Findings, implications and lessons learned

Some statistical implications of this case study are as follows.

  1. It is seen from this case study that survey analysis does not always require formal statistical evaluation. Inferences can be made from tables and graphical presentations when the results are obvious. The graphical presentations give overall trends. Of course, when the results are not obvious, then they must be statistically quantified, and inferences made only if the results are statistically significant.

  2. The case study has shown how sample size determines how far the data can be classified. Thus, it was appropriate to present the distribution of the gender and age of head of homestead only down to the sub-regional level and not dip-tank area level. It is important when planning a survey to decide what questions need to be asked and whether the sample size is large enough. It is generally necessary in terms of cost and manpower to limit the sampling to what can be managed. However, there may be a corresponding 'cost' in terms of how far the data can be classified.

  3. One also has to be aware that proposed sample sizes are not always achieved. Thus, some questionnaires in this survey had to be discarded due to incomplete or missing information, particularly the key information needed before any subsequent entries could be made, and so the sample size in each dip tank area often ended up being less than the target of 10 homesteads.