Re: repeated measures question

You just have to watch out, endlooser, with what Statistica calls "manova". I'm not used to working with this package, but I tried it once, thinking I was doing a manova (through the "anova" interface), but ending up with a common repeated measures anova. That was in a very old version. I just tried it again, this time in Statistica 8, and it correctly tells me what it is doing (although selecting multiple dependent variables might still trick you into thinking that you are performing manova). In the ANOVA module, I can only define one within-subjects factor, in which case it does manova (Wills lambda). To analyze a multifactorial repeated-measures design, it refers me to GLM:
"GLM. Note that the General ANOVA/MANOVA module only allows one within-subject (repeated measures) factor. If your design has multiple within-subject (repeated measures) factors, you need to use the General Linear Models module." (That is in Statistics -> Advanced Linear/Nonlinear models -> General Linear Models)
Statistica doesn't seem to be the right tool for a very extensive Manova, with several dependent and independent variables.
If I were you, I would do four repeated-measures ANOVAs, as has been suggested before, using Huynh-Feldt or Greenhouse-Geisser adjustments for the degrees of freedom if epsilon is below 0.75, to accomodate for sfericity issues. The homoscedasticity of some of your dependent variables might benefit from log or other transformations. The p-values for those 4 omnibus anovas could be corrected through bonferroni, although it is a bit conservative, especially for positively correlated test statistics. I would definately go look for linear trends in your time and speed variables, through orthogonal polynomial contrasts, to boost power and interpretability.
Mixed linear models, in which you define yourself the variance-covariance structure of your measurements, provide a technique that is much more customizable, of which manova and repeated measures anova are special cases. However, it is not trivial for multifactorial designs with possible interactions. I don't think that Statistica handles that.
.

Relevant Pages

  • Re: Doubt in Regression
    ... Dependent <- Group 1 ... First Problem: ... This is my independent variable scale structure. ... All my independent variables are significantly ...
    (sci.stat.consult)
  • Re: JSH: The Math
    ... > Not surprisingly, since mathematics IS mathematics, my position can be ... I have picked m and f, a and b are dependent on them, though ... Therefore by the Harris Independence Principle (HIP), ... > The independent variables are just independent. ...
    (sci.math)
  • Re: time series multiple regression
    ... from its dependent variable. ... independent variables and the dependent variable, ... For a single predictor, I'd be tempted to use an ARIMA transfer function model. ...
    (sci.stat.math)
  • Re: time series multiple regression
    ... from its dependent variable. ... independent variables and the dependent variable, ... In econometrics it's pretty standard to run a multiple regression ... tried the Farnsworth package, but that does not appear to support ...
    (sci.stat.math)
  • Re: Univariate Analysis
    ... regression" should be used when we have more than one dependent ... variable and one or more independent variables (in the latter case, ... should probably speak of "multivariate multiple regression"). ... multiple independent variables, then we may want to call it ...
    (sci.stat.math)