4-7 logically makes more sense than 3-6. Why would you leave out no. 7 when analysing bowling the the middle order? Nonsensical. And no.3 comes out against the new ball very regularly.
Again, no. I didn't use Rabada or Cummins because they were change bowlers and (with Cummins at least) have usually been change bowlers, so of course they're going to have a higher percentage of middle order wickets compared to Starc. I picked other opening bowlers to compare to Starc, who also plays mostly as an opening bowler. (I think Morkel has been a change bowler a lot, but he opened this game which is why I chose him)
There was no bias here, stop trying to look for one because you'll get shot down every time. You're entire argument was based on a false assumption, just accept it and move on.
Don't you think there is a reason these guys are playing as opening/change bowlers? Players get selected as opening and change bowlers because of how well they bowl in those respective positions.
Rabada is a better opening bowler than Morkel (he has opened on numerous occasions). He bowls at the middle more because it is better for the team that he do so, because he is relatively better- a Philander/ KG opening combo would be better, but not so sufficiently better as to outweigh the change disadvantage. KG often bowls with an older ball because the others are worse with an older ball than he is.
I don't know how Cummins ranks as an opening bowler compared to his teammates, but he is bowling change at least in part because his teammates are poor at the middle. Marsh bowls at the middle because the relative disadvantage of having him bowl at the middle is less than the relative disadvantage if having him bowl at the top/tail- Starc and Hazelwood are significantly more likely than Marsh to take wickets with a moving ball but only slightly more likely to take wickets with a ball that is not moving, because they are significantly more effective with the moving ball and less effective with a ball that isn't moving.
Im not sure if you have encountered the term statistical bias before. It has a different meaning to bias in everyday speech, and is not an insult. Selection bias is at work here- choosing a sample of statistics that are not broadly representative of the entirety. I'm not saying the bias is intended, but there is always bias in stats, intentional or not. It is impossible to get a perfect representation of a sufficiently complex system. You seem to get the impression I'm accusing you of cherry picking based on a misunderstanding of the term.
You are comparing him to 2 bowlers who are similarly known to be good at the top and poor against the midde. Even if you compared him to every opening bowler ever there would still be bias- why only openers? Change bowlers bowl at the top order and opening bowlers bowl at the middle order with sufficient regularity to impact results, so why discard this information? Sure, you can argue that they have different roles, but you are just chosing between two different forms of bias here. How about spinners or medium pacers that opened the bowling...
Then you would want to filter for statistical noise (it's a different game to the 19th century...). But to do so introduces further bias.
Any selection of batting positions represents a bias. You can argue that your selection is a less imperfect selection, which I wouldn't disagree with, if for no other reason than the fact that I cherry picked mine. But you can't argue that it is a perfect selection. A number 3 can functionally be an opener (1st ball wicket) or a middle order player (long opening stand). A number 7 is not typically a batman, but can be specialist bat quality or not. And there are a practically unlimited number of other variables at play here. Would it make much of a difference here chosing 3-6 or 4-7. Likely not, or at least not compared to the bowlers selected, but this has no impact on it being an imperfect selection.
Stats are useful in cricket in that they can take out subjective (non-statistical) bias, but a much larger sample size is required in order to be statistically meaningful, and some level of subjective interpretation is still required. The less perfect the representation is (ie. your sample of a handful of bowlers), the greater the level of interpretation required becomes.
I've given you a logically coherent argument argument that you don't disagree with why Starc can't be very good without swing. You can determine pretty easily by eye that Starc isn't a great bowler without swing. Either of these alone should be sufficient to convince you that your reading of Starc's stats is a result of statistical bias and/or noise given how small the sample is.
