I apologize if there's something I'm missing here or if this has been asked already, but I couldn't find this issue in past discussions and it's been vexing me for the past while. Basically, I put together a linear regression model for a dataset, and got some good, significant results (shown below on the right). A couple hours later, when working on another model, I accidentally put the same response variable in as before instead of the new one I was trying to measure. However, something didn't look right about the resulting model. When I pared it down to the same effects as before to double check, I got a wildly different result (shown below on the left).
Any ideas about what could be causing this? At first I thought it was the result of me hiding/excluding some key data points in the interim. However, toggling inclusion of various data points seems to cause corresponding change in the p-values of the right (original) model, but not the left, which remains constant for some reason. I've double checked and all the data types are the same for both- it is the same dataset, as far as I can tell. The blank R-squared field on the right, as well as the fact that the degrees of freedom differ by one, indicate that something's not right here, but I've got no idea as to what. Any insight would be helpful!
Accepted Solutions
