Larger tables - attendance at diabetic clinics

As an example of a more complex case, we might want to compare various methods of encouraging diabetic patients to attend their next routine clinic appointment. During their May visit to the clinic, all patients will be issued with an appointment card that details when they should attend in June, but then various additional measures may be taken. Patients are randomly allocated to 4 groups  

We then count those who do/do not attend their June appointment. The null hypothesis is that all four methods are equally effective. The results, expressed as a contingency table are shown in Table 16.3. 

Ideally the total number allocated to each approach would have been the same. However, this was not quite achieved. This is not a problem, because the chi-square test takes account of different column totals . For optimum power, extreme variation in group sizes should be avoided. 

The column percentages have been included and visual inspection of these suggests that the rate of attendance is higher where additional reminders have been used, but a formal statistical test is required to see whether the differences within these small samples would continue to be seen in the longer term. The result of a contingency chi-square test will be a P value of 0.006 (highly significant). 

1 The difficulty of interpreting the results from larger contingency tables 

---