Logistic Regression
The concept of confounding and how you can use multivariate tests to limit the effect of confounders on the outcomes of tests of the association of dependent and independent variables has been discussed.
Focus on the definitions of confounding and effect modification as presented in the text (Chapter 9):
Essentials of Biostatistics in Public Health, Second Edition
Lisa M. Sullivan
Jones & Bartlett Learning, 2012
978-0-7637-9531-3
Also, consider the examples given in the text and the PowerPoint (see Additional Materials) on multiple linear and logistic regression of potential confounders. Then consider the following:
A researcher plans to use a sample of 500 men and women attending a well-known online university to study the relationship between the number of fast-food hamburgers consumed in a month and BMI (body mass index). The null hypothesis is that there is no relationship between fast-food hamburger consumption and BMI.
1. What statistical tests could the researcher use to test the null hypothesis? How might those change if the dependent variable was yes/no obesity instead of BMI?
2. What confounders would you include in your analysis?
3. Explain to your fellow students why it is important to include these potential confounders in any tests of this relationship. Support your opinions with the literature and cite your sources.