Hi forum,
Ive inherited data from a, lets say, rather “tricky” quasi-experimental design. Its “basically” repeated measurement. 8 classes of 3 schools were measured 4 times.
Altogether, each student was measured 4 times. Each cluster has 20+ students. The outcome measures were school scores. That’s the easy part. Now comes the tricky part:
1. The four tests used were not standardized tests (different topics, difficulty may be also inconsistent). In fact, every time (t1, t2, t3, t4) a different test stimulation was applied.
2. Every second test situation was a different test motivation for the students. T1 and t3 were “easy” testing situations. T2 and t4 were “serious” testing situations.
Yes, before you wonder: No controls, no baseline, just 4times measured differently and confounding effects abound.
And, one of the favoruite resreach questions is: Are there differences between the uneven and even measurement times (t1,3 vs. t2,4)?
Any idea whether this could be modelled by HLM or mixed modelling?
I appreciate every bit of advice!
Cheers!
HLMmouse