Reflection on Malcolm et al.
The Malawi study had significant issues with both the theoretical framework of its model and the data collected to run though the model. The theoretical framework was supported by meetings with stakeholders, including government officials, local xxx, and NGOs. This theory first approach led to a model in which factors, such as the presence or absence of cell phones, the gender of the head of the household, and income were weighted componnents of a vulnerability score. One issue of the model is that the weights are determined by how important the researchers think they should be, not determined by data. A model like this, which attempts to measure future instability, cannot be checked for accuracy until a future calamity occurs. In addition, the ideas of what makes a stable and functioning household and what “stability” is are highly variable, and the model is invisibly influenced by those who weren’t consulted or deemed important enough to consult.
The data itself is another issue. In an attempt to preserve the provacy of those surveyed, Malcolm et al. randomized the location of household points within a 5km radius. What the study could have done instead was aggregate the household characteristics upward into the districts, which would have preserved privacy. This is a huge problem as some households change districts and some even end up in lakes or other non data units.