I had this idea that measuring an automobile's "handling" ability is sort of like measuring a person's intelligence. It's a kind of nebulous concept that summarizes stuff like turning radius, brake sensitivity, center of mass, and a whole bunch of other known and unknown characteristics.
Car experts could probably create a strict ordering for the handling of different models just by driving them, and from that you could fit the results to a curve and come up with a single metric for how well a car handles, a "general handling factor," h.
In talking about intelligence I've had a lot of debates break down to something along the lines of "g (or IQ) is real" vs. "g is not real," and I think this analogy reveals how those sides are talking past each other: h is pretty clearly "real" in that it represents a real property of a car and it would be useful to us in talking about cars and comparing different cars to one another.
On the other hand, h doesn't correspond to any single physical component of the car's construction: you can't just pop out one "handling mechanism" and replace it with a new one. And a statement like "this car corners well because it has a high h factor" ends up sounding kind of incoherent: it's probably sort of true, because cornering ability and handling are going to correlate very highly, but it implies a causal mechanism that isn't there. I think that's the problem of "reification" that people complain about when it comes to g.