# Moderation analysis problems

#### ChristophR

##### New Member
Hi girls and guys

I'm pretty stuck rigth now with my moderation analysis.
First of all, this is for my bachelor thesis, psychology.

I've got:

-several continous dependent (lets call one of them HADS-A T1)
-a categorical independent variable (research condition)
-and a buch of moderating variables (lets call one of them CES-D T0)

First of all, I googled, but I didn't find anything as the things I found were either not relevant to my questions, or I didn't understand.

Now I know how to do the regression analysis (I use SPSS 21 btw).
Dependent variable (HADS-A T1) goes in the dependent variable field.
Independent vairable (condition) and moderating variable (CES-D T0) go in the independent variable field. There, in the second step, add the interaction term (CES-DT0Xconditon).

All continous variables were centered beforehand.

Now here comes my problem:
The categorical variable (condition) was scored as 1 and 0.
My instructor instructed me to recode it as 1 and 2. According to her a value of 0 would ... well, mess up the regression analysis, as anything X 0 = 0.

I did as she told me, and reported my beta and t values in a nice table.

I wanted to make graphs to depict the moderation effects. I tried ModGraph (http://www.victoria.ac.nz/psyc/paul-jose-files/modgraph/) to do so.

But it appears that ModGraph wants my categorical variable coded as 1 and 0, not 1 and 2 as I did. So I recoded again, did all analysis again. Put it into ModGraph.

The resulting graph does not make much sense to me. And I am not shure where I went wrong.

Also, my beta values are now (of course) different.

What I want to know:
- do I code my categorical as 1 and 0, or as 1 and 2? What differnce does it make?
- What do I report? I thought the betas would be easier to compare, as they are standardized.
- How do I make my graph? I want the kind where my continous dependent is on the Y-axis, my (continous) moderator is on the X-axis (but in catgeories -1std;mean;+1std), and two lines, one for each level of the categorical.

Any help would be appreciated greatly.

Will give honor and/or cookies as reward
Christopher

#### Karabiner

##### TS Contributor
The categorical variable (condition) was scored as 1 and 0.
My instructor instructed me to recode it as 1 and 2. According to her a value of 0 would ... well, mess up the regression analysis, as anything X 0 = 0.
This is silly. Categorical predictors are either coded as 0 and 1, or as 1 and -1.
It is about presence or absence.

EVERY textbook will tell her so.

With kind regards

K.

#### noetsi

##### No cake for spunky
That instructer's comments makes no sense. The norm for dummy variables is to code it 1 and 0