Statistical modelling/Results of data exploration

Surveys often need less formal statistical analysis than experiments. Tables can often be compiled from the frequency tables calculated during the exploratory phase and presented without further analysis. It is only when one is unsure whether an observed differences is real or not that some statistical verification may be important. This may result in the use of simple t- or Chi-square tests or sometimes, where the data warrant it, in more complicated analysis of variance or logistic regression techniques. So far, under data exploration, we have discovered the following:

1. Males headed four fifths of the homesteads. No females 30 years of age or younger were heads of homesteads.

2. Livestock was important, as it was ranked as one of the three major sources of income. The others were crops and salary/wages. Homesteads ranking livestock as the primary source of income tended to have slightly more cattle on average than other homesteads although the average difference might not be statistically significant.

3. Cattle were important to the homesteads and were reported as the species of primary importance by over 90% of the homesteads. Cattle were mainly kept for the purposes of work/draft and cash from sales, though meat, milk and manure were also ranked as important. A method for summarising the relative importance of the different purposes by pooling the different rankings might be helpful.