The power of genetic diversity

[Jacob]

Seems pointless to argue about the borders of Europe and Asia.

Yeah. Better to have a land war.

[DGuller]

the utility of proxy variables is in direct proportion to the degree of their relation or connection to that which they are proxying for, but you already knew that.

Of course.  However, I'm not picky, I'll take any incremental improvement in predictive power.

If the thing one is attempting to proxy for is some sort of genetic characteristic, then traditional racial categories can be problematic because their connection to genetic characteristics is typically very loose at best and accidental. (not surprising given that, as stated before, these categorizations preceded any clear understanding of genetics).  To take derspeiss' example, if one wants to proxy for risk of sickle cell anemia, then what one might want to look for is the extent to which one descends from persons who lived in tropical areas at high risk of malaria.  Traditional racial categories like "Black" are rather suboptimal for that purpose, as they are simultaneously overinclusive (many "Blacks" do not trace descent to such locations and at are low risk) and underinclusive (e.g. failure to include certain persons tracing descent from persons living in certain South Asian locales who are at high risk).

This is a good example of how proxy variables, and true underlying causative variables, compare.  Of course it is more useful to know where someone is descended from for the purpose of screening for sickle cell anemia.  If you know that, knowing race is probably useless, because someone's descent would explain all the variation in outcomes that race would, and then some.

However, so what?  That's the nature of any proxy variable.  They're by definition a rather "dirty" variable that you'd rather not be using.  You'd rather be using the underlying causative variables.  You'll get more predictive power that way, and you'll get a better understanding of dynamics at play.  You can also effect outcome changes by controlling the causative variables, in situations where it's applicable, while controlling the proxy variable is just shooting the messenger.

So why use proxy variables at all then?  You use them because you either don't know the causative variables, or it's impractically hard to observe the causative variables.  If you know someone's descent for screening for sickle cell anemia, use that.  But that seems like a more difficult variable to observe and ascertain than just eyeballing someone's race.  Depending on the application, greater predictive power may not be worth the cost.

[derspiess]

Seems pointless to argue about the borders of Europe and Asia.

I find geographical arguments like this entertaining, since there seems to be a lot of variation as to how world geography is taught in different parts of the world (e.g., the debate I started here a while back as to how many continents there are).

[The Minsky Moment]

But that seems like a more difficult variable to observe and ascertain than just eyeballing someone's race. 

Except you don't eyeball race.  You eyeball physical appearance and that is the true proxy that is being used here.  "Race" in this example is just a rather sloppy way of categorizing physical appearances.

[DGuller]

But that seems like a more difficult variable to observe and ascertain than just eyeballing someone's race. 

Except you don't eyeball race.  You eyeball physical appearance and that is the true proxy that is being used here.  "Race" in this example is just a rather sloppy way of categorizing physical appearances.

Yes, you are correct.  The true proxy variable is the perception of what race someone belongs to.  That's a shorthand that I've used for pretty much the whole thread.

[The Minsky Moment]

Seems pointless to argue about the borders of Europe and Asia.

I find geographical arguments like this entertaining, since there seems to be a lot of variation as to how world geography is taught in different parts of the world (e.g., the debate I started here a while back as to how many continents there are).

I've seen the Caucuasus referred to both ways (ie part of central asia and not).  It's an arbitrary characterization; couldn't care less either way.  If one wants to be literal, geographically speaking the center of Asia would be around Irkutsk; on a population weighting probably somewhere around Lhasa.

[Razgovory]

Seems pointless to argue about the borders of Europe and Asia.

I find geographical arguments like this entertaining, since there seems to be a lot of variation as to how world geography is taught in different parts of the world (e.g., the debate I started here a while back as to how many continents there are).

When Languish was down, there was this huge argument on EUOT whether or not Poland is in eastern Europe or not.  I didn't know that Poles were so sensitive about such things.  I said that if they didn't like being Eastern Europeans they could be West Asians if they like.

[garbon]

But that seems like a more difficult variable to observe and ascertain than just eyeballing someone's race. 

Except you don't eyeball race.  You eyeball physical appearance and that is the true proxy that is being used here.  "Race" in this example is just a rather sloppy way of categorizing physical appearances.

Yes, you are correct.  The true proxy variable is the perception of what race someone belongs to.  That's a shorthand that I've used for pretty much the whole thread.

That's terrible. It likely wouldn't do me much good if my physician thought my background was the Middle East.

[DGuller]

That's terrible. It likely wouldn't do me much good if my physician thought my background was the Middle East.

The cases where race is most difficult to classify are also the cases where race would be at its most useless as a predictive proxy variable.  Think about it.

[Neil]

"Race" in this example is just a rather sloppy way of categorizing physical appearances.

It's not very sloppy at all.

[Jacob]

When Languish was down, there was this huge argument on EUOT whether or not Poland is in eastern Europe or not.  I didn't know that Poles were so sensitive about such things.  I said that if they didn't like being Eastern Europeans they could be West Asians if they like.

Of course they're East Europeans:

1. They speak a Slavic language.
2. They're a former Soviet vassal.

That makes them East Europeans. Once upon a time, they may have been part of "Central Europe" or some such place, but there's no "Central Europe" after WWII.

Now, given time the categories may shift again but right now, Poland = Eastern Europe.

[Ideologue]

Or, as it is also known, "Bad Europe."

[garbon]

That's terrible. It likely wouldn't do me much good if my physician thought my background was the Middle East.

The cases where race is most difficult to classify are also the cases where race would be at its most useless as a predictive proxy variable.  Think about it.

Of course, my example easily works the opposite way.

[Crazy_Ivan80]

It's apparently Ohio state law that all newborns be blood-tested for sickle cell.  They had to draw some blood from my baby girl's foot a couple weeks ago & I asked if we could just skip it.  The nurse rolled her eyes & said she wished they could since the hospital had never had a test come back positive at this hospital (it's located in my township which is like 98% white).  She did say they did have several that did come back positive when she worked at University hospital downtown.

So my poor little girl had to have her foot poked for no reason

odd that it would only be for that. Imho, it's more likely that it's the Guthrietest that happened
http://en.wikipedia.org/wiki/Guthrie_test

[The Brain]

Killing fascists test?