ShotStat provides two tools for statistical hypothesis testing: the T-Test and the F-Test. ``Statistical Hypothesis Testing'' is a fancy way of saying you may wish to know if two sets of data are different or not, within the framework of the random errors in each set. For example, the T-Test (10.3) is used to determine if two averages are significantly different. The averages you may wish to test may be average point of impact for two shot groups, average case weight for two lots (or brands) of cases, etc. The Null Hypothesis (10.3) that you are testing in a T-Test can be stated as ``the sample averages are not significantly different.''
The F-Test (10.3) is used similarly to evaluate if the standard deviations or variances of two samples are the same. If you wish to know if the group size for two Lots of ammunition are significantly different, you should use the F-Test. In this case, the Null Hypothesis being tested can likewise be worded: ``the sample standard deviations are not significantly different.''
It is important to understand what these tests can tell you. Essentially, you want a good guess as to whether the populations for each sample are different. The Null Hypothesis says these populations are the same. This means that your samples are two samples from the same population if the Null Hypothesis is true. In other words, any difference in the averages (or standard deviation) is due to the errors in measurement, not due to the samples being representative of populations that are different. With this kind of test, the handloader can put a quantitative value to the statement Load A is better than Load B. This quantitative value is the Significance Level (See 10.3).
For both types of test, the user needs to decide the Significance Level before the test is conducted. The Significance Level is the chance of incorrectly rejecting the Null Hypothesis. Typical values are 10%, 5% and 2.5%, though you can select any value between 0.001% and 45% using ShotStat.