# How to interpret output of a moderation with 2 moderators (1 Dichotomous, 1 continuous) and a dichotomous predictor?

I have conducted a moderation analysis with two independent moderators. (Using Hayes Process model 2). I have attached the output for your reference.

I am looking for the impact that financial wellbeing and gender has the relationship between marital status and loneliness. My variables are:

• Y: Loneliness (continuous variable - low values indicating low loneliness)
• X: Marital status (single or married- dummy coded to have single as the reference category)
• W (mod1): Financial wellbeing
(continuous- low values indicating low financial wellbeing)
• Z (mod2): Gender (male or female- dummy coded to have male as the reference category)

All the p-values show significance

I have two questions:

1. Is this selection of variables appropriate for the moderation analysis I am undertaking? Is the combination of a dichotomous predictor and moderator suitable for this?
2. Can someone please help me interpret the output? When looking at the conditional effect, I am seeing all the values under 'effect' are negative- some closer to 0 than others but I am not sure how to interpret this in relation to the dichotomous predictor (being single coded as 0).

Any guidance would be greatly appreciated!

• Please edit the question to provide the program output as text. If you paste text into the editor, select it, and use the code {} tool, it's easier to read than in a picture, values can be copied for further analysis, and those who use speech-to-text can "see" the values.
– EdM
May 16 at 13:42
• Thank you for your suggestion! Each time I copy the text into the question to edit - it forces it into an image format without the option to use the code tool. Even when first selecting the code tool to then copy and paste the output in -It still wants to convert it into an image. I'm unsure why it's doing that and how to prevent it, I'm afraid! May 16 at 14:08

Y ~ X + W + Z + X:W + X:Z

The "moderation" is via the interaction terms X:W and X:Z, with X (marital status here) being of primary interest but with individual terms also needed for W and Z. Interactions are possible in principle between categorical predictors, continuous predictors, or any combination, so you have no problem there. The interaction terms are simply products of the included predictor-variable values.
I'm not sure, however, how this PROCESS Procedure works together with standard SPSS coding of categorical predictors like X and Z here. The SPSS default is to express coefficients as differences from the highest level of a categorical predictor, unlike R which uses differences from the lowest level. You will have to check the manual(s) or play with some example data to determine whether, for example, the coefficient for Z (Male) is for (Male - Female) or (Female - Male). (Highly software-specific questions are off-topic on this site.)