Nested within-subjects repeated measures stats?

#1
Hi people,

I am having some confusion over how to carry out the stats for my experiment.

In the experiment, there is two condition of teaching method "prof-generated" and "student-generated" and the DV is students' quiz scores over 4 quizzes. As I carried out the experiment in a real classroom setting under my prof's module, I could not do a fully crossed design and had a lot of restrictions. Instead, all students first go through the "prof-generated" condition in the first half of the semester and took quiz 1 and 2 in the same period. Following that, they then go through the "student-generated condition" in the 2nd half of the semester and took quiz 3 and 4 in the same period.

I am trying to find out the effectiveness of student-generated condition over prof-generated but realised time could be a factor that could affect students' improvement in quiz scores as a natural course of the module.

At first, I did a 2-way repeated measures anova and got significant results in treatment factor and insignificant results in time and interaction. However, my stats tutor told me that my time factor is nested within treatment and I cannot do that. I tried doing linear mixed model in SPSS but got very very confused as I have never learnt that before (even after hours of youtube videos).

I tried putting DV as quiz scores, treatment as fixed variable, time as fixed variable and random variable and subjects as random variable. Time was insignificant and treatment was significant but had no idea if i analyze and interprete the results correctly and how to show that time is not affecting the quiz results and only treatment does (specifically student-generated > prof-generated).

Thanks in advance:)
 
#2
Depends on what you want to do. Are you trying to determine rate of growth? Or do you only care about change from quiz one to four?

Also you do have nesting to worry about but it's in the nesting of students with different teachers. This is pretty easy to take care of with a binary variable.
 
#3
Depends on what you want to do. Are you trying to determine rate of growth? Or do you only care about change from quiz one to four?

Also you do have nesting to worry about but it's in the nesting of students with different teachers. This is pretty easy to take care of with a binary variable.
Hi,

Just to clarify.

In the experiment, all students go through two conditions consecutively - "prof-generated learning" followed by "student-generated learning" as part of a module in a normal semester.

Throughout the semester, they undergo 4 quizzes (2 during each of the two conditions). Hence Quiz 1 and 2 were conducted during "prof-generated learning" period and Quiz 3 and 4 were conducted during "student-generated learning" period.

H1: Student-generated condition is more effective than prof-generated condition (in terms of quiz results)

H2: Time course of the module does not influence results of H1 (Not a linear function of quiz results improvement with time).

Could you explain on why nesting of students with different teachers occur since all students go through both conditions consecutively. I understand that Quiz 1 and Quiz 2 is nested within prof-generated condition and Quiz 3 and Quiz 4 is nested within student-generated condition. Am I suppose to use linear mixed model? And if so, what should I be doing? Binary variable as in putting results of quiz 1 and 2 = 1 (prof-generated) and quiz 3 and 4 = 2 (student generated) under a variable of treatment?