What would be the appropriate test for a study that has:

2 categorical independent variables with 2 levels each, 1 moderator variable (and would it matter if my moderator is categorical or continuous?), and 1 continuous dependent variable

I am still undecided on whether I will propose my variables as being correlated or as predictor/criterion variables.

If the results of my moderator (which will most likely be measured on a scale of 1 to 7) are scored as low (1-2.9), medium (3-5), and high (5.1-7), would I be correct in assuming that, as there are more than 2 'levels' of the moderator, I will have to encode them into dummy variables if I'm using SPSS?

Would my choice of carrying out multiple regression be appropriate for this study?


Yes, you can carry out multiple regression. Alternatively, you can perform an ANOVA. You will get the same outcome either way, so choose one that better fits your context.

As to your moderator variable, yes, if you code it into low-medium-high then it becomes a categorical (dummy) variable regardless of the number of levels.

  • 1
    $\begingroup$ Do not recode a 1-7 ordinal variable into intervals. This loses precision and power and is completely arbitrary. Spend more time understanding the data, and don't launch into statistics without studying a good intro book. $\endgroup$ – Frank Harrell Feb 6 at 12:40
  • $\begingroup$ Hello and thank you for the response. I should clarify that the Multidimensional Scale of Perceived Social Support (MSPSS) scoring guide instructs as i have mentioned above with regards to assigning the low, medium, and high values based on the score. May I ask: what would you advise is the most straightforward way to work with my 3 level categorical moderator on SPSS? Would choosing model 1 of the PROCESS macro be sufficient? $\endgroup$ – Aamir Kapasi Feb 6 at 17:27

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.