I ran multiple linear regression by using six IVs to predict household water insecurity in rural Ghana. My output came out with R-squared and Adjusted R-Squared of .981 and .962, respectively. Can these be too high in social sciences research? Does it raise any question(s)? What should I do in this situation?
It's a very high value, indeed: the model would explain almost all the variability of the response data around its mean. Maybe to predict household water insecurity is valid but, in my experience, such a high value is not common in social sciences, where many causes are usually involved. Check the calculations. If that's the final result, explain it very well, both the analytical process and the model.
I agree with Pablo Delatorre that such a high value would be very unusual in the social sciences. So, I would start by examining your correlation matrix to see if one or more of your independent variables is highly correlated with the dependent variables. You could also look at the descriptive statistics for your variables to make sure that the mean values make sense.
- Badges
- Science topic
- Similar topics
- Social Science