quant-model-multinomiallisted

# quant-model-multinomial — Multinomial Logistic Regression ## 1. Objective Fit a multinomial logistic regression model predicting an unordered categorical outcome (3+ categories) from one or more predictors. Uses `nnet::multinom` in R with standard Wald standard errors. Outputs K−1 sets of odds ratios (one per non-reference category), fit indices, predicted probabilities per category, assumption flags, and a dual-tab HTML report. **Route here when:** - The dependent variable has 3+ categories with **no inherent order** (e.g., market: US/EU/APAC, brand preference: Nike/Adidas/Puma, channel: online/store/app) - If the categories **are ordered** (e.g., Likert scale), use `quant-model-ordinal` instead --- ## 2. Pre-flight Required columns in the input CSV: - `outcome_col`: categorical with 3–10 unique values (nominal — no ranking implied) - `predictor_cols`: one or more columns (numeric or categorical) Minimum n: 50 complete cases; at least 10 per outcome category recommended. **⚠ IIA assumption:** Multinomial logistic regression assumes Independence of Irrelevant Alternatives (IIA) — adding or removing an outcome category should not change the relative odds between remaining categories. This is a theoretical constraint, not tested here; documented as a caveat in the report. --- ## 3. AskUserQuestion protocol ### Q1 — Input file ``` Which file contains your data? Options: [user provides path] ``` ### Q2 — Outcome column ``` Which column is your outcome variable? (It