Dear Researchers.
The Visual Analogue Scale (VAS), and other similar scales, has a floor and a ceiling score; e.g.: 0 to 100.
Usually, these scales are analysed with linear regression models, but the predicted scores could appear outside the conceptual range of the scale.
In this manner, What is the appropriate regression model for the analysis of the Visual Analogue Scale and other scales with floor and ceiling scores?
Thank you.
Andrés.
I would begin by looking at limited dependent variable models. See the link below. Best wishes, David
http://www.bing.com/search?pc=COSP&ptag=D071616-A0A14E5293D1943C986F&form=CONBDF&conlogo=CT3330954&q=limited+dependent+variable
In general, this should not trouble you too much. However, have you checked the assumptions of your regression model? The described scenario sounds as if you were very likely violating homoscedasticity.
You need to provide more information, for example, about the distribution of variables and the research purpose. For example, if they theory that you are evaluating stipulates a linear relationship, and your goal is a test of falsification, then go with linear (your p values might be off, but that is not the issue). If the theory says the relationship is quadratic, that is what you would examine.
- Badges
- Science topic
- Similar topics
- Psychology
- Psychometrics