Multiple or simple linear regression?

#1
Hi all

I'm fairly new to statistics and have a project where I investigate if the difference between two variables can predict the value of a third variable. And I am unsure if I should use simple or multiple regression.

The participants in my sample has answered three surveys:
A) Emotions they have had during the week
B) Emotions they ideally want to have
C) Their life satisfaction

My hypothesis is that the difference between the mean score of emotions people report they have (A), and the mean score of the emotions the same people report they ideally want to have (B) can predict the same people's score on life satisfaction (C).

Can I just calculate the mean difference of A and B and use that in a linear regression to see if it predicts the value of C?

Or should I use a multiple regression instead where A and B are predictors of C?

I ask because I don't know if it is a no-go for some reason to just use the raw difference between the two variables in my case.

I hope you can help. Thank you so much!
 

Karabiner

TS Contributor
#2
Do you have reference literature on the effect of the actual/desired emotions on life satisfaction?
It would be interesting to know how they handled it there.

I supppose that a regression model including actual emotions plus the discrapancy between actual
and desired emotions could be used. If you only use the discrepancy, one could suspect that a
correlation between discrepancy and satisfaction is just an effect of "low" emotion.

With kind regards

Karabiner
 
#3
Thank you for the reply.

This article shows the effect of actual/desired emotions on depressive symptoms: https://doi.org/10.1037/0022-3514.90.2.288

In this, the researchers investigate the difference between high aroused positive emotions (HAP) and low aroused positive emotions (LAP) and its effect on depression among other things.

They describe that they perform a hierarchical regression analysis to assess the incremental change in R-square for depression, when the discrepancy between actual and ideal HAP is entered at the first step, and the discrepancy between actual and ideal LAP was entered at the second step. But they don't say any more than that.
 

Karabiner

TS Contributor
#4
So, discrepancy between actual and ideal/desired emotion was used as a predictor only once in
depression and/or life satisfaction research. This seems a bit surprising. But at least you know now
that using the difference value was considered as acceptable for publication in an APA journal,
so it might also be o.k. for you.

With kind regards

Karabiner
 
#5
Thank you again. It is just one of the references I have. I couldn't find anything about this topic in my statistics book.
And with your help I now know that I can use the difference value to predict life satisfaction in a regression model.