The following are instructions for using the interActive data visualization application. We designed this application to help researchers analyze, create graphics for, and ultimately better understand interaction effects. The application is located at the following link:
https://connorjmccabe.shinyapps.io/interactive/
Please be sure to read the accompanying manuscript before using this application. You can find an OnlineFirst version of this manuscript on the webpage for the new APS journal, Advances in Methods and Practices in Psychological Science:
http://journals.sagepub.com/doi/full/10.1177/2515245917746792
The following steps for using interActive are based on a simulated dataset of 150 observations. These data are provided in the GitHub repository, or accessible directly for download at the link below:
Before proceeding, be sure to download this data file to your local computer for upload to the interActive application.
Step 1: select a data file by clicking the “browse” button on the upper left-hand side of the screen.
Acceptable file types are .csv, .xls, .xlsx, and .sav (SPSS data files).
The model being estimated is Y = X + Z + XZ. This model can be specified in interActive using the dropdown menus provided (e.g., see below).
Note that X and Z are centered and Y is untransformed (raw) by default, though multiple scaling options are available. No covariates or quadratic effects will be specified in the current example.
Step 3: Click the “Run Analysis” button (depicted above) to produce a summary table of coefficients and interaction graphics.
This will generate a summary table of coefficients, conduct regions of significance analyses, and will create the interaction graphics.
Each multiple corresponds with a particular level of the moderator. By default, these range from -2 to 2 standard deviations because this will generally be representative of the moderator range given the moderator is approximately normal. However, we encourage you to modify these values to explore this functionality. Once small multiple values are decided upon, proceed to step 5 to customize the plot.
Step 5: Click on the “Customize Plot” tab to change axes and plot titles, select the “greyscale” option, and download the finalized plot.
Users may use these customization options to ready this plot for use in a manuscript. Clicking the “Download Plot” button will output a .png file.
Click on the “Marginal Effects Plot” tab to view a depiction of the regions of significance analyses. Note that this plot displays standardized versions of X and Z predictor variables. Customization options for this plot are provided on this tab as well, and users can download this plot using the “Download Marginal Effects Plot” button below the graphic.
Users may view the uploaded data by clicking on the “Raw Data” tab to help identify problems in their data (e.g., missing values, outliers, etc.). Users can also view the plot estimates directly by clicking on the “Plot Estimates” tab to view the estimates used to create the small multiples graphic.
Users can download an .rds file using the “Download Plot Data” button located on the upper-righthand side of the screen under the "Plot" tab. Advanced users can read this file into R or Rstudio to customize their plots or examine their models in greater detail. Example code for reading an .rds file into R or Rstudio is provided below:
readRDS("~/Downloads/plotdata.rds")
Provided in this data file are positively-skewed predictor variables simulated from exponential distributions. We encourage the reviewers to conduct the above steps using these variables as well, by specifying the following model in interActive:
yexpXZ = Xexp + Zexp + Xexp*Zexp
The screenshot below depicts the dropdown menu options for producing this graphic:
