One of the consequences of owning a farm in Central Brazil is that I frequently meet people who do things and have life narratives that we are unlikely to come across in texts on sociology or epidemiology. In the nearby town they live in streets that have postal codes that would place all the residents in the same cell of a contingency table.
But in each residence lives a family with a separate and unique story. I believe that it is worth asking whether it is really the perceived similarities that determine their health outcomes. Yes, perhaps when sanitation is involved, but there are other health outcomes that would not be centered on the postcode.
I have spent the last two years helping to link notifications of interpersonal violence with birth and mortality records. The idea is to find some way to stop men harming their partners. If this can be achieved by changing the way the law reacts to the violence then these linkages may prove useful, especially when legal and penal records are included in the studies.
But what if what needs to be changed is the way a boy is treated in his first year of life? It is unlikely that information collected at the time of the violence would be accurate about events in early childhood. How could record linkages tell us that we should be looking elsewhere? I have no idea. But I believe it to be the most important question in population studies today.
Statisticians are always pleased to tell us that we have failed to prove something. We need a methodology that tells us where to look. Also it must be based on something with more possibilities than those currently offered by diluted Marxism.