What Do You Like? — Preference Prediction & False Consensus (Paper)

This repository hosts a paper-only case study on how consumers predict others’ choices and the role of false consensus / projection.

Contents

• reports/paper.pdf — Full paper (methods, results, discussion)
• abstracts/executive_summary.md — 1-page summary for quick review

One-minute summary

• Question: How well do people predict others’ choices, and do they project their own tastes?
• Design: N=32 Google Form with own-choice + predicted group-share per domain. Bias = predicted − actual; false-consensus = supporters’ mean prediction − non-supporters’.
• Findings: Under-estimation of majorities (Pretzels 53.1%, Coke 56.3%) by ~−2.6 pp, and over-estimation of Pop (40.6% actual vs 50.0% predicted, +9.4 pp). False-consensus deltas ≈ +8–12 pp.
• Implications: Adjust gut estimates (≈±10 pp), randomize question order, and favor measured panels.
• Limits: Small local sample; binary categories; no confidence calibration.

Citation

Armstrong, I. (2025). What Do You Like? A Study on Consumers’ Predicting Other People’s Preferences. Marketing 301 – Consumer Analysis.

Last updated: 2025-09-20