I have a binary dependent variable model with five independent explanatory variables. If the p-value of the coefficients were insignificant, what shall I do? can I still interpret the coefficients without caring to the p-value ? because we just care for the sign of the coefficient not to the magnitude.
No because if the p-values are large you can't reject a hypothesis of no difference so it doesn't matter what the sign on the estimate is. If your sample size was say 50 (you didn't say what it was) you should be ok. Even 30 is probably enough. Are you sure the model represents what you want to test. I am attaching a very good set of notes that may help. Best, David Booth
P-values shows the level of confidence on how accurate a finding is. Confidence interval @1% means you have 99% assurance that you found was not by chance, @5% and @10 means 95% and 90% assurance that you found was not by chance. Anything above 10%@ means that result has no significant impact.
P-values show the level of confidence on how accurate a finding is. Confidence interval @1% means you have 99% assurance that what you found was not by chance, @5% and @10 means 95% and 90% assurance that what you found was not by chance. Anything above 10%@ means that the result has no significant impact.
- Badges
- Science topic
- Similar topics
- Mathematics
- Statistics
- Regression