Parametric and Nonparametric Tests Transcript
Today’s topic is about parametric and non-parametric tests, and let’s get started. Why are you interested in this topic? You want to calculate a hypothesis test but you don’t know exactly what the difference is between a parametric and a non-parametric test, and you’re wondering when to use which test. If you want to calculate a hypothesis test, you must first check the assumptions. One of the most common assumptions is that the data used must show a certain distribution, usually the normal distribution. Simplified, we could say that if your data is normally distributed, parametric tests are used. For example, the t-Test, the ANOVA, or a Pearson correlation. If your data is not normally distributed, you use a non-parametric test. For example, Mann-Whitney U test or Spearman correlation.
What about the other assumptions? Of course, you still have to check if there are further assumptions for the respective test, but in general, there are less assumptions for non-parametric tests than for parametric tests. So, why then do we need parametric tests? Parametric tests are generally more powerful than non-parametric tests. What does that mean? Let’s say you have formulated your null hypothesis, which is, for example, the salary of men and women does not differ. Whether the null hypothesis is rejected depends on the following things, on the difference in the salary and also on the sample size. In a parametric test, a smaller difference in the salary or a smaller sample is usually sufficient to reject the null hypothesis. If possible, always use parametric tests.
Now there’s still two open topics. First, I’ll show you what the most common parametric and non-parametric tests are, and then I will explain to you how you can easily calculate these tests online with DATAtab. Usually for the most common parametric tests, there is always a non-parametric counterpart. If you only have one sample, the parametric test is the simple t-test and the non-parametric test is the Wilcoxin test for one sample. If you have two dependent samples, on one side it is the paired t-Test and on the other side it is the Wilcoxin test. If we look at independent samples, it’s the unpaired t-Test and the Mann-Whitney U test. If you don’t know exactly what dependent and independent samples are, just watch my [inaudible 00:03:24] about it. You can find the link in the video description below. If you have more than two independent samples, you use the analysis of variants or the Kruskal-Wallace test.
Finally, if you have more than two dependent groups, you use the ANOVA with repeated measures or the Friedman Test. If you want to calculate the correlation between two variables, you can use either the Pearson correlation or the Spearman correlation. You can find the link to this overview in the video description.
If you want to analyze your data by using DATAtab, you just visit datatab.net and you click on the statistics calculator. Then you can copy your data into this table and click on this tab, which says t-Test, Chi square-test, ANOVA, and so on. Let’s say that you want a test if there is a difference between men and women in the salary. You choose salary as metric variable and as a nominal variable gender. Then automatically a t-Test for independent samples is calculated, which is a parametric test. If you want, you can also click here, then a non-parametric test will be calculated. Then you can see that the Mann-Whitney U-Test is calculated, which is the non-parametric counterpart to the t-Test for independent samples. I hope you like this video and thank you for watching.