Hi,
I think I have a problem with a variable in my dataset.
I have 2 dependent variables and I am regressing them on a categorical predictor.
I graphed the residual vs fit plots for both of them.
For one I get a plot with 2 straight lines and for the other one, I get a cloud of dots.
What is the problem do you think?
See pictures attached.
Thank you,
Claire
What is the problem do you think?
(the answer to this question will help us to help you more specifically)
I'm not sure... The two outcomes are composite scores of similar items (measured on Likert scales), so although the distributions of the errors can be different, I'm surprised one has 2 straight lines and the other one has a cloud of dots.
I was wondering if when a the predictor is categorical I should have two straight lines. In which case, I have a problem with my first outcome for which I have a cloud of dots.
Is that clearer?
I see. There is one categorical predictor that can take two values. There is nothing else in the model (except the response, for sure). This means the model can predict two different values. Therfore, the x-coordinates of the plots should have two distinct values (leading to the alignment of the points in two parallel, vertical lines). If the model includes only the categorical predictor, the picture must look like this (like in fit2). The "cloud" of points (like in fit1) indicates that model contains a predictor can take many different (numerical) values - usually a continuous/metric variable.
- Badges
- Science topic
- Similar topics
- Mathematical Sciences
- Statistics
- Statistical Software
- SPSS