News article - "white" names on resumes.

[William Barnes]

One of the many articles about the study which shows that "white" names on rusumes got more responces than "black" names.
Article

[frank davis]

Originally posted by William Barnes:
One of the many articles about the study which shows that "white" names on rusumes got more responces than "black" names.
Article

There was a 3.4% (10.1 vs 6.7) difference in the response rate based on name. What would the standard deviation and range of error? Normally in such studies the range of error is given and you would expect MIT people to know that. Why didn't MIT relesse those numbers? Notice the timing of the release of the study (Upcoming high profile Supreme Court case, MLK Day). Someone is playing political games.
Even if the range of error were zero, is the study really enlightening? Does the study really tell us anything new, not about black vs white discrimination, but standard vs non-standard names? Since the 1940s even white movie actors, singers, and other performers have changed their names to be more standard or mainstream and gain greater acceptance. Any statiscally unusual name could have the same response. Although in the IT field I don't believe that would be the case with Chinese or Indian names.
Given that the range of error could be as large as the 3 or 4, which is likely given that identical resumes except for the names were not sent to the same businesses, is a 3.4% difference something to get really "disturbed" about?

[William Barnes]

To start I was not attempting to make any political point. I was posting a current news item I found interesting.
But I do find some errors in your post.

given that identical resumes except for the names were not sent to the same businesses

Actually the news article states that identical resumes were sent to the same business.

There was a 3.4% (10.1 vs 6.7) difference in the response rate based on name.

Well I don't know squat about stats but it sure looks like you doing your math different than I would. Sure those number look close, but not if they always fall in the favor of the "white" name. Isn't that the point?

... something to get really "disturbed" about?

Yes and no. But when you put these results along with studies which consistently show (no I don't have any URLs) that minorities who show up for job interviews, or to rent apartments, they are less successfull/rejected at a higher rate than white applicants who have the exact same skills/backgroud. Than ya, it shows a pattern of discremination.
[ January 22, 2003: Message edited by: William Barnes ]

[Thomas Paul]

The issue is that if the error percent is within the range then the study proves nothing. It's like those political polls that say Gore is leading Bush 46% - 44% but the error range is 4%. Which means Gore could be leading 50% - 40% or behind 42% - 48%.
There are other variables. Assuming the two resumes arrive at the same time, which got read first? Would you as an HR person think that two identical resumes might be odd? Do we know the race of the person that viewed the resumes? It doesn't sound like a very scientific study to me.

[Paul Bull]

William and Thomas,
You both bring up probing questions about the statistical significance of this study. However, to me, your responses seem to be a tad reactionary. I think its safe to assume that there still are significant problems in the area of discrimination in this country and I'm sure there are plenty of studies out there to support this assumption. Although we hear plenty of reactionary rhetoric from the certain individuals
on the left that exaggerate the issues of race, I also find many people, including many my of friends and family, who are too quick to dismiss any suggestion that discrimination is still a problem in America

[Thomas Paul]

I'm not dismissing racism... I'm dismissing the study! Just because a study gives a result that makes sense does not make the study worth anything.

[Pakka Desi]

BTW, how would you make out black names from the white ones? Don't blacks and whites (in US) use the same set of names?
I have observed that hispanic names are very typical..they sort of end with -ez (Lopez, Harnandez, Martinez etc.) It would be very easy to make out Indian names from American names too. But black and white names? :roll:

[frank davis]

Originally posted by William Barnes:
To start I was not attempting to make any political point. I was posting a current news item I found interesting.
But I do find some errors in your post.

Yes and no. But when you put these results along with studies which consistently show (no I don't have any URLs) that minorities who show up for job interviews, or to rent apartments, they are less successfull/rejected at a higher rate than white applicants who have the exact same skills/backgroud. Than ya, it shows a pattern of discremination.
[ January 22, 2003: Message edited by: William Barnes ]

I did not say or mean to say that you were making a political point; I was referring to the authors of the study. Nowhere in the URL that you posted did I see the phrase "identical resumes". Maybe they did use identical resumes, but the URL given does not say that.
How are HR people going to react when they see identical resumes? The situation of receiving word for word resumes is so unusual that we could be narrowing the validity of the study to that narrow set of situations rather than making a broad case for racism. Not that I'm denying racism, just that I'd prefer the debate not to be further inflamed by exagerating the significance of a potentially flaws study.

[frank davis]

Originally posted by Paul Bull:
William and Thomas,
You both bring up probing questions about the statistical significance of this study. However, to me, your responses seem to be a tad reactionary. I think its safe to assume that there still are significant problems in the area of discrimination in this country and I'm sure there are plenty of studies out there to support this assumption. Although we hear plenty of reactionary rhetoric from the certain individuals
on the left that exaggerate the issues of race, I also find many people, including many my of friends and family, who are too quick to dismiss any suggestion that discrimination is still a problem in America

I didn't see where anyone dismissed racism. As I originally mentioned, if we don't know the standard deviation or the range of error, then the study really means absolutely nothing.

[William Barnes]

> BTW, how would you make out black names from the white ones? Don't blacks and whites (in US) use the same set of names?
There are names which are associated with Afican Americans in the US. The news story below explains it a little.
Having a hard time finding the actual study on the net. But here is an article from the NY Times which has a little more info.
Article

[Thomas Paul]

You have to register to read it... I hate that!

[frank davis]

Originally posted by William Barnes:
>
Having a hard time finding the actual study on the net. But here is an article from the NY Times which has a little more info.
Article

Now, I'm really confused. First William Barnes says I'm in error because I said the employers were not given identical resumes, then he gives a URL for the NYT were they explicitly say :
"No single employer was sent two identical r�sum�s".
What's the real story?
Still no mention on the range of error and standard deviation. I would expect that MIT people
would have that info. Why the NYT chooses not to publish it (other studies and polls routinely give range of error) is interesting.

[Mark Herschberg]

Originally posted by herb slocomb:

Still no mention on the range of error and standard deviation. I would expect that MIT people
would have that info. Why the NYT chooses not to publish it (other studies and polls routinely give range of error) is interesting.

The MIT people didn't write the article. It was the journalist/editor who choose what information to include in the article. Why not try to find the original study?
--Mark

[frank davis]

Originally posted by Mark Herschberg:

The MIT people didn't write the article. It was the journalist/editor who choose what information to include in the article. Why not try to find the original study?
--Mark

Original article here :
http://www.econ.yale.edu/seminars/strategy/st02/bertrand-021204.pdf
Range of error is not mentioned, but I think it is equal to 3 standard deviations which in this case appears small which is good.
Interestingly enough, the black names "Jermaine" and "Leroy" did better than 65% of the white male names. Meaning that blacks with those names got higher call backs than 65% of the whites (!?). So a black name is an advatange if its the right black name. But again, I think this just underscores that some of the issue revolves around people having preferences for certain familiar names. To white people, some of the black names sound unfamiliar and the negative impact could result more from that than a prejudice against black names in general. A familiar name has been known to be an advantage in marketing people since at least the 1940s when white actors and other performers changed their names to something easily pronounceable or at least more mainstream sounding (Kirk Douglas's real name was Demsky Isadore).
[ January 24, 2003: Message edited by: herb slocomb ]

[William Barnes]

Thanks for finding the original, it was interesting to read.