Hi everyone,
Our team wants to see any difference between the clinical factors that affected the radiologic studies' cancellation rate between the preCOVID vs. COVID period.
In this way, we have two different groups (preCOVID and COVID), a dichotomous dependent variable (cancelled vs. no-cancelled), and various independent variables. I wanted to do a binary logistic regression to see which factors were associated with the dependent variable on each group.
However, I also want to compare these regression results between the two groups. I want to know if the cancellation rate factors differ during these periods. I don't know how I can do this. What test should I use?
I tried to fix 2 CARTs (classification and regressions trees) using the rpart package in R and offer a “visual” comparison of both groups. However, I also want to perform a numerical test to justify my discussion.
You have to possibilities: (1) Like Jos Feys said, you can interact the covid-dummy with the other independent variables: y = cons + covid + gender + age + covid*gender + covid*age and so on. (2) Or you could do y = cons + gender + age if covid = 0 and repeat this analysis for covid = 1. Afterwards, you then have to test the difference in effects between the two sub-samples.
You have to possibilities: (1) Like Jos Feys said, you can interact the covid-dummy with the other independent variables: y = cons + covid + gender + age + covid*gender + covid*age and so on. (2) Or you could do y = cons + gender + age if covid = 0 and repeat this analysis for covid = 1. Afterwards, you then have to test the difference in effects between the two sub-samples.
In terms of numerical comparisons between groups and factor level combinations you might also want to use the emmeans package to display and analyze mean values as predicted by your obtained models. This package might also provide a good way to display group differences in a plot.
- Badges
- Science topic
- Similar topics
- Biological Science
- Histology
- Fixatives