True or False?

(a) Robust regression is primarily for variables that are nonlinearly related; (b) if the Y errors are nonnormal and/or outliers are present, the KTRL provides an alternative to data transformation; (c) as with OLS regression, KTRL can be used to predict the mean Y value for specified X value; (d) the main idea behind weighted regression is that observations that have greater variance or uncertainty should be down-weighted in the computation of the regression coefficients; (e) although primarily intended for limiting the influence of outliers, iteratively reweighted least squares regression can also help in the case of nonconstant error variance by down-weighting the influence of observations with larger residuals (i.e., that have greater variance); (f) when the regression assumption of independence and/or homoscedasticity is violated, it is still possible to compute conventional OLS regression, but then use robust standard errors for the tests of significance; (g) a regression method is described as “bounded influence” regression because it limits the potential influence of outliers on the regression results; (h) bounded influence regression methods are typically characterized as having a “breakdown point” of 50% because up to 50% of the data points can be outliers without affecting the regression results; (i) LMS and LTS are both bounded influence regression methods, but LMS is based on the median while LTS is based on the trimmed mean; (j) since OLS regression can be affected by even a single outlier, it has a breakdown point of 100%.