0
$\begingroup$

I have the following data:

IV = 2 levels (Experimental Group + Control Group).

DV = I'm using signal detection variables i.e. (Hits, CRs, FAs & Misses)

Both groups were tested twice so therefore I have 2 sets of DVs for each group (i.e. Hits 1 & 2, CRs 1 & 2) and so on.

I want to compare between-group differences as well as within subject differences on both tests. How do I carry a repeated measures analysis?

Thanks!

$\endgroup$
  • $\begingroup$ What software are you going to use for that? $\endgroup$ – ttnphns Aug 31 '11 at 17:12
  • $\begingroup$ Hi there. I'm going to use SPSS. $\endgroup$ – Tanuja Aug 31 '11 at 17:51
  • $\begingroup$ Just one more thing, I originally carried out independent samples t-tests to check for differences in the DVs between the two groups. Additionally I also checked for differences between the 2 sets of DVs, i.e. was there anyy differences between Hits1 and Hits2? $\endgroup$ – Tanuja Aug 31 '11 at 18:00
  • $\begingroup$ I was told that I shouldn't be using multiple comparisons without a Bonferroni type correction when my variables are not independent. My supervisor suggested I use a tabular method like chi-square and a statistician has suggested repeated measures. Now I'm unsure of how to proceed! $\endgroup$ – Tanuja Aug 31 '11 at 18:03
  • $\begingroup$ Pointless to speak about correction for multiple comparisons - you have only 2 groups and only 2 time levels. $\endgroup$ – ttnphns Aug 31 '11 at 20:02
3
$\begingroup$

OK, you have variable GROUP with 2 values and dependent variables HIT1 HIT2 CR1 CR2 FA1 FA2 MISS1 MISS2. I assume that the DVs are all interval (scale) measurement level and approximately normal distribution.

You go to General Linear - Repeated Measures menu. Enter within-subject factor name TIME with 2 levels. Enter 4 measures names: HIT, CR, FA, MISS. Press Define, allocate your DVs properly in Within-subject variables field, move GROUP in Between-subject factors field. If you Paste your syntax you will get this

GLM
  hit1 hit2 cr1 cr2 fa1 fa2 miss1 miss2 BY group
  /WSFACTOR = time 2 Polynomial
  /MEASURE = hit cr fa miss
  /METHOD = SSTYPE(3)
  /CRITERIA = ALPHA(.05)
  /WSDESIGN = time
  /DESIGN = group .

Now you may run your analysis. Find within-subject (TIME) effects and interaction effect (TIME*GROUP) in table Univariate tests. Find effect of GROUP alone in table Test of Between-subject effects.

There are also other tables, including Multivariate effects, which I won't comment on in this pretty rough instruction. Good luck.

$\endgroup$
  • $\begingroup$ Hi, thank you so much. This definitely answers my question. I appreciate that you answered so fast to my urgent request =) $\endgroup$ – Tanuja Aug 31 '11 at 20:18
  • $\begingroup$ Hi there. I have run the analysis and I have one more questions. I understand how to interpret the table for the within subject effects. However, I am confused regarding the effect of GROUP in the table Tests of Between-subject effects. I found that the effect of GROUP is significant for FA; however, how will I know if there was a differences in FA1 between GROUP 1 & 2, or if this difference is in FA2 between GROUP 1 & 2? Do I need to carry out post-hoc analyses to know which one is involved? Thanks! $\endgroup$ – Tanuja Aug 31 '11 at 23:39
  • $\begingroup$ One more thing. On top of the Tests of Between-subject effects, it's written "Transformed Variable: Average". Did SPSS automatically average HITS 1 & HITS 2 etc? $\endgroup$ – Tanuja Aug 31 '11 at 23:48
  • $\begingroup$ I want to be analyse on there were group differences on Hits 1 or 2. $\endgroup$ – Tanuja Sep 1 '11 at 4:18
  • $\begingroup$ Yes. In Between-subject effects table, 2 times (eg FA1 & FA2) are averaged, so it tells whether 2 groups differ on that averaged variable. To see whether groups differ on one time (eg FA1) while don't differ on the other (FA2), look TIME*GROUP interaction effect. $\endgroup$ – ttnphns Sep 1 '11 at 7:05

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.