Statistical Analysis for Education and Psychology Researchers

Class Levels Values

Religion 3 1 2 3

Number of observations in data set=24

General Linear Models Procedure

Dependent Variable:ATTRIB1

Source DF Sum of Squares Mean Square F Value Pr>F

Model 2 10.5833333 5.2916667 0.9 0.4671

Error 21 140.7500000 6.7023810

Corrected Total 23 151.3333333

R-Square C.V. Root MSE ATTRIB1 Mean

0.069934 15.08095 2.58890 17.1667

Source DF Type I SS Mean Square F Value Pr>F

Religion 2 10.5833333 5.2916667 0.79 0.4671

Source DF Type III SS Mean Square F Value Pr>F

Religion 2 10.5833333 5.2916667 0.79 0.4671

Figure 8.14: Analysis of variance from

PROC GLM

*** rank the residuals ***;

proc rank data=new ties=mean normal=blom;

var res1; ranks norml;

run;

*** plot to check the models assumptions ***;

proc plot;

plot res1*norm1;

plot res1*pred1;

title4 'test of normality';

run;

The option ties=mean is used so that any tied ranks would take the mean value, and the
option normal=blom converts the ranked residuals into normal scores. The variable
(norml) is thus a normalized score of the ranked residuals. A plot of the residuals (res1)
against the normalized rank residuals (norml) should give a straight line plot, and a plot
of res1 against pred1 should give a random scatter of points if the assumptions of
normality and homogeneity of variance are valid. Output from these plots is shown in
Figure 8.15.

Inferences involving continuous data 321