The issue with this line of reasoning is that it avoids complex notions of cyclical reinforcement, as well as lower level, and likely systemic, bias. That is,

The issue with this line of reasoning is that it avoids complex notions of cyclical reinforcement, as well as lower level, and likely systemic, bias. That is, there are potential ideological underpinnings that lead to the rather by-the-book statistical analysis. When Lombroso attempted his measurements of criminality, and leveraged what appeared to be an exhaustive bit of data against the subject, inevitable analysis of the data ensued. What was missed however, were the vastly more complex subjects of “What is criminality” and “Who defines criminality” for example. In a way, the data became a cyclical reinforcement of the (false) hypothesis, while ignoring the entire sociological dimension. Very few have stopped to question the metrics and models, and instead only see the data. In this case, it may be prudent to not stop at the 70% data, and dig beneath it to evaluate the ideology and biases that _may_ have played a role in the symptom. It is seductive to lean on Western Rationalism / Scientism, but we conceivably could do well to question the systems that give rise to the models that yield the data.