Date: Tue, 14 Jun 2005 21:09:40 -0700
Reply-To: thecrickethunter@YAHOO.COM
Sender: "SAS(r) Discussion" <SAS-L@LISTSERV.UGA.EDU>
From: thecrickethunter@YAHOO.COM
Organization: http://groups.google.com
Subject: ANCOVA with repeated measures
Content-Type: text/plain; charset="iso-8859-1"
Dear all,
I posted the following question at the SPSS discussion group, and I've
got no response. Hopefully, some of you are bilingual and respond to my
question. I recorded cricket calling songs of four populations that
were divided into two zones, allopatric and sympatric zones. Sympatry
means the cricket species occurs with other cricket species of
interest, and allopatry means the cricket species occurs alone. In
addition to these two variables ('zone' and 'population'), I also noted
'temperature' and 'age' for each calling song recording, which
may be covariates.
A list of variables:
Zone: (allopatric or sympatric)
Pop: population, nested within zone, four populations, 2 for each zone
Indi: individuals, 8-14 individuals per population
Tmp: temperature, varying covariate
Age: age, varying covariate
Calling song characters: response variables
Both covariates, tmp and age, are varying covariates. Each cricket in
a population was recorded once in one of three temperature ranges
(18-20, 22-24, 26-28 degree C). I tried to record calling songs for all
three temperature ranges for each cricket, but not all crickets were
recorded for all three temperature ranges. I included crickets that
were recorded at least twice for the analyses.
My ultimate goal from this analysis is whether there is a significant
zone effect. That is, whether calling song characters were different
between allopatric and sympatric zones throughout the temperature
range. The following is the syntax of my analysis.
MIXED cro BY zone pop indi WITH age tmp
/FIXED = zone pop(zone) indi(pop(zone)) | SSTYPE(3)
/METHOD = REML
/PRINT = COVB LMATRIX TESTCOV SOLUTION G
/RANDOM = age tmp age*zone tmp*zone | COVTYPE(ID)
/REPEATED = age tmp | SUBJECT(indi)
/TEST = 'Contrasts of zone' zone 1 -1
/TEST = 'Contrasts of population' pop(zone) 1 0 -1 -1
/TEST = 'Contrasts of population' pop(zone) 0 1 -1 -1
/TEST = 'Contrasts of population' pop(zone) 1 -1 0 0.
In the results, the zone effect seems to be nothing as you can see
below.
Type III Tests of Fixed Effects
Source Num df Denom df F Sig.
Intercept 1 61.0 18.3 .000
Zone 1 9.3 .40 .541
Pop(Zone) 2 39.2 28.9 .000
Indi(Pop(zone)) 42 203.3 2.9 .000
However, in an separate test when I asked the contrast question, the
allopatric and sympatrix zones are different.
/TEST = 'Contrasts of zone' zone 1 -1
whether the sympatric and allopatric zones are different?
Contrasts Estimates SE DF TestValue t Sig.
L1 1.44 .70 15.8 0 2.1 .055
Why are these two results different? In a graphic analysis, there seems
to be interaction between zone and temperature. Is this interaction
anything to do with these results? Any advice would be appreciated.
Cricket Hunter
