Over ANOVA

An example research question with statistical tests.

A researcher decides to study how a new drug (drug A) affects blood pressure in patients. They choose a control group size of 200 and a test group size of 300. In this situation, would it be best to recommend a z-test, a t-test, or an ANOVA?

For this, we need to define the null and alternative hypotheses.

Null Hypothesis: Drug A does not affect blood pressure in patients.

Alternative Hypothesis: Drug A has a statistically significant effect on the blood pressure of patients.

First, consider what each test is designed to do. This will let us make an informed decision and shed light on the right course of action for the researcher. Overall, these tests will allow the researcher to determine if Drug A has an effect on the blood pressure of patients. These test statistics are best described as a ratio of variance or noise. Andy Field does a good job defining this concept as the “variance explained by the model divided by the variance that the model cannot explain” (Field, 2013). Before we dive into the tests, it will be beneficial to consider what our variables are.

We have two samples: a control group and an experimental group. We know the sizes of these groups as well. The control group consists of 200 individuals, and the experimental group…