Multiple dependent variables in factorial design I have a 2x2x2 factorial design with two dependent variables (lets say height and weight).  I can examine the effect of the three factors for each dependent variable separately. But I also want to check if height has anything to do with weight?  I.e., is the tallest subject the heaviest?


*

*Any idea on how to analyse this in spss??

*Also what would the assumptions be?

 A: If you are up for a challenge, look into MANOVA (multivariate analysis of variance) and its assumptions.  It would fit the case where you have multiple categorical predictors and multiple continuous outcomes.  In essence, MANOVA would help you describe the web of relationships connecting your predictors to your outcomes.
If you want to keep things simpler, you can analyze the height-weight relationship separately using a scatterplot and (if you find a linear relationship) correlation.  For your main research questions you could conduct separate ANOVAs (analysis of variance) for height and for weight.  Again, you'll want to look into the assumptions underlying best-practice use of ANOVA.
Depending on your version of SPSS, for modeling you'll want to go into Analyze...General Linear Model...[Univariate or Multivariate].  The Help files there should prove somewhat useful, and if you run into a link saying "Show me," it should take you to a tutorial; most of these are pretty good.
For the scatterplot, type graph/scatter height with weight.
For correlation, it's simply corr height weight.
A: Effect of factors on 2 Dependent Variables (DVs)
Some options include:


*

*MANOVA

*Two separate ANOVAs

*Create a composite out of your two DVs and run an ANOVA with the composite as a DV


Effect of DV1 (e.g., height) on DV2 (e.g., weight)
Some options include


*

*Correlation or simple regression of DV1 with DV2

*ANCOVA

*

*You could include your three factors (in SPSS, see GLM, fixed factors) and DV1 (in SPSS, see GLM, covariates) as predictors and DV2 as the dependent variable.

*The ANCOVA option is arguably the most directly related to your research question: Check out this discussion of assumptions by David Garson.


