Inferential statistics permit experimenters to determine whether their findings can be generalized from their samples to the populations they represent. Different tests are used depending on the number of means being compared, whether the means come from the same or different groups, and how many independent variables were used in the study. The statistics resulting from these tests are compared to critical values to determine whether they are likely to have been obtained by chance. Consider a simple experiment in which an experimental group that is exposed to a condition is compared to a control group that is not. For the difference between the means of the two groups to be statistically significant, the difference must have a low probability (usually less than 5 percent) of occurring by normal random variation. Improving the power of a study by measuring more participants or improving measurement techniques will increase the likelihood that the study will allow rejection of a false null hypothesis, thus reducing the probability of a Type II error.