I'm not talking about statistical significance. Statistical significance requires that your model at least attempts to correct for confounding factors. This one doesn't. It's a ballpark figure, and as such, it can tell you only whether you are at least roughly in the right ballpark. 1:4 is in the right ballpark. It is possible, of course, to be in the correct ballpark and still be offside, and a ballpark test won't allow you to decide that. To do that, you need something more precise. And if you want something more precise, you have to base your model on assumptions that aren't pulled out of my ass. But you're the one claiming offside. I'm not. So all I had to show was that the ball was not self-evidently offside. Which it isn't.

Besides, it's only 1:4 by one of the measures. The two other measures that have been put forward in this thread call it even. Furthermore, the measure that's 1:4 is the least appropriate one, because it assumes that all sides had an equal hand in all deaths that weren't from their own side, which is obviously nonsense.

- Jake

Friends come and go. Enemies accumulate.