(go to Outline)Are the conclusions appropriately based on the survey results or other data presented in the discussion section? It is very difficult to prove cause-effect based on the results of a single survey. Finding both a high cumulative prevalence of diarrheal disease and a high prevalence of acute malnutrition does not necessarily mean that the diarrheal disease caused the malnutrition. Authors who state that their survey does this are probably over interpreting their results. In addition, the conclusions in many survey reports seem to have little relationship to the data gathered by the survey and the results of data analysis.
Are the results of a survey extended beyond the sampling universe? A survey done with a sample taken from one province cannot be generalized to a neighbouring province. The survey report authors may point out reasons why the neighbouring province may have a similar health or nutritional status as the surveyed province, but they should not try to reach absolute conclusions.
Are the conclusions consistent with known scientific fact? Many things are already very well known about malnutrition and other health indicators. If the survey seems to contradict these well-known principals, perhaps there is something wrong with the survey. For example, if a survey shows a very high prevalence of acute malnutrition in young children in the midst of a dysentery outbreak, but a mortality rate in young children which is below normal, there is probably something wrong. We know that malnutrition, especially when combined with disease outbreaks, kills children. If there is a lot of both malnutrition and disease, children are probably dying at an elevated rate, and the survey results should reflect this.
Do the conclusions seem to be motivated more by politics than survey results? This is a very common problem. No one is completely neutral. Survey managers must be careful to prevent their opinions from influencing their survey results or conclusions. A survey report that shows this type of influence is automatically suspect.
Are the results put into a larger context? Because of the inherent limitations of surveys, survey results must be supplemented by other information which can explain the results and provide program guidance. The results should be compared to those of prior surveys or surveys done in similar populations.