Analyzing individual quiz scores with group quiz scores

[ashar malik]

Dear Statlist experts,

I have a data set of the performance of graduate course participants (n=30). Each course participant first attempted a quiz (n=4 quizzes and five BCQs in each quiz) individually. Next, the same quiz was retaken as a group activity by randomly assigning participants to groups (4-5 participants). We hypothesize that the quiz scores of groups are higher than the individual quiz scores. One option is to use an interclass correlation between individual scores and group scores. But it is a cumbersome process to compare individual and group scores for 20 questions/4 quizzes. Is there a better way to analyse this data?
An example data set is attached for 9 students for the week three test/quiz.iq31-iq35 individuals scores in questions 1-5 of the quiz, gq31-gq35 group scores of same questions. iq3Tot and gq3Tot are the total scores in the individual quiz and group quiz respectively.

[George Ford]

" We hypothesize that the quiz scores of groups are higher than the individual quiz scores."
ttest iq3tot = gq3tot
reg gq3tot iq3tot

[ashar malik]

Thanks, Ford, isn't it will give the difference between the two quizzes? Whereas I am exploring some sort of aggregate measure to prove if the intervention was effective.

[George Ford]

Are you trying to deal with the fact each student has 4 individual tests versus 1 group test?

[George Ford]

The means difference between individual and group scores is an aggregate measure of the intervention. (It seems you'd want to say: On average, group scores are higher than individual scores.) There could be other aggregation measures, but that is surely one. Maybe the variance of test scores is different (I suspect the group scores have lower variance).

[ashar malik]

#4 each individual had four individual tests and four group tests, each test's questions, group members and timings were different

[ashar malik]

#5 You are right: median scores are the same but the IQR of group scores is smaller than the IQR of individual scores. However, I want to understand the group dynamics. For example, member's role, how many members in the group had correct answers to a particular question, and whether they were able to influence the choices of those whose answers were wrong.

[George Ford]

are all the tests independent or is the same test given multiple times?

[ashar malik]

# tests are independent. Each test is based on a unique topic