Hello, I have a question concerning predictivity/ or R squared of two sets of variables.
what id like to do, is report r2,r2adjusted...
for two models
but i cant do hierarchical regression, because in the first set of variables I have two categorical variables (for ex. gender).
Now i fear I have to use GLm with dummies... I don't know how to do this. Wouldnt i be testing for difference between variances when using GLM?
Im working with spss, could someone give me a hint or a link to an explanation?
thank you very much
Joa
Hello Joa,
To report R-squared and adjusted R-squared for two models, one with categorical variables and another without, you can use the Generalized Linear Model (GLM) in SPSS. The GLM allows you to handle categorical variables through the use of dummy variables.
First, create dummy variables for your categorical variables. In SPSS, you can use the "Transform" -> "Recode into Different Variables" option to create dummy variables for each category of your categorical variables. For example, if you have a gender variable with two categories (male and female), you would create two dummy variables (e.g., male_dummy and female_dummy) where 1 represents the presence of that category and 0 represents the absence.
Once you have created the dummy variables, you can include them in your GLM model along with your other predictor variables. In SPSS, go to "Analyse" -> "Generalized Linear Models" -> "Generalized Linear Models" to open the GLM dialog box. Here, you can select your dependent variable and the predictor variables, including the dummy variables.
Make sure to specify the appropriate link function and distribution based on the nature of your dependent variable. For example, if your dependent variable is continuous and normally distributed, you can select the "Identity" link function and "Normal" distribution.
After running the GLM, you can find the R-squared and adjusted R-squared values in the "Parameter Estimates" table. The R-squared value represents the proportion of variance in the dependent variable explained by the predictors, while the adjusted R-squared adjusts for the number of predictors in the model.
It's important to note that GLM is not specifically testing for differences between variances. It is a general method for analyzing the relationship between predictors and a dependent variable, whether they are continuous or categorical.
For a more detailed explanation and step-by-step instructions, you can refer to the SPSS documentation or search for tutorials on GLM with dummy variables in SPSS.
I hope this helps! Let me know if you have any further questions.