0

I have designed a study and am not sure which statistical analysis can best answer my research question.

Just to simplify the problem, I have two IVs. IV1 is continuous, and IV2 is a categorical one with three levels. The DV is a continuous numeric value.

The thing is the IV2 restricts the range of the DV, making the DV confounded with IV2.

For example, in one of the three levels of IV2, the possible DV can range from -2 to +2. However, in the second level of IV2, the possible DV can range from -1 to +3. In the last level of IV2, the DV can range from 0 to +4.

In this case, which analysis should I use to adjust this inherent difference in the range of DV that varies depending on the IV2?

Any suggestions will be appreciated!

May
  • 101
  • 2
    It depends on your model and what you're trying to achieve. After all, you can eliminate this problem simply by subtracting 0, 1, or 2 from the DV depending on the level of IV2: but what effect would that have on your model or your interpretation of the results? – whuber Oct 09 '19 at 21:47
  • @whuber Thanks for your comment! I was considering Linear Mixed Models. If I subtract 0,1,2 from the DV, then it would change the interpretation, because DV = 0 represents equality (which I'm interested in measuring). DV= -2, -1, +1, +2 etc all represent the deviation from equality. – May Oct 09 '19 at 22:42

0 Answers0