Laboratory of Synapse Structure and Function
About Anatomy Links Publications Search Tools

KSTAT

JCF 1/15/01

KSTAT is a Windows (Win32) application for visually and statistically comparing two data sets.

Download Version 0.5a (39K zip) [Last update: Jan 15, 2001]

Description:

A data set is an unsorted list of single real numbers. Data sets are loaded in KSTAT one at a time. Data can either be imported from a text file, or pasted from another application such as a spreadsheet column. A data set is sorted and displayed in the window as a frequency histogram from the minimum to maximum values. You can adjust the histogram bin size interactively by using the up and down arrow keys.

When two data sets are loaded, the background will indicate rejection (red) or acceptance (green) of the null hypothesis that the data sets come from the same distribution, using a significance level of 0.005. Detailed information on the statistical test can be obtained from the Statistics menu item.

By default, KSTAT applies the Kolmogorov-Smirnov test to compare distributions. The Kolmogorov-Smirnov distance is the maxmimum value of the absolute difference of the cumulative distributions of the unbinned data sets. For more about this test see Numerical Recipes in C, Section 14.3 pp. 623-626. An alternative distance measure can be used to compare the data sets by selecting Kuiper's test from the Options Menu. Kuiper's distance is the sum of the absolute values of the maximum positive and negative differences between cumulative distributions. (See Numerical Recipes in C pp. 626-627.) While the Kolmogorov-Smirnov statistic is good at finding shifts between distributions, especially changes in the median value, it is not so good at finding spreads that affect the tails of a distribution but leave the median unchanged. Kuiper's statistic can provide a better test for spreads.

Return to Protocols

Last Updated: 1/15/01