How to conduct Repeated Measures MANCOVA in SPSS

I was at the analysis stage of my thesis and was facing problems with ‘Repeated Measures’. On consulting with my supervisor I got to know that over 89% of research candidates face the same problem and seek for guidance and help. I decided to overcome this fear of analysis, so, I researched through the method of conducting repeated measures MANCOVA in SPSS. With this article I want to help my fellow researchers and provide them with an insight of conducting repeated measures MANCOVA in SPSS, so they won’t face the same problem.

I think, it would be better if we have some basic understanding of ‘Repeated Measure’ before getting started.

So let me help you with the basics first.

‘Repeated Measures MANCOVA’ is used to test how a dependent variable varies over time, using multiple measurements of that variable, with each measurement separated by a given period of time. In addition to determine whether the dependent variable itself varies, a MANCOVA can also determine other variable are predictive of variability in the dependent variable overtime.

For your convenience and better understanding, I’m taking an example of a marketing campaign with three different strategies (variable: promo; valuable labels: Strategy A, Strategy B, Strategy C), the campaign is launched in market of three different sizes (variable: mktsize; value labels: small, Medium, large) and the measured sales for every quarter over for an year (variable name: sales 1, sales 2, sales 3, sales 4).

The data is shown as below:

Image: Data Value

The sales, shown in the data image are scaled in thousands (i.e. 68.42 is actually $68,420). Note that the data should be in person level format (wide).

To start the analysis using SPSS, click on Analyze > General linear model > Repeated Measures 

Image: SPSS MANCOVA MENU

Do the following in the Repeated Measures Window:

1. Mention your own name for the concept of time replacing the default within-subject factor name. I’m putting ‘Time’ in box no. 1.

   

2. Mention the number of dependent variables you have (number of times the dependent variables are measured) in the number of level box and click on ‘Add’ button.

   

3. Select a name for dependent variables (repeatedly measured variables) and mention it in the measure name box. Click on ‘Add’ button.

   I’m selecting the name ‘Sales’

   

4. Click on ‘Define’ button

   

   After hitting ‘Define’ button, the next repeated measures window appears which shows the four sales (variable) in the within-subject variable box. Continuous variable would go into the covariates box.

Image: Repeated Measures Window

Now, by clicking on the ‘Model button’, the Repeated Measures: Model dialogue window appears which is used to specify the model. Basically, it is used to select which variable have main effect on the ‘dependent variables’ and which variable might interact with each other to predict the dependent.

Full factorial is the default option which examines each variable’s main effect, also checks for the possible interaction among variables.

Here, I’m using full factorial, however, if you want to build your own model, you can choose custom option given in the Repeated Measures: Model dialogue window.

To exit the dialogue window, click on ‘Continue’ button.

Image: Repeated Measures: Model Dialogue Window

Next, in the Repeated Measures window, by clicking on the ‘Contrast’ button the ‘Repeated Measures: Contrast Dialogue Window’ appears. This window is used to change the type of contrast for each variable. Keep ‘Time’ as default i.e. polynomial, change ‘Promo’ to ‘Simple’ and ‘mktsize’ to ‘First’.

Image: Repeated Measures: Contrast Dialogue Window

Now, click on the ‘Plot’ button which will take you to the ‘Repeated Measure: Profile Plot’ dialogue window. You can change the type of plot or graph by using this window. I’m taking ‘Time’ on horizontal axis and ‘Promo’ factor variable as a separate line on vertical axis and hit ‘Continue’

Image: Repeated Measure: Profile Plot Dialogue Window

If you want the comparison between each of the factor level, you can click on ‘Post Hoc’ button in the ‘Repeated Measures’ window.

Now, back to the main ‘Repeated Measures’ dialogue window, you can click ‘OK’ to execute the analysis but I would like to recommend to click the ‘Paste’ button to paste the commands used in the analysis into the syntax window. This will allow you to easily recreate the analysis for later use.

Image: Repeated Measures MANCOVA Analysis

Now, you need to paste the command in the Syntax window. It would be recommended if you know the syntax as it improves your efficiency.

Image: Repeated Measure- Syntax Editor Window

To execute the analysis, highlight all the text and hit the green play button or you can use the menu as Run> Selection.

The output files are in the form of different tables and graphs. Below is the descriptive statistics table which shows Mean, Standard Deviation and Sample Size for each dependent variable. The other tables are Mauchly’s test of Sphericity, within-subject table and polynomials tables.

Image: Output Table1- Descriptive Statistics

Image: Profile Plot

The plot above shows the sales at each of the four data collection using three promotional strategies. You can interpret from the graph that there are difference between the sales numbers of three strategy groups mentioned initially (Strategy A, Strategy B, Strategy C).

And we are done!

I hope, my experience with Repeated Measures MANCOVA will help you sail your boat.

For any query and info, post your comments below.