I could see this making any kind of analysis of outcomes from arbitrary cohort extremely complex. This is outside my area of expertise and I'm definitely not a scientist (though I use it constantly). Perhaps I'm looking at this from the wrong perspective? It just seems that for any cohort the amount of variance would be so high that if you were to draw a conclusion from data it could already be wrong.

These would be two distinct yet connected sorts of data. The qualitative and the quantitative. One can disprove or question theories defined by the other, which is one reason why social sciences are in a permanent state of revision. Another reason is because your understanding of society is contingent on the very questions and framing devices you utilize.

Let's say you are investigating the birth of a city from an economic standpoint. Meaning that you wish to know what sort of trade or productive activity jumpstarted those first few cycles of capital accumulation. In the Americas you have cities like Rio de Janeiro which are so new that the documentation goes back to when there were only a few hundred families in the urban center + surrounding hinterland. So you have both the potential to realize patterns and variance, or the quantitative and the qualitative sides of your analysis.

You can look at specific families, what they did for a living, how they did well for themselves, what happened to their estates, what economic role mothers and fathers played, etc. And you can look at all families at once to find patterns. The quantitative analysis might help you make blanket statements such as 'the first and most primary form of capital accumulation for Rio de Janeiro was slavery, specifically the enslavement of native groups as part of allied wars and enslavement campaigns'. But if you go on to make another blanket statement such as, say, 'social mores in the wealthier sectors of colonial society followed a rigid patriarchical structure', then you'll find numerous examples of families where sometimes 3 full generations were outright led by the mothers and wives rather than the fathers and husbands. Sometimes because they appear to just be better at it than the men, sometimes because there were no men.

I'm framing it like this because that's how it works. We write our sociological narratives, and then someone else re-frames things a little differently and comes up with new and interesting inputs. It's not often that we completely disprove old, well supported theories. Rather, we end up refining them as scholars argue forever to the smallest detail.