If multicollinearity inflates standard errors, what is the practical impact on confidence intervals for affected coefficients?

• A. Wider confidence intervals
• B. Narrower confidence intervals
• C. No change to confidence intervals
• D. Confidence intervals become invalid

Answer: (A)

Multicollinearity inflates the standard errors of the coefficient estimates, and the width of a confidence interval is driven by that standard error. As the standard error grows, the margin of error around the estimated coefficient increases, so the confidence interval becomes wider. This means less precision in pinpointing the exact effect of a predictor, and it can make it harder to show that the effect is significantly different from zero or to determine its direction with confidence. In other words, the practical impact is greater uncertainty about the coefficient’s true value.

If people wonder about the other possibilities: narrowing the interval would require a smaller standard error, which isn’t what multicollinearity does. No change would ignore the inflated uncertainty, and calling them invalid would only apply in the extreme case of perfect multicollinearity, whereas typical multicollinearity enlarges the interval but keeps the estimate usable.