Anova - repeated measures

[Andrzej Andrzej]

Hi, I am completely new to Stata and want to do anova with repeated measures (in GUI).
1. My question is what to put in: Repeated measures variables line in a window: anova - Analysis of variance and covariance ?
I have got DV-'score" variable, IV: categorical (between subject) - "liquid" variable and IV: indexing, repeated measures variable: "time_points".
When I mark: two-way full factorial in "Create varlist with factor variables" window it seems to be working, but
when I additionally tick "Repeated measures variables" in a window: anova - Analysis of variance and covariance, placing time-points variable over there, that throws an error:
"could not determine between-subject basic unit; use bseunit() option".
2. Does Stata have got an option to quickly bring last analysis window back, sort of like in SPSS, in order not to go through all steps all over again ?
3. What does Stata regard as factor variable ? Is it a long, float, double with a values' labels assigned ?
Any help would be very much appreciated, thank you.
Andrzej

[Carlo Lazzaro]

Andrzej:
welcome to this forum.
As far as I know, Joseph Coveney is the lister who contributed the most on this (and many other) topic(s).
Take a look at his post on this forum.

[Joseph Coveney]

For your first question, looking at the error message that you got, I assume that you neglected to include the variable for subjects. You mention variables for outcome, between-subjects and for distinguishing time points, but I don't see one for subjects.

I cannot answer your second question, but I recall that the GUI allows you to "submit" a tentative set of instructions without closing the fill-in window. I know that you can use various function keys to recall recent commands entered at the command line, but I'm not sure about restoring a fill-in window on the GUI.

For your third question, the variables specified by factor-variable syntax as categorical may be in any of the numerical datatypes (float, double, byte, integer, long integer). They do not need to have value labels assigned to them. The only requirement that I am aware of is that they need to be nonnegative integers. In your case, the assumption for Stata's anova and manova estimation commands is that all of the predictors (explanatory variables) are categorical unless otherwise indicated with a c. prefix. So, for those two commands you don't need to put the i. factor-variable prefix, even for specifying an interaction, unless you want to make the command line syntax explicit for the record.

My recommendation, in general, is to skip the GUI and get familiar with the command syntax. There are worked examples for repeated measures ANOVA in the online help file for the anova estimation command that you can use as a template to get yourself started.

[Andrzej Andrzej]

Thank you very much indeed to both of you for your kind help and explanation. I am coming here from SPSS world trying to understand by analogy what's possible to do and how.