Statistical Analysis for Education and Psychology Researchers

101

001

;

** Do not edit beyond this

line **;

Figure 6.8: Data entry section of SAS

programme for computing Cochran’s

Q

The statement data a; is the beginning of the data step. Following the input statement
are the variables t1 t2 and t3. These need to be numbered consecutively and separated by
a space. They correspond to the treatment periods (measurement occasions) in the
repeated measures design. The cards statement indicates that data lines follow. The SAS
system recognizes the end of the data when the first line after the last data line contains a
single semicolon.
Output from this analysis is shown in Figure 6.9.

Cochran Q Test Results

Cochran Q test value=1.75

p-value with liberal df:

df=2

Not significant at 5% level (p-value

=0.417)

p-value with conservative critical value:

Not significant at 5% level

Figure 6.9: SAS output from

Cochran’s Q analysis for vocabulary

data

Interpretation of computer output

The value of the Q test statistic is the same as that calculated in the previous worked
example. Interpretation is also the same. The SAS programme prints out the significance
levels for both liberal and conservative tests, the p-value is given for the liberal degrees
of freedom. The programme automatically tests alpha at 1 per cent and alpha at 5 per cent
and prints a p value for either 1 per cent or if this is not significant the p value for the 5
per cent level. A warning is printed if the number of cells is less than 24.

6.8 Summary

In this chapter the use of tests has been discussed where data has been discrete and
observations have been classified in contingency tables. The role of scores has been to
label and classify data. Whereas these procedures are suitable for many occasions they do
not make use of as much of the information contained in data as is sometimes possible.

Inferences involving binomial and nominal count data 201