2
$\begingroup$

In my study involving three conditions (control, treatment 1, treatment 2), I am interested in measuring 2 DVs at a pre- and post-exposure. So effectively a 3x2 design (participants randomly allocated).

Research question: Does treatment reduce body dissatisfaction and negative mood?

I need to decide what type of analysis to use and I want to know if using 2 mixed-method ANOVAs for each DV (with bonferroni adj) is theoretically the same as performing a between subjects ANOVA, repeated measures ANOVA and then a between subjects ANOVA, for each DV (with bonferroni adj)?

Thank you!

Improve this question
$\endgroup$
2
  • $\begingroup$ Your first alternative I understand but it is not clear to me what your second one amounts to. Can you clarify it by editing your question? Also expanding your abbreviations would be good: what are P and RQ? $\endgroup$
    – mdewey
    Commented Apr 29, 2017 at 12:11
  • $\begingroup$ One of my students asked me about "Mixed Anova" this semester. She said it was on a video she had seen. $\endgroup$
    – Jamie Watts
    Commented Jan 2, 2020 at 5:59

1 Answer 1

Reset to default
0
$\begingroup$

First off, your terminology is incorrect. I am not aware of a "mixed methods" ANOVA. In research methods, mixed methods are studies with both qualitative and quantitative research design elements. In classic ANOVA terminology, you can have fixed and random effects. You can also have between and within factors (as you have). These are not mixed methods.

What you describe is a fairly basic and typical two-way factorial design with a within subjects factor (time) and a between subjects factor (condition). The standard way to analyze this would be with a two-way ANOVA, specifying one of the factors as within subjects (this affects the error term). The interaction effect between time and condition is the effect of interest, assuming you are interested in knowing if pre-post change differs across conditions. Splitting this up into separate models is considered poor practice. It will also not answer the main hypothesis unless the between subjects model is on the pre-post differences.

Improve this answer
$\endgroup$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.