Check your understanding
Quiz
You have worked through the modern regression toolkit: robust and quantile methods that resist outliers, ridge and lasso shrinkage, flexible GAMs, count models for Poisson and negative-binomial data, zero-inflated and hurdle models, Gamma and Tweedie for amounts and losses, beta and ordinal models for bounded outcomes, and mixed models for grouped data. This quiz checks what stuck. The last two steps are live R you can run.
Question 1 of 10
When one point tilts the line
A single gross outlier in y has dragged your OLS slope far from the trend the bulk of the data follows. Why does an M-estimator like rlm handle this better?
Correct: OLS squares every residual, so one huge error dominates. An M-estimator uses a loss that grows more slowly, so extreme points pull far less.
The whole point is a different, outlier-resistant answer, not a faster OLS.