The F stat for my model is very high (F= 3128851.522, p=.000), but only four of my predictors are significant out of a total of nine predictors. The no. of observations is 20. What should i do?
Asked
Active
Viewed 18 times
0
-
Maybe do forward selection or backward elimination of predictor variables to get rid of some inconsequential predictors. – BruceET Apr 20 '20 at 20:03
-
This may happen, see https://stats.stackexchange.com/questions/151403/significance-of-individual-coefficients-vs-significance-of-both. – Christoph Hanck Apr 21 '20 at 14:09
1 Answers
0
You need to either gather a lot more data or formulate different research questions.
Having 9 predictors with 20 observations is going to overfit your model dramatically. A general rule of thumb is that you need 10 observations for each independent variable.
Peter Flom
- 94,055
- 35
- 143
- 276