In video 7 we learn the chi-squared test for association in a contingency table. The premise is that we have given two treatments to groups in a clinical trial (just as an example). After administration of the treatments, we note whether there was worsening, no change, or improvement in the patients' condition. The variables are categorical and the test to see if there is a treatment effect is the chi-squared test.
One quick note: in the video, in the presentation of the observed results and the expected results, the letter "E" (expected value) should go where I have written "Pr". Blame it on senility.
Review of and comments on the video:
The chi-squared is the classical test here. One problem with this approach is that other factors in addition to treatment effect, such as gender, lifestyle, concomitant conditions, etc. cannot easily be tested with the chi-squared. In later videos we will learn more modern, model-based methods that are more expandable.
A couple of the commands used in this video:
observed <- matrix(c(15,18,35,5,4,42),ncol=3)
results <- chisq.test(observed)
Contact me at: email@example.com