Summary

This case study describes the design of a questionnaire survey to assess the importance of cattle to subsistence farms (homesteads) of the Hhohho region in Swaziland and the methods of cattle production under which they are raised. These data represent a small subset of data taken from a large livestock breed survey carried out in Swaziland on cattle, sheep, goats, donkeys, pigs and chickens. One of the main purposes of the case study is to demonstrate methods for handling data from surveys and in particular the application of methods of analysing multiple response data. The statistical package INSTAT is used as well as GenStat.

Data exploration using tables and graphs, keeping in mind the multiple response nature of the data, are first undertaken to determine the general data patterns, trends and distributions. To confirm the exploratory results, further data analysis is carried out using an index method (based on ranking) to assess the importance of cattle to the homestead and the purposes of raising cattle. 

In cases where conclusions are not obvious, Pearson's chi-square test is applied to test the differences in proportions of homesteads giving different responses. REML is also used to compare average cattle numbers raised in homesteads in which livestock was either the primary source of income or not.

Finally, aspects of methods of reporting of results are briefly described in terms of the different types of people (from researchers to farmers) who will read the information.