Different results when conducting regression and ANCOVA in SPSS I'm conducting a regression analysis with two binary categorical variables and one continuous variable as independent variables. From my understanding, the results should be identical as the results from an ANCOVA analysis in SPSS (general linear model - > univariat -> treat the binary variables as fixed factors and the continuous variable as covariate). I have to add, that I don't model any interactions (just main effects). Now, the results are 100% identical but the constant is a little bit different. Why is that? From my understanding, the constant should also be the same. 
Thank you!
 A: The difference is probably due to how the reference categories for your categorical variables are defined.
Okay: to elaborate.  Suppose you regress salary on a minority dummy (0/1) variable and jobtime using the employee data.sav file shipped with Statistics.  The regression produces
(constant) 23248.908
Minority  -7499.316
jobtime   158.009
In the Univariate output for the same variables, you get
intercept 15749.591
minority=0 7499.316
minority=1 0 (omitted because it is redundant.
jobtime   158.009  
Notice that Univariate has omitted minority = 1 while the regresssion spec omitted minority = 0.  The opposite minority coefficient exactly offsets the difference in the intercept.  Tbat is, if you calculate the predicted value for minority = 1, in the regression case it is the sum of the intercept and the minority=1 coefficient, while in the Univariate parameterization it is just the intercept.  
So the two sets of results are exactly equivalent.  They are just parameterized differently.
