Hi. I need some help. Im currently analyzing my data. I couldn't run regression test in SPSS since the data for my DV is in ratio (Percentage) while my IV's are in likert scale. Unable to detect the variables in the test. Even the correlation test between the IVs and DV showed negative. How to solve the problem?
Do your Likest-Scored items group together into scales? If so, you should use Cronbach's alpha as a first step in assessing those scales.
If instead you have single items (rather than scales), then you will need to use non-parametric statistics.
Hello Subramanium,
I agree with David L Morgan , that if the Likert-type response scale items are related, it makes more sense to combine them into a summated scale. If these have not previously been so grouped, you might consider exploratory factor analysis as a way to determine how/which items to combine. Alternatively, you could try polytomous IRT modeling to create one or more scale scores for cases.
I'm not sure why you think that having a DV which is a percent (of something) restricts you from using OLS regression. While percents (and proportions) are bounded, and therefore not genuinely continuous variables (e.g., to infinity in either tail), they often behave reasonably like continuous variables. The arc sin transformation for proportions is a common method for "stretching" the scale.
The problem with percent variables that is a real concern is when you have a set, and they collectively sum to 100 (for percent) or 1 (for proportion). In such instances, there are linear dependencies (and negative intercorrelations) induced by virtue of the fact that the values sum to a constant. That doesn't appear to be the case for your situation, however.
Good luck with your work.
Have you checked the assumptions for regression?
You have 2 continuous variables - so a regression should not be an issue.
The issue might lie in the fact that your data does not meet the assumptions, and the model does not converge. Hmmm.
- Badges
- Science topic
- Similar topics
- Mathematical Sciences
- Statistics
- Statistical Software
- SPSS