# Formulating the general purpose of statistical tests in plain language

I am a bachelor student trying to understand the purpose of statistical tests.

Explanatory data analysis is very straightforward to understand: we compare the means of a variable between two groups. Let's say we have a body mass index (BMI) of two groups, men and women. The difference in BMI is 4 points. Now we have these statistical tests that we can use. Is my understanding correct that these tests do the following:

1. allow me to say if the 4-point difference is statistically significant or not (hypothesis testing, p-values);
2. allow quantifying uncertainty for the 4-point difference (confidence intervals). In other words, how certain I am that the mean BMI of these groups is actually (statistically) different.
• First, your characterization of EDA is so much more limited than what EDA actually does that you ought to consider learning more about EDA as soon as you can. Second, this flowchart ("tree") is only for those who haven't any idea of what to do with data and, IMHO, it's not going to be of any help in learning good techniques of data analysis. Because it presents such a false and limited perspective on statistics, you are quite correct to react to it by questioning the objectives of statistical testing. I trust you will be able to find better resources for learning and doing statistics.
– whuber
Jan 8 at 17:31