my version 26 of SPSS for PLUM ordinal regression, produces an output of parameter estimates that indicates the location but uses categorical codes. My predictor is categorical with four levels, and the output looks like level =1, level = 2, level =3, level =4. however, I've seen examples of parameter estimates where the location is given for the categorical variable without the levels. I attached an example of the two versions of the output. why am I getting the first version with estimates for each of the category levels, and not a parameter estimate for just the one categorical variable? what commands or options should I be using to obtain a parameter estimate for just the one categorical variable?
I am confused by your question. I am not Familiar with SPSS nor PLUM. However, Predictors are Xs not Ys. Ordinal regression refers to the DV as ordinal not one of the IVs.. If i have misunderstood your question, please try again. You may find the attached reference useful. Best, D. Booth
Hello Peter. PLUM (which is short for Polytomous Logit Universal Model) is among those newer SPSS procedures that use key words BY and WITH to distinguish between factor variables and quantitative variables (called covariates in SPSS lingo) respectively. From the Command Syntax Reference manual (link below):
- Factor variables can be of any type (numeric versus string). Factor levels are sorted in ascending order. The lowest value defines the first category.
- Covariate variables must be numeric.
- Names of the factors follow the dependent variable separated by the keyword BY.
- Enter the covariates, if any, following the factors. Use the keyword WITH to separate covariates from factors (if any) and the dependent variable.
HTH.
https://www.ibm.com/support/knowledgecenter/en/SSLVMB_26.0.0/statistics_reference_project_ddita/spss/advanced/syn_plum_variable_list.html
@ Bruce Weaver, that was the problem. The variable if continuous was supposed to be in the With syntax and entered as covariates, and the categorical variable was supposed to be in the By syntax. Since my variables were categorical anyway, I did the analysis correctly and entered them into the By syntax as factors. I just couldn't figure out why some websites had the variables as covariates with only one estimate. I'm sure I'm not making sense here, but at the end of the day, the distinction between the With (continuous, covariates) and By (categorical) did the trick. Bruce Weaver
- Badges
- Science topic
- Similar topics
- Mathematical Sciences
- Statistics
- Statistical Software
- SPSS