Statistical Analysis for Education and Psychology Researchers

Interpretation

The obtained p value of 0.5715 (two-tailed) is clearly not significant at the 5 per cent
level.

Computer Analysis

The calculation of exact probabilities for each contingency table is tedious and time
consuming. PROC FREQ in SAS produces both one-tailed and two-tailed probabilities
for Fisher’s exact test. Data from the worked example using the following lines of SAS
code is shown:

data fisher;

input row $ col $ count @@;

cards;

1 1 3 1 2 1 2 1 3 2 2 3

;

To analyze this data PROC FREQ is used with the weight statement (this specifies the
variable that contains the cell frequencies). The appropriate SAS code is:

proc freq data=fisher;

weight count;

table row* col/nopercent chisq;

title “Fisher’s exact test”;

run;

The complete programme is shown in Figure 6, Appendix A3.

Interpretation of Computer Output

Computer output for the 2×2 contingency table relating to Fisher’s exact test for the
group and handedness data (data used in the worked example) is shown in Figure 6.5. For
clarity of presentation only the relevant sections of output are illustrated.
The probabilities shown in Figure 6.5 are the same as those calculated in the worked
example. The null hypothesis cannot be rejected, and we conclude there is no difference
between the two age groups in the proportions of left-handed subjects with upper limb
injury. In other words the observed frequency distribution could have arisen by chance.
Table of row by col
Row Col^
Frequency^
Row Pct^
Col Pct 1 2 Total^
1 3 1 4

75.00 25.00

50.00 25.00

Inferences involving binomial and nominal count data 183