    "## The mean answer\n",
     "\n",
     "Standard practice first: difference-in-differences on mean spend, with robust\n",
     "standard errors.\n"
@@ -298,7 +298,7 @@
    "id": "1e750447",
    "metadata": {},
    "source": [
-    "## 4. Changes-in-Changes: the full picture\n",
+    "## Changes-in-Changes: the full picture\n",
     "\n",
     "CiC (Athey & Imbens 2006) treats the control group's period-to-period change as a\n",
     "*transformation of the whole distribution* and applies that transformation to the\n",
@@ -307,7 +307,7 @@
     "treated customer the outcome move that a control customer at the same rank\n",
     "experienced.\" Identification needs a monotone outcome model in a scalar unobservable\n",
     "and time-invariance of that unobservable within groups - *not* parallel mean trends\n",
-    "(more on that in Section 6).\n",
+    "(more on that in the levels-vs-logs section).\n",
     "\n",
     "Inference is bootstrap-only (999 replicates here, seeded). The point estimates match\n",
     "the R `qte` package exactly; see the methodology registry for the parity details.\n"
@@ -446,7 +446,7 @@
    "id": "a655da72",
    "metadata": {},
    "source": [
-    "## 5. Which quantiles moved? Joint claims and the interior range\n",
+    "## Which quantiles moved? Joint claims and the interior range\n",
     "\n",
     "Reading 19 pointwise confidence intervals off the plot and declaring \"significant\n",
     "below the median\" is a multiple-testing mistake - pointwise intervals over-reject\n",
@@ -621,7 +621,7 @@
    "id": "2b463a4c",
    "metadata": {},
    "source": [
-    "## 6. Scale is not a nuisance: levels, logs, and why CiC doesn't care\n",
+    "## Scale is not a nuisance: levels, logs, and why CiC doesn't care\n",
     "\n",
     "Every mean-DiD practitioner has faced the question: *levels or logs?* Parallel\n",
     "trends on one scale generally fails on the other, so the modeling choice changes the\n",
@@ -742,7 +742,7 @@
     "round-trip, at most a few units in the last bit - on this machine it is exactly\n",
     "zero). The levels-vs-logs debate simply does not arise for CiC's distributional\n",
     "answer. (The scope matters: this exact equivariance holds for *unconditional* fits -\n",
-    "the covariate branch in Section 7 runs linear quantile regressions, which are not\n",
+    "the covariate branch below runs linear quantile regressions, which are not\n",
     "equivariant to nonlinear transforms. And means never commute with transforms, so the\n",
     "ATT is scale-specific for every estimator, CiC included.)\n",
     "\n",
@@ -826,7 +826,7 @@
    "id": "71f41d36",
    "metadata": {},
    "source": [
-    "## 7. When the mix differs: covariates\n",
+    "## When the mix differs: covariates\n",
     "\n",
     "CiC's time-invariance assumption can fail *observably*: if the treated and control\n",
     "groups differ in composition on a variable that also drives the period-to-period\n",
@@ -929,7 +929,7 @@
    "id": "8caefb49",
    "metadata": {},
    "source": [
-    "## 8. What next\n",
+    "## What next\n",
     "\n",
     "Fitted results know their own follow-up work: `practitioner_next_steps()` reads the\n",
     "estimator type, covariate status, and bootstrap health, and returns the remaining\n",
diff --git a/docs/tutorials/README.md b/docs/tutorials/README.md
index 7024dad45..19c1d0abc 100644
--- a/docs/tutorials/README.md
+++ b/docs/tutorials/README.md
@@ -85,7 +85,7 @@ Practitioner walkthrough for marketing analytics teams measuring geo-experiment
 - Synthetic panel of 80 markets with simulated campaign launch
 - `SyntheticDiD` fit, diagnostics, and inference (placebo + bootstrap)
 - Unit weights and time weights interpretation
-- Stakeholder communication template (Tutorial 17 Section 9 pattern)
+- Stakeholder communication template (Tutorial 17 leadership-communication pattern)
 
 ### 19. dCDH Marketing Pulse Campaigns (`19_dcdh_marketing_pulse.ipynb`)
 Practitioner walkthrough for measuring lift from on/off promotional pulses across markets, where treatment can switch in both directions: