Output from this analysis is shown in Figure 8.19. The lsmeans statement produces least-
square means for the response variable attrib1 split on the independent variables religion
and sex. The standard error of the lsmean is also produced (code is stderr) along with a
test of the null hypothesis that the lsmeans are equal (code is pdiff). The means statement
produces means instead of lsmeans (in this example they are both the same). Use of the
estimate statement is described in section 8.7 under the heading ‘A priori and post hoc
multiple comparison procedures’.
General Linear Models Procedure
Class Level Information
(^) Class Levels Values
(^) sex 2 0 1
(^) rel 2 2 5
Number of observations in data set=24
Dependent Variable: ATTRIB1
Source DF Sum of Squares Mean Square F Value Pr>F
Model 3 47.0000000 15.6666667 3.00 0.0547
Error 20 104.3333333 5.2166667
Corrected
Total 23 151.3333333
R-Square C.V. Root MSE Mean
0.310573 13.30487 2.28400 17.1667
Source DF Type SS Mean Square F Value Pr>F
sex 1 42.6666667 42.6666667 8.18 0.0097
rel 1 4.166666 4.1666667 0.80 0.3821
sexrel 1 0.1666667 0.1666667 0.03 0.8599
Source DF Type III SS Mean Square F Value Pr>F
sex 1 42.6666667 42.6666667 8.18
rel 1 4.1666667 4.1666667 0.80
sexrel 0.1666667 0.1666667 0.03
Least Squares Means
Sex Attribl lsmean Std Err lsmean Pr>|T| H0: lsmean=0 Pr>|T| H0: lsmeanl=lsmean2
0 18.5000000 0.6593347 0.0001 0.0097
1 15.8333333 0.6593347 0.0001
Rel Attribl lsmean Std Err
lsmean
Pr>|T| H0: lsmean=0 Pr>|T| H0: lsmeanl=lsmean2
2 16.7500000 0.6593347 0.0001 0.3821
5 17.5833333 0.0001
General Linear Models Procedure
Level of -----------ATTRIB1--------
(^) SEX N Mean SD
0 12 18.5000000 2.39317211
1 12 15.8333333 2.03752672
Level of -----------ATTRIB1--------
REL N Mean SD
Statistical analysis for education and psychology researchers 336