Right, I think you’d want to include a control group for carries (i.e., groups where one player is notably better than the others). So the actual comparison would use the “carry” control group’s %-of-the-time-that-green-circle-getter-dies-first as the baseline, rather than just a naive 25%.
I don’t think you could establish anything definitive about a particular match, but I think it’s possible to draw inferences about matches in general. I agree that if you actually wanted to do this, you’d want to include some kind of control for ordinary carries. And of course, as you say, it’s just finding correlations (you could certainly set up an experiment to actually get at causation, but I can’t imagine FS did anything so elaborate). But, if you’re a game company trying to decide on whether to add a feature to your game or not, a rough correlation of the “this thing seems to correlate a bit with selfish/anti-teamwork behavior” variety could very well be the thing that makes them shrug their shoulders and think, “well alright, let’s just leave it out then”.
Having observed green-circle-chasing behavior myself anecdotally in VT2 – and also having watched players chase K/D ratios in other games, to the detriment of their teams – I do think that their explanation re: “toxicity” is a plausible one. But who knows, really – they could have just left it out for the same mystifying reasons they left out a lot of stuff from VT2, and all this stuff about toxicity is just some kind of post-hoc rationalization. FS works in mysterious ways.