“The closer to the train station, the worse the kebab” – a “study”

Original Article ↗ Hacker News Discussion ↗

Study result vs. expectations

  • Many commenters initially assumed the study had confirmed the aphorism; others repeatedly pointed out that the author did not find a meaningful correlation between kebab rating and distance from stations (Pearson ≈ 0.09, statistically very weak).
  • Some argued the title is misleading because it sounds like a confirmed result rather than a mostly-null finding.
  • A few tried to reinterpret the plots visually, claiming patterns the statistical summary does not strongly support.

Google reviews as a proxy for “quality”

  • Strong debate over whether Google ratings meaningfully measure food quality:
    • Critics: ratings are noisy; they mix food, service, ambience, delivery issues, tourist expectations, and even one-off rage or hype.
    • Defenders: with 50–100+ reviews, averages often become surprisingly reliable, though cultural taste differences and tourist-heavy areas can skew scores.
  • Several note that the study really measures correlation with Google ratings, not intrinsic kebab quality.
  • Suggested improvements: classify review text (possibly via LLMs) to separate food-specific sentiment from other factors, or focus on proportion of food-related complaints.

Stations: metro vs. “real” train

  • Multiple people argue the original French saying refers to big intercity “gares,” not dense metro networks.
  • In Paris, almost everything is near a metro stop, so including metros dilutes any effect. Filtering to train-only still showed little change, but many want a re-run limited to major rail hubs and other cities.

Interpreting the scatter and possible one‑way effect

  • Some see an empty “far & bad” quadrant as suggestive: very bad kebabs appear only near stations, whereas far-away shops have at least decent minimum quality.
  • Others counter that this can arise from selection/collider bias (only surviving businesses are analyzed) and human pattern-finding in noise.

Restaurant economics and rules of thumb

  • Popular heuristic: restaurants trade off quality, location, and price—near-station spots pay high rent and rely on captive or transient customers, so can survive with mediocre food.
  • Counter-argument: higher revenue potential in good locations could also support better food; reality depends on tourist vs. commuter mix and competitive pressure.

OP clarifications and future work

  • The author notes the post began as a tongue-in-cheek “meme study,” acknowledges linear correlation may be the wrong test, and plans a follow-up:
    • More cities (Berlin, London, Stockholm mentioned).
    • Possibly non-linear or quantile approaches, and better distinction between metro and major train stations.