I have a linear model with 4 independent variables.
To confirm that there are not omitted nonlinear variables left out, I ran the RESET test. It turned out that a multiplication by two variables (AB) is highly significant. Therefore I also included it in the model such that there are now 5 independent where one is a multiplication of two. (AB)
However, when I now run the entire regression of the model again, the A*B term is not significant anymore. Therefore I need to remove it again. Why is it like that? I know that the RESET test sometimes can be flawed.