Let's say that I am interested in how exam test score varies with a person's height. To do this, I take a sample from a school, and then I split this sample into two categories, (1) shorter than 165cm and (2) taller than or equal to 165cm. I then perform a one-way ANOVA test to compare the mean exam test score in the two subgroups.

With this setup, are the observations in the two groups independent?

  • 1
    $\begingroup$ You should not binarize the continuous heightvariable. Keep it as is, and use a regression model. $\endgroup$ – kjetil b halvorsen Dec 6 '19 at 4:08

Yes, the groups are independent, but, as @kjetil pointed out, this is not the way to answer the question you have. Dichotomizing a continuous IV is nearly always a bad idea.

I wrote a blog post that shows, graphically, what happens when you categorize an IV. Also see Frank Harrell's book Regression Modeling Strategies which gives a list of the many things that go wrong when you do this.


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.