{
  "$type": "site.standard.document",
  "bskyPostRef": {
    "cid": "bafyreiba3oi2ne3fn7733y6lhxjxlupub3naq6aquw54pzhbqxbhmze3oy",
    "uri": "at://did:plc:wwyqal4cnqhuwyacdj7rqq3n/app.bsky.feed.post/3micr6atrii62"
  },
  "path": "/t/clustering-in-the-denominator-non-independence-of-starts-in-racing-fatality-studies/28662#post_4",
  "publishedAt": "2026-03-30T13:52:01.000Z",
  "site": "https://discourse.datamethods.org",
  "textContent": "I might be thinking of this wrong but my first thought would be to treat each start as an exposure and have n horses with different numbers of races so you’d get something like an exponential survival model? Something like death ~ XB + offset(starts)?",
  "title": "Clustering in the denominator: non-independence of starts in racing fatality studies"
}