Abstract
In this paper, we compare nonparametric kernel estimates with smoothed histograms as methods for displaying logarithmically transformed dwell-time distributions. Kernel density plots provide a simpler means for producing estimates of the probability density function (pdf) and they have the advantage of being smoothed in a well-specified, carefully controlled manner. Smoothing is essential for multidimensional plots because, with realistic amounts of data, the number of counts per bin is small. Examples are presented for a 2-dimensional pdf and its associated dependency-difference plot that display the correlations between successive dwell times.