Determining which test done to same subject at different intervals correlates with outcome
Jan 20, 2020 6:57 AM(196 views)
Please help describe how to set up this situation in jmp.
There are two tests completed on the same individual. It is the same test completed several weeks apart. (so call it Test A and Test B). Both of these variables are continuous variables.
The outcome is then confirmed and also is a continuous variable.
I want to know which test correlates stronger with the outcome, i.e. is the amount found in Test A more likely to predict the outcome or is the amount found in Test B more likely to predict the outcome.
Set up your data table with a separate data column for each test and the outcome. Enter the results for an individual on separate rows. Then select Analyze > Fit Model. Enter the outcome in the Y role. Select the data columns for the tests and click Add. Click Run. This is a multiple regression model. The effect tests should help you determine which test is correlated with the outcome.
There are potential pitfalls with regression, when your data (situation) does not follow the assumptions. Are the two tests correlated? That is a common issue with regression analysis.