Linear Regression Calculation differences

[dregelyi]

Dear All,
I have a data set consist of 2 factors and 1 output variable. I would like to perform linear regression with the base Jamovi modul.
The calculated ANOVA (Omnibus ANOVA test) is good (pic1.jpg).
The next step is to take into account the interaction to the model. In this case the calculated ANOVA table is and the modell coefficients (pic2.jpg).
When I have repeated it with the Linear Models modul, there the calculation is different(pic3.jpg).
If the variable is take into account as factor, then the ANOVA is the same. If the variable is take into account as covariates the results are different.
Could you explain the received difference?

Best regards,
Agota

[mcfanda@gmail.com]

General Linear Model by default centers the covariates to their mean, whereas Linear Regression does not. So, when your variables are entered as covariates and there is an interaction in the model, the first order effects are different. If you center your variables or in GLM set the scale to "none", you get the same results.

If you define your variables as factors you get the same results because both modules code the factors in the same way (centered dummy coding)

[dregelyi]

Thanks, it is clear now.
BR,Agota