CONFIDENCE BANDS IN NONPARAMETRIC REGRESSION

New bias-corrected confidence bands are proposed for nonparametric ker nal regression. These bands are constructed using only a kernel estima tor of the regression curve and its data-selected bandwidth. They are shown to have asymptotically correct coverage properties and to behave well in a small-sample study. One consequence of the large-sample dev elopments is that Bonferroni-type bands for the regression curve at th e design points also have conservative asymptotic coverage behavior wi th no bias correction.