Maybe: it’s easier to capture a whole load of diverse stuff if you don’t care about numerically quantifying it, and dont care about statistical significance tests, multiple testing, etc. Once you have a a candidate list of qualitative features that might be interesting, you can then ask: ok, how do I numerically measure this thing?
Maybe: it’s easier to capture a whole load of diverse stuff if you don’t care about numerically quantifying it, and dont care about statistical significance tests, multiple testing, etc. Once you have a a candidate list of qualitative features that might be interesting, you can then ask: ok, how do I numerically measure this thing?
But if you have a whole load of diverse stuff, how do you know what matters and what doesn’t matter?