[Pilfer]

>but does not strongly suggest that close to normal people on average have a difference

A valid point, but that does not apply to Google engineers because their hiring practices directly discriminate for ability, as they select the best ~15% of people they interview. Statistically, someone with better than normal ability is far more likely to receive a job from Google than 'normal people on average'.

[ggm]

I contend that even seeking to select for the best 15% they are unlikely in practice to get people who actually lie outside a range we'd consider normal. Not the least because I believe a significant number of the savant are functionally unable to survive an interview process which extends beyond the machine.

The above-normal extends from the mildly interesting to the truly unbelievable. Perhaps I'm seduced by the latter and you want to consider the former. I'd agree that google and like companies hire above-normal. I still content that the profile of ability is probably within spec of normal, even if they aim high.

I hasten to add, I doubt I'd pass a google interview. Haven't tried, don't expect to be tapped. And, I know many fine current and ex- google people and they are smart. The thing is, they all show remarkably high verbal skills, men and women both. Which makes me (bias alert: particular to general warning) wary of a measure which implies there is a detectable skew. These people, display both abilities.

[Pilfer]

You are playing word games with the phrase ‘considered normal’. The phrase has different meanings whether you consider it from a statistical (as the submitted scientific paper would) or colloquial (as you are) standpoint.

Google does not publish hiring statistics, but I genuinely believe most Google engineers are hired from the top 5% of the general population. In a statistical sense, that could be considered as above normal. However, in a colloquial sense, there is no clear definition of what is ‘considered normal’, hence you could consider persons with top engineering ability as normal. Heck you could even correctly consider everyone as normal, and therefore all Google engineers are normal.

Since this thread is about a statistical interpretation of a population in the US & India, I hope you can see how your use the colloquial sense of the phrase is misleading here.

[ggm]

Interesting you chose 5 and others say 15. because, 15 is close to 1 std deviation. So, my sense that at 15% its likely that people are basically normal, is a crossing point of what the statistics say as a heuristic, approximate signal, and common colloquial meaning. somewhere at 15-20% its really not a big deal to be in that space.

you went 5%. thats the tip of the iceberg. if google can succeed in hiring only from the 5% and get no fake-it-till-you-make-it, or mismeasures, Then sure: its beyond normal in any sense we know.

I'm not a believer. I don't think they can hire to that at the scale they want to be, and not break a huge barrier of cultural assumptions about scores, and measures. They can certainly hire only the top 5% or 1% by degree outcome, and I'd believe they do that too.

[Pilfer]

I am sorry, but you are still making the same mistake I pointed out earlier, conflating statistical concepts with colloquial ones. You don’t seem to understand the point I am making at all. To understand a statistical argument like the paper submitted, or respond to the statistical argument in the ‘diversity’ memo, you require a basic understanding of statistics. 5%, 15%, those exact numbers I was throwing out, are meaningless here. From a statistical standpoint, what matters is the distribution. The distribution of traits for individuals at the 50th percentile (using whatever metric), is not the same as the distribution of individuals at the 85th percentile, nor the 95th percentile. What you need to understand is that for the general population, the distributions of traits are not identical at every percentile. That is the key misunderstanding you demonstrate in your original comment. Further, your claim that “somewhere at 15-20% its really not a big deal” is not easily disprovable, but statistical nonsense (that’s a colloquial statement).

I do encourage you to re-read my comments and understand where 15% and 5% come from. They are from different populations, something which you do not appear to notice.

Sadly there’s a litany of basic mistakes in your comment. It’s not worth my teaching people high school statistics, so I won’t be replying.

[ggm]

I got schooled. I guess I deserved it.