From the course: Designing Big Data Healthcare Studies, Part Two
Unlock the full course today
Join today to access over 22,400 courses taught by industry experts or purchase this course individually.
Selective stepwise modeling: Breaking the working model - R Tutorial
From the course: Designing Big Data Healthcare Studies, Part Two
Selective stepwise modeling: Breaking the working model
- [Teacher] Welcome to this section where we do round two of SS modeling and have a little fun breaking our working model. But for now, let's get down to business. In round two, you start with the working model you developed in round one where you tried all the covariates and only kept the ones that were significant, or almost significant in the model, along with the exposure variable. So basically, your working model has a bunch of covariates in it and most, if not all of them, are statistically significant. Remember table one, your descriptive table? You listed all of your potential covariates in that. So I use table one as a guide as to what order to go in, when putting covariates in the model, both in round one and round two. So go pick out the first discarded variable that's not in your working model. Go ahead and add that one to the model and run it, then look at it and update your metadata. And under the common column, state your verdict. Keep it, or toss it. Should it stay, or…
Practice while you learn with exercise files
Download the files the instructor uses to teach the course. Follow along and learn by watching, listening and practicing.
Contents
-
-
-
-
Descriptive vs. regression analysis3m 58s
-
Considerations for categorical outcomes4m 59s
-
Structure of descriptive table: Table 13m 45s
-
Descriptive table example2m 13s
-
Stepwise modeling to answer a hypothesis5m 17s
-
Establishing the working model: Part 13m 30s
-
Establishing the working model: Part 22m 20s
-
Documenting model metadata4m 15s
-
Selective stepwise modeling: Breaking the working model4m 27s
-
Considering model fit4m 20s
-
Selecting and interpreting the final model4m 13s
-
-
-