Set out the kinds of heteroskedasticity and their consequences for statistical inference

Heteroskedasticity means the error variance is not constant. Unconditional heteroskedasticity is unrelated to the independent variables, leaves coefficients unbiased and is not treated as a serious problem, although strictly the conventional standard-error formula assumes constant variance, so robust errors are the safe choice. Conditional heteroskedasticity, where the variance moves with the independent variables, leaves coefficients unbiased but makes standard errors unreliable, usually too small, so t-tests and the F-test reject too often. Detect it with the Breusch–Pagan test: regress the squared residuals on the independent variables and compare n × R² from that regression with a chi-square critical value with k degrees of freedom. Correct it with robust (White-corrected) standard errors. The traps are using the original regression's R², or believing the coefficients themselves are biased.