Dependent variable- 7 scale double-pole (-3, ...,0,..+3)

Predictors (Total 24 numbers) comprising of:

Continuous variables - 6 number

Nominal - 3 number variables, 5 scale double pole (-2,...+2)

Ordinal - 6 number variables, 4 -5 scale unipolar (1, 2, ..., 5)

Ordinal - 9 number variables, 5 scale double pole (-2, ...,0, ...+2)

Need to incorporate all 24 variables in a single equation. Can I get some feedback on whether to approach with stepwise/ backward elimination or what?

(I am a fresher in statistics, as I tried to fit with backward elimination, I get plenty of unusual observations)

I have around 900 samples,

Would prefer to use Minitab, or maybe SPSS