Starting your analysis

Please try to go through these steps before you come to the methodology shop. Should you experience any difficulties while doing so, we are more than happy to help.

1. Identify your research question and/or specific hypotheses.
2. Identify the dependent variable(s) (what do you want to predict?)
3. Identify the independent variable(s) (what are your predictors?).
4. Determine the levels at which your variables are measured (are the variables categorical or continuous?) and whether your variables are measured once or repeatedly.
5. Pick an analysis according to the decision trees below.
   • Decision Tree part 1
   • Decision Tree part 2
   • Decision Tree part 3

With more than one dependent variable in the same analysis, multivariate models (e.g. MAN(C)OVA) are needed.
If your dependent variable is measured repeatedly, either a paired t-test (2 measurements) or a repeated measures ANOVA (more than 2 measurements) is appropriate.

Often, assumptions (e.g. normality) are tested on the residuals of your (already calculated) model, not the dependent variable.

On the internet, you can find many p-value based tests to check if the assumptions of your analysis are met (e.g. the Shapiro-Wilks test). We would not recommend using these as your main assumption checks, because of the drawbacks from NHST. A more informative way to check your assumptions is plotting (e.g. Q-Q plots and plots of residuals versus predicted values).

Step-by-step SPSS tutorials and basic analysis guides:

• StatisticsSolutions

For a recap of the basics:

• Online Statistics Education: A Multimedia Course of Study

For mediation and moderation analysis (we recommend PROCESS in SPSS):

• Mediation/Moderation analysis

For ANOVA:

• ANOVA in SPSS
• ANOVA Assumptions in R