# How to Graph Moderation Effects on SPSS

I am conducting an independent research but am completely stuck on how to visualize my data.

So I am looking for a moderation effect of peer sociometrics (ex. negative nominations, social preference) between response inhibition scores and later intimate partner violence.

For example, I would like to show that people with lower response inhibition, along with negative peer status, predict higher rates of intimate partner violence later in life.

All of the variables (moderators, independent, and dependent) are continuous. In this case, how do I graph it?

I heard that I could create a 3D model, but it would be very difficult to comprehend for the readers and was advised to use the xy graph instead.

However, with 3 different types of variables (I have more in total, as I deal with 5 different types of peer sociometrics), I wasn't sure how all of the variables can be coherently put in one graph.

Also, when writing out the results in the manuscript, which information do i have to include (from the analyses)?

For example, I would put (β=0.000, p<0.000, SE=0.000) for a regression analysis - but what for the moderation analyses?

• What are the variables? What is moderating what? Can you post the regression equation? May 29, 2016 at 12:01
• (1) independent variable: response inhibition / moderators: social impact, social preference / dependent variable: intimate partner violence in later life (2) I am trying to graph how with moderators significantly impact the association between the response inhibition (at childhood) and later intimate partner violence. (3) I used PROCESS in SPSS to find the significance! I'm not sure how I'd be able to show you the equation..Sorry :( Thanks for your help!!
– Kim
May 30, 2016 at 4:56
• There is an example here; ruf.rice.edu/~lane/papers/designing_better_graphs.pdf I don't like the graphs that include only lines and no points. Mar 19, 2017 at 0:15

Regarding how to plot the interaction/moderation: You can plot the simple slopes on the y-axis and the moderator on the x-axis. By doing that, the reader can easily see how increasing the moderator will change the linear relationship between predictor and your dependent variable.

You can add a confidence band to your graph, so that the reader can see which values of your moderator will result in significant simple slopes. That procedure is known as "regions of significance".

Alternatively, you can calculate simple regression equations (regression lines that would be truth if alle observational unit would have one and the same value on the moderator).

You can finde a good explanation here: http://www.quantpsy.org/interact/interactions.htm

You can plot the simple regression lines by using SPSS or Excel. However, plotting regions of significance in SPSS is difficult. The following link leads to a online calculator for computing simple regression lines and the region of significance: http://www.quantpsy.org/interact/mlr2.htm

The code is generated by the website and send to Rweb. Thus, you do not need to know R-Code on your own. However, it may be useful to consult a R-Programmer if you need to modify the for publication.

Moreover, the calculator needs the covariation matrix of your regression-estimates. However, SPSS will not give you all of them until you trick it into doing so. the following link shows you how to get the covariance matrix: http://www.quantpsy.org/interact/acov.htm