2D Histograms - Hanging Bars
The hanging bars histogram
offers a "visual test of normality" of the distribution that
helps identify the areas of the distribution where the discrepancies (between
the observed and expected normal frequencies) occur. While the standard
way of presenting the normal distribution fitted to an observed distribution
is to overlay the best-fitting normal curve over a histogram, the hanging
bars histogram does just the opposite: it suspends the bars representing
the observed frequencies for consecutive ranges of values from the best-fitting
normal curve. If the investigated distribution can be well approximated
by the normal curve, the bottoms of all bars should form a straight, horizontal
line. If more than one variable is selected, then one graph will be produced
for each variable selected.

See also, 2D Histograms
and Graphs
- 2D Histograms.