ez_e
11-03-2006, 12:41 AM
I have two independent groups which I am looking at whether the type of responses differ across groups e.g. Group 1 produces significantly more type one response than Group 2 does (which I have categrised in to 3 diffeent responses). I guess I want to see if different frequencies of responses vary from one group to another but also withingroup differences regarding the type of responses that are most prominant.
I orginally thought a chi-square test but the independency of assumptions is violated because particpants within each group could produced 4 responses each and then each response was then categorized into one of three types (or other) for example:
particpant1: 2 *type 1 resonses/ 0* type 2 responses/ 2*type 3 responses (total 4 responses)
particpant 2: 1* type 1/ 1*type 2/ 2*type 3 (total 4 responses)
etc
combinations varied for each particpant then I added up total number of each response within each group- and I want to see whether group allocation had an effect on the types of response frequencies? Can not figure out how to statistically test this!
had it thought out- if only chi square could handle particpants contributing to more than one cell!
Then if there wasn't a sig difference in group- can you add these together and discuss as if it were one big group and look at difference between total frequences of types of response- same question- how to statistically look at this?
Another question- does any one know if you can use pie charts in APA style formats?? And whether this would be good to demonstrate above comparisons?
HELP HELP HELP
I orginally thought a chi-square test but the independency of assumptions is violated because particpants within each group could produced 4 responses each and then each response was then categorized into one of three types (or other) for example:
particpant1: 2 *type 1 resonses/ 0* type 2 responses/ 2*type 3 responses (total 4 responses)
particpant 2: 1* type 1/ 1*type 2/ 2*type 3 (total 4 responses)
etc
combinations varied for each particpant then I added up total number of each response within each group- and I want to see whether group allocation had an effect on the types of response frequencies? Can not figure out how to statistically test this!
had it thought out- if only chi square could handle particpants contributing to more than one cell!
Then if there wasn't a sig difference in group- can you add these together and discuss as if it were one big group and look at difference between total frequences of types of response- same question- how to statistically look at this?
Another question- does any one know if you can use pie charts in APA style formats?? And whether this would be good to demonstrate above comparisons?
HELP HELP HELP