I am creating 5 different regression models to measure a single phenomenon, tuition discounts. Tuition discounts are my independent variable in each regression. The Dependent variables are revenue, SAT scores, and percentage of underrepresented races. Each regression also has control variables such as net tuition price, endowment, total enrollment. My question is: can I use percentage of race as a control variable for my regression measuring the relationship between tuition discounts and SAT scores? Percentage of race is a good predictor for SAT scores so it makes sense to use as a control, but it is a dependent variable in one of my other regressions. It seems a little cannibalistic to use the variable in two places.
Hello Daniel,
It sounds to me as if tuition discount is the DV (dependent variable), especially if you're trying to evaluate how the other variables (e.g., revenue) combine to explain differences in discounts. It also sounds as if your data are at an institutional level, not at the individual person level (I only mention this because you need to make sure that all variables are quantified at the same level; otherwise you'll need to use a multilevel model).
As for whether population characteristics (e.g., proportion of clients who are minority) may be used as a control/IV in one model and as a DV in another model, the answer is, of course...so long as each model addresses one of your specific research questions. If you think about path models, one variable is often the "downstream" or DV in one portion of a model, and simultaneously, an "upstream" or IV in another portion of a model.
Good luck with your work.
Yes. BUT IF A NEW PREDICTED DV IS ENTERED IN A SECOND REGRESSION AS IV THE STANDARD ERROR OF THE OLD DV (NEW IV). MUST BE CONSIDERED. THIS MEANS USE AN ERRORS IN VARIABLES MODEL. GOOGLE THIS TERM PLEASE 🙏 IF YOU ARE DOING THIS WHICH IS COMMON AND GETS A WRONG ANSWER IF YOU DON'T DO IT. David Booth PS percentage of race seems to be a such a variable. If it is Don't skip the errors in variables model or YOU WILL BE WRONG. Your friendly statistician
Oops I almost forgot. Percentage is not a continuous variable so Ols regression is not appropriate. See Rosner Fundamentals of Biostatistics in the z-library for methods that are . Once again Good luck, your friendly statistician
Oops oops one last point. I meant in the second post if the percentage variable is used as a DV ols regression is not appropriate. Rosner has a good discussion., Your hopefully still friendly, statistician
- Badges
- Science topic
- Similar topics
- Mathematics
- Statistics
- Regression