Dependence you should expect: panel data | Post hoc, without refitting | Clustering repairs inference, not efficiency | Choice-based sampling | References

The data | Fit a multinomial logit | Predicted shares | Elasticities and diversion | Willingness to pay | A policy counterfactual | One interface, every model | Choosing among choice models | How choicer compares | References

Simulate correlated random coefficients | Fit | Parameter recovery | Substitution through taste heterogeneity | Identification and tails | Panel data: one taste draw per decision maker | Many alternatives: constants, convergence and standard errors | A research checklist | References

Build a population | Draw a choice-based sample | Weighted estimation and inference | Starting from an existing sample | The shipped mode_choice data | References

Simulate a panel | Fit | Convergence is part of the result | Did we recover the truth? | Willingness to pay and predicted shares | A policy counterfactual | An entry counterfactual | Price endogeneity | The probit sibling | Further reading | References

Simulate from a probit process | Run the sampler | Compare posterior summaries with the truth | A quick MCMC diagnostic | What this fit does not provide | References

Simulate a known data-generating process | Fit | Did we recover the truth? | Post-estimation | Substitution restrictions | Empirical checklist | References

Simulate a nested process | Fit | Parameter recovery | What identifies the dissimilarity parameters? | Nest-consistent elasticities | Share inversion with BLP | Empirical checklist | References

Capturing lineage without taps | Three ways in | What a snapshot knows | Lifecycle events | Detecting value-only changes | Lineage resolution limits | Exporting the report

Tracking what happens in your pipeline | Your first trail | Operation-aware taps | Join taps | Filter taps | Comparing snapshots | Full audit report | Domain-specific diagnostics | Snapshot controls | Tabulation in pipelines | Standalone mode | Exporting and sharing trails | Converting to R objects | Saving and loading trails | HTML visualization

Before you join | Different key names | Tracking the impact on a metric | Are your keys unique? | Beyond uniqueness: what's the relationship? | What changed between versions? | Filter with visibility | filter_keep | filter_drop | Warning thresholds | Diagnose data quality | Missing values | Column summaries | Frequency tables | One-way tables | Sorting and cutoffs | Two-way crosstabulations | Weighted tabulation | String quality | Audit your transformations | Character vectors | Numeric vectors | Date vectors | Factor vectors | Logical vectors

Example data | Step 1: Validate primary keys | Step 2: Validate the join | Step 3: Filter with diagnostics | Step 4: Compare before and after | Step 5: Check data quality | Missing values | String quality | Date coverage | Step 6: Summarize columns | Auditing cleaning operations | Further reading