“Squeeze” Significance Tests

Summary A new test statistic is described and defined. This is called the “squeeze” statistic, its name deriving from the method used to perform the test, whereby data points plotted on the appropriate probability paper are squeezed between parallel rules. The narrowest setting of the rulers, such that no points are completely covered, gives rise to a value called the squeeze statistic. The distribution of the squeeze statistic for the normal distribution, and for certain alternative distributions, has been determined by Monte Carlo methods, and is tabulated for a range of sample sizes. The squeeze statistic has the merit that it is a practical and visual test that can be carried out very quickly on plotted data.