To be fair, you would need a bigger more representative company. If your slice is CEOs' then that is pre-filtering for a lot of things. Tech role at apple skewed a fair bit, general role less so, a retail apple role thena different population all together.
Amazon possibly has enough employees in the fuller distribution of income / qualifications that would be an interesting dataset for slicing, but of course you will always find rouge slices and while they may have a reasonable explanation othertimes they do not - why are 90% of CEO of a listed companies are >50 years old, 22% of the population is under 18 its unfair they are not represented (lets not get into the number of old people in politics)
I don't think the question of why CEOs are older than non-CEOs is interesting. Highly important roles require suitable experience, which of course is going to bias the sample towards older people.
The problem is that, if the only answer if there's no "reasonable explanation" is to imply the existence of some kind of discrimination, you're going to include a lot of weird things under that category, and often make complex problems appear simple.
Scott Alexander wrote a good article "Black People Less Likely" that discusses a similar idea. There's often possible explanations for local bias, ideas of unfair treatment and rude old white people, but sometimes you hide a lot by addressing at a local level. If what is identified as a problem is present at all levels of the hierarchy from social clubs to CEO positions, it's useless to ask why more xyz people aren't CEOs.
Amazon possibly has enough employees in the fuller distribution of income / qualifications that would be an interesting dataset for slicing, but of course you will always find rouge slices and while they may have a reasonable explanation othertimes they do not - why are 90% of CEO of a listed companies are >50 years old, 22% of the population is under 18 its unfair they are not represented (lets not get into the number of old people in politics)