Say one has a large data set of cross-sectional data over a number of years (independently pooled cross section across time). If one regresses a subset of that large data set – a cross-sectional model for the data for one year with no year dummy variables included - can the goodness of fit value from this model be compared directly to the goodness of fit value from the pooled regression model using multiple years and dummy variables for the years? Why or why not?