Categorical IV and DV, continuous covariates.

[markhutton]

Hi all, I am doing a study on the specific criminogenic needs of prolific offenders and I have come up with a fair bit of data. Using 1000 participants and logistic regression I found that those who re-offend have high drug misuse, alcohol misuse, financial management problems and attitude problems towards crime.

I am now performing a secondary phase and am trying to assess whether a specific programme can address these 4 needs. I have A categorical IV (whther they have taken the programme or not) and a categorical DV (Reoffender yes/no). I have measured the four variables for each participant along a continuous scale but I can't seem to find a way to do this using SPSS.

Is it possible to use continuous variables as covariates in what is essentially categorical data?

Thanks,

Mark.

[Bruce Weaver]

markhutton wrote

Hi all, I am doing a study on the specific criminogenic needs of prolific offenders and I have come up with a fair bit of data. Using 1000 participants and logistic regression I found that those who re-offend have high drug misuse, alcohol misuse, financial management problems and attitude problems towards crime.

Your wording makes it sound as re-offender (Yes/No) is the explanatory variable, and the other things are outcomes.  But if you used logistic regression, then re-offender was the outcome (or dependent) variable in the model, so it would be more natural to word it something like:  the odds of re-offending are higher among those who have high drug misuse, alcohol misuse, financial management problems and attitude problems towards crime.

I am now performing a secondary phase and am trying to assess whether a specific programme can address these 4 needs. I have A categorical IV (whther they have taken the programme or not) and a categorical DV (Reoffender yes/no). I have measured the four variables for each participant along a continuous scale but I can't seem to find a way to do this using SPSS.

Is it possible to use continuous variables as covariates in what is essentially categorical data?

Thanks,

Mark.

I think you are suggesting another logistic regression model, with re-offender as the outcome variable, participation in the program (Yes/No) as the main explanatory variable, and you are asking if you can include continuous variables as additional explanatory variables.  If that is what you are asking, the answer is YES.  But do bear in mind that for logistic regression, you should have 15-20 events per model parameter in order to avoid over-fitting the model.  (An "event" is whichever category of the outcome variable occurs less frequently--e.g., a YES if there are fewer YES than NO outcomes.)

One other thing:  In SPSS dialogs, categorical explanatory variables are FACTORS, and continuous explanatory variables are COVARIATES.  

HTH, and good luck.

[markhutton]

Sorry, yes I did word it poorly. That's great, thank you very much for your help.

Mark.