How Could the Estimates Obtained through OLS and GLS Vary in Their 95% Confidence Interval Values While They Have the Same P-Value?

Question

I am working on an interrupted time series analysis. The data actually meets most of the assumptions of the OLS (the outcome variable is normally distributed, the Beusch-Pag