Regression Models in UX Research - Cheatsheet for Practitioners
·
1 min
Key Decision Flowchart for UX Researchers
1. Identify your outcome variable.
| Outcome |
Recommended model |
| Binary |
Logistic Regression |
| Ordered categories |
Ordinal Regression |
| Unordered categories (>2) |
Multinomial Regression |
| Counts |
Check for overdispersion → Negative Binomial |
| Proportion (0–1) |
Beta Regression |
| Continuous |
Proceed to step 2 |
2. Examine your data structure.
| Data issue |
Recommendation |
| Repeated measures / clustered data |
Use mixed‑effects (multilevel) version of the chosen model (random intercepts/slopes) |
| Time‑based observations |
Use time‑series or temporal models (ARIMA, state‑space, or models with time terms) |
| Many zeros |
Consider zero‑inflated or hurdle models (ZIP/ZINB, ZINegBin) |
| Outliers / skewed distributions |
Apply transformations or use robust or quantile regression |
3. Define your analytical goal.
| Analytical goal |
Recommendation |
| Explain drivers |
Use interpretable models (Linear, Logistic, etc.) |
| Predict accurately |
Consider regularized or machine‑learning models (Lasso, GAMs, Boosting) |
| Inform design decisions |
Prioritize models with communicable outputs (odds ratios, probabilities) |
Regression Models Cheatsheet in UX Research
Image created by Gemini 3
Refferences
