Q. What types of statistical tests should I use with the different data types?
While the data type is relevant in determining what statistical test to use, it’s equally (if not more) important to consider the research context of the problem.
Suppose you are in charge of two studies. In one study, you want to compare the weights of participants before and after participating in a dieting program. In the other, you want to compare the average weight difference between men and women. In both cases, weight has a numeric (more specifically, ratio) data type. However, you would apply a paired t-test in the first study, and a two-sample t-test to the second. The key difference is that in the first study you have two records per participant, and in the second study you only have one. As a result, the data in the first study is not independent. In this scenario, the structure of the study affected the statistical test.
Data type is still an important consideration. Numeric data often results in the calculation of means, which can lead to t-tests and ANOVAs. Categorical data generally results in the calculation of proportions, which can lead to z-tests and chi-square tests. If you are working with categorical data that has more than two groups, the type of test you perform will be affected by whether you have ordinal or nominal data.