JoeWhiteJr/ooa-hypothesis

Out-of-Africa migration distance as a continuous encoding of ancestry — methodological replication across 16 datasets # OOA Hypothesis: Out-of-Africa Migration Distance as a Continuous Encoding of Ancestry A methodological replication project testing whether **route-aware migration distance from East Africa** (Addis Ababa, with Beringia + Sundaland waypoints) functions as a useful continuous covariate across health, behavioral, and societal outcomes — and whether it outperforms naive baselines like alphabetical ordering. ## What's here - **`OOA_explained.md`** — full walkthrough of the project (what we asked, what we tested, what we found, what the theory is, where it fails). Casual / conversational. - **`findings_summary.md`** — formal 3-page review document. Page 1-2: theory + headline findings. Page 3: dense analysis tables. This is the version for collaborators to mark up. - **`ramach_codes.txt`**, **`ramach_coords.txt`** — Ramachandran et al. (2005) population codes and coordinates used to validate the route-distance model. - **`rosenberg_freqs.txt`** — Rosenberg et al. STRUCTURE cluster frequencies used in early sanity checks. - **`data/FETCH.md`** — where every raw dataset came from. Re-run the fetches to reproduce. ## What's NOT here (and why) All raw data dumps (NHANES `.XPT`, World Bank `.json`, WHO `.json`, NCD-RisC `.csv`, WHR `.xls`, Wikipedia HTML scrapes) are `.gitignore`'d. Reasons: 1. Most have non-trivial redistribution terms (WHR, WHO). 2. Everything is re-fetchable from stable public URLs (see `data/FETCH.md`). 3. ~164 MB of frozen JSON would slow collaborator clones for no scientific gain. ## Headline findings (very short) - Route-distance vs HGDP microsatellite heterozygosity: **r ≈ -0.87** (validates the encoding against the genetic ground truth it's meant to proxy). - Beats permutation null on multiple cross-sectional outcomes (obesity widening, life expectancy, fertility, internet penetration). - Robust to a 4-covariate confound stack (log GDP + |latitude| + urbanization % + colonial-history binary) on some DVs, collapses on others — and the *patter …