4
$\begingroup$

I am analyzing student performance on a pre/post test designed to assess gains for a week long high school biology curriculum on natural selection.

After comparing pre/post performance using a variety of different metrics (test score, Rasch measure, chi-square per item) between the paired samples, I conducted an ANOVA to gauge whether there were differences between teachers. Two of the (seven) teachers are significantly different from the others (their students have drastically lower mean Rash measures and did worse overall), so I'd like to account for these differences in the analysis.

Is a nested ANOVA model appropriate here? I know nested models aren't in the SPSS dialogs, but I'm not much of a programmer and thus do not know what syntax to produce.

Moreover, each teacher has a different number of students. Would a random sample (ensuring the same amount from each teacher) be useful for paired comparisons of means (like pre vs. post), or does it not matter since it's a comparison of means between groups anyway?

Apologies for any obvious questions. Education is not my original field (I'm a bio-anthropologist), and it's been a bit since I took multivariate stats in grad school.

$\endgroup$
2
  • $\begingroup$ Hi @mfernan5, I guess I need a little help understanding your research question. If the question is pre/post changes in scoring, I am not sure why differences in teacher + student cohorts is important. You say some groups of students did drastically different; does this mean they changed on the pre/post less than the other groups, or that their pre-scores were lower? Rasch (and everything else) is quite sensitive to sample size, could you please share the sample sizes for each teacher-student cohort and let us know the size of the standard errors reported. Thanks! $\endgroup$ Aug 1, 2014 at 3:03
  • $\begingroup$ Hi Doug, I managed to run a nested model ANOVA in SPSS by tinkering with the syntax a bit, but to answer your questions: 1 - Two teachers (from lower SES schools) had drastically lower Rasch measures, overall scores, and correct response %'s than other teachers 2 - Sample size is a bit small (n=308), with anywhere between 30-80 students from any of seven teachers 3 - Standard error in the nested model (for rasch measure) was .069, I can't seem to find the output for the non-nested analysis of the population but could find it for you. $\endgroup$
    – mfernan5
    Aug 12, 2014 at 18:19

1 Answer 1

5
$\begingroup$

I figured this out after some searching and consultation with an old stats professor.

To employ a nested model ANOVA in SPSS, some editing of the syntax must be done, and the nesting effect is determined in the /DESIGN command. In the end, my syntax looked something like this (depending on your preference of parameters and posthoc tests):

UNIANOVA

  Y  BY A B

/METHOD = SSTYPE(2)
/INTERCEPT = INCLUDE
/POSTHOC = A B ( BONFERRONI )
/EMMEANS = TABLES(A) COMPARE ADJ(BONFERRONI)
/EMMEANS = TABLES(B) COMPARE ADJ(BONFERRONI)
/PRINT = DESCRIPTIVE ETASQ
/CRITERIA = ALPHA(.05)
/DESIGN = A B(A).
$\endgroup$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.