Also, statistically test if there is any commonality between people with the same aspects (sun square jupiter for example) or aspect patterns (t-square for example). Of course dozens upon dozens will have to be tested and the birth dates would have to be checked (using 2 or more sources). Maybe such tests could be refined with certain factors being isolated to check for statistical commonality.
But it isn't going to work. There's a huge difference between a received square - in this case Aries to Cancer, and an irreceptive square - Libra to Capricorn.
Is it a day chart or a night chart?
Is either the Sun or Jupiter besieged? Which houses are they in, and which do they rule?
Are they in aspect to other planets?
And any number of other variables.
You could concentrate on something like determining temperament (Lilly or Bonatti style, perhaps?), as neuroscientists are now calling it 'seasonal biology but definitely not astrology' - only astrologers used some of the same formulas (the season you were born in counts heavily towards temperament and has done for many, many centuries - we're going back to Aristotelean philosophy here). But astrology does have the jump on neuroscience in that regard, as we also use moon phase, planets in or aspecting the ascendant, the chart almuten, etc., to describe temperament, which is both physiological, and for want of a better term - psychological.
You can look up the studies on neuroscience and temperament by birth season, so you can see where they've got with it.
That might prove fruitful. It's a reasonably standard measuring technique within astrology, and whilst it cuts fine enough to give definite results, it doesn't cut so fine as trying to sort out what an aspect in isolation means. And you've got the advantages that it's standardised, and something that science is looking into now.
The societal and especially corporate misuses possible - pretty much inevitable - if it's proved true are terrifying, but if you want to do this, that's probably as good a starting point as any.