Graduate Level ANOVA with SPSS for Psychological Research
Below are links to multiple labs I presented for the incoming graduate student cohorts at UCLA. This course revolved around ANOVAs using SPSS. The professor who led the lectures preferred SPSS for ANOVA-based research since its very easy to specify them and SPSS automatically uses the correct statistics (like Type III sum of squares). Some of the simplest models you can also run in other software like R, however, quickly you might see that many of the types of analyses we want to do can’t be done in R (besides hard coding everything yourself, which we didn’t want to teach to psychology students).
Examples of things SPSS can do that R cannot:
- In between-subjects designs, R functions like
anova()give wrong results when you have unequal group sizes (which is very common). - In within-subject designs, functions such as
manova()rely on assumptions such as homskedasticity and sphericity, which are often violated in real data.
Software such as SPSS or SAS provide better solutions. It’s possible to do these ANOVAs in R, but it will require fine grain math and coding skills from the researchers. Supposedly, in the future the professor who taught this course will be forced to switch to R, so maybe I can ask them for their material so I can redo some of this.
The caveat of using SPSS is that I can’t run SPSS code natively in Quarto, which is how I am constructing this website. So all lessons will give you the syntax, but output will be done via screenshots as best as I can. But due to this limitation, the quality of the lesson material might ebb and flow depending on the topic.
The SPSS data sets can be found on this Google Link.
The lessons build on each other sequentially. So you will not understand as well later lessons if you do not work through previous lessons. Also, some of these lessons were written months apart, so I apoligze if the formatting/notation is inconsistent throughout.
Introduction to SPSS and One-Way ANOVA
This lesson introduces the logic of ANOVA through the one-way between-subjects design, covering how the \(F\) test compares group means by partitioning total variance into between-group and within-group components. It also provides a practical introduction to running and interpreting ANOVA output in SPSS.
Contrast Testing
Contrast testing allows you to ask focused questions about specific group comparisons rather than relying on the omnibus \(F\) test alone. This lesson covers how to specify and interpret planned contrasts in SPSS using the full/reduced model framework.
Type I Error Correction: Bonferroni, Tukey, and Scheffe Corrections
When multiple comparisons are made simultaneously, the probability of a false positive increases. This lesson covers the most common correction procedures and when to apply each one.
Trend Analysis: Linear and Nonlinear Trajectories
Trend analysis tests whether group means follow a systematic pattern across ordered levels of a factor, such as a linear increase or a quadratic curve. This lesson covers how to specify and interpret polynomial contrasts in SPSS.
Two-Way Between-Subjects ANOVA
This lesson extends the one-way design to two categorical factors, introducing the concept of an interaction effect. You will learn how to interpret main effects and interactions and how to follow up a significant interaction with simple effects tests.
Higher Order Between-Subjects ANOVA
Higher order designs include three or more between-subjects factors, producing main effects, two-way interactions, and a three-way interaction. This lesson covers how to interpret and follow up these more complex effect structures.
One-Way Within-Subjects ANOVA: Univariate and Multivariate Method
Within-subjects designs measure the same participants across multiple conditions, reducing error variance by removing individual differences from the error term. This lesson covers both the univariate and multivariate approaches to testing within-subjects effects in SPSS.
Two-Way/Higher Order Within-Subjects ANOVA
This lesson extends the within-subjects design to two or more repeated factors, covering how to interpret interactions among within-subjects effects and how to conduct follow-up analyses.
Split-Plot ANOVA
The split-plot design combines between-subjects and within-subjects factors in a single analysis, making it one of the most common designs in psychological research. This lesson covers the interpretation of main effects, interactions, and simple effects in mixed designs.
ANCOVA
ANCOVA extends the ANOVA framework by adding a continuous covariate to the model, reducing error variance and adjusting group means for pre-existing differences. This lesson covers the two-step procedure of testing homogeneity of regression slopes before evaluating the factor effect.