I am trying to proceed with nonparametric tests on SPSS with all my collected data.
I have approximately three groups of six subgroups and I am trying to calculate the signaficativity of these tests (Whitney...)for all my data.
My problem consists in attributing to every subgroup a value in my working sheet that will specify it belongs to one of the three groups.
I looked for a solution in manuals, web helpers and in my surrounding (I am working in a laboratory) but nothing for the moment.
Thank you in advance to all the IBM community...
J
Topic

Re: non parametric test on groups with subgroups of data
20130104T14:37:25ZThis is the accepted answer. This is the accepted answer.As Jon said please describe you data in more detail.
What are your independent and dependent variables? How are they measured?
Are the same 6 subgroups in each of the three groups? What are the groups and subgroups?
Why do you feel you need to use nonparmetrics? 
Re: non parametric test on groups with subgroups of data
20130104T18:02:17ZThis is the accepted answer. This is the accepted answer. SystemAdmin
 20130104T14:37:25Z
As Jon said please describe you data in more detail.
What are your independent and dependent variables? How are they measured?
Are the same 6 subgroups in each of the three groups? What are the groups and subgroups?
Why do you feel you need to use nonparmetrics?
All my data is from a behavioral experiment.
It involves different parameters : time, performance, and conditions of experiment.
We have three major groups of data respectively with three different conditions for the same parameter (here is frequency) . Inside each group we decomposed the experiment in six sessions.
My question is about the interdependency of all six sessions but the independence between the three groups.
Accordingly to literature these nonparametric tests are commonly used to evaluate the samples.
Either we can merge data from two groups and form two independent samples which according to my manual will allow me to apply either the independent sample ttest or the MannWhitney U
In order to test my sessions (supposing they're dependent) I would like to use the withinsubjects ANOVA (Post hoc test : Tukey's HSD)
and then analyse globally my data for the three independent samples (groups) and use betweensubjects ANOVA (Post hoc test: protected ttest).
Correct Incorrect
Ses1 Ses2 Ses3 Ses4 TO Ses5 TO Ses6 Ses1 Ses2 Ses3 Ses4 TO Ses5 TO Ses6
839,78 631,17 720,67 972,30 918,13 848,89 1390,89 1025,11 879,86 1030,47 1262,11 1248,40
this is only an example of the data we are manipulating with several software.
Thank you.
J 
Re: non parametric test on groups with subgroups of data
20130107T13:50:37ZThis is the accepted answer. This is the accepted answer. SystemAdmin
 20130104T14:12:10Z
Please describe your data in more detail. A small extract that shows what you want to do would be helpful.
my data is a combination of related performance and time measured in ms retrieved from several protocols. These protocols constitute our groups and each protocol is divided into sessions that we can eventually bin but which constitute our subgroups ultimately.
these are the general lines that sketch my data.
Thank you in advance.
J