The public life of a dissertation

by Eszter Hargittai on March 20, 2004

It is not often that “a dissertation gets written up in the New York Times”:http://www.nytimes.com/2004/03/20/arts/20DEVA.html so I thought it was worth a mention here. Kieran has written here about Devah Pager‘s work earlier including a bit of context. Since Devah is a friend of mine, I would like to add that not only is she really smart and great at finding innovative approaches to research questions, but she’s also a delightful person. It’s wonderful to have people like her in academia and in sociology in particular using her talents to work on important questions… and it’s also nice to see good academic work get public recognition for a change!

Update: Be sure to check out Kieran’s note in the comments for more details about the public life of this dissertation.

{ 12 comments }

1

Rook 03.20.04 at 4:34 pm

Could ya fix that link please…………..

2

Herman 03.20.04 at 4:39 pm

the hell with the link..I turned to other link and found One very Hot Chick! post more pics, please and cut the sociology nonsense.

3

Steven D. Krause 03.20.04 at 6:42 pm

A very interesting article and I would agree that it is nice to see someone getting attention for their scholarship. But in a way, the conclusions of her study don’t really strike me as that “new” or “shocking.” I recall seeing a piece on one of these TV magazine news shows like NBC’s “Dateline” or ABC’s “20/20” several years ago where two groups of young men with identical references and resumes went about to get an apartment, a job, etc. One group was white and one group was African-American. The results turned out pretty much the same as this study.

4

DJW 03.20.04 at 6:58 pm

Agreed that it’s not that surprising, but since so many people wish to deny this aspect of social reality, it’s very good to have solid research to point to (in addition to honing our understanding of racial phenomenon such as this).

Eszter/Kieran/other sociologists, do you understand Heckman’s objection? In the article, it seems rather incoherent, except the sample size complaint.

5

Kieran Healy 03.20.04 at 7:16 pm

I’m going to claim a small measure of credit for this, in my capacity as the plankton in this media food chain. Brad DeLong saw the post that “I wrote”:https://www.crookedtimber.org/archives/000386.html about Devah’s dissertation and “blogged about it himself”:http://www.j-bradford-delong.net/movable_type/2003_archives/002052.html, which in turn was was read by David Wessel who wrote “a column”:http://www.careerjournal.com/myc/diversity/20030916-wessel.html for the _Wall Street Journal_ about the research. From there it was on to Howard Dean’s stump speech and now the _Times_. Of course, Devah’s research has intrinsic value and immediacy that it was only a matter of time before it bubbled up to media prominence by itself.

Steven says:

the conclusions of her study don’t really strike me as that “new” or “shocking.”

That’s not the point. It’s one thing to do a quick comparison on a TV show with one or two people and another to do a properly controlled, well-designed audit study that allows you to pin down the effect. Excellence in social science isn’t always about coming up with something no-one’s ever thought of before. You need to get the nuts and bolts right because the belief that the labor market itself is basically unbiased with respect to irrelevant characteristics like race or gender is phenomenally strong amongst many people. (Ascriptive characteristics are thought to really boil down to differences in innate ability, human capital or preferences.) Which brings us to

Heckman’s objection? In the article, it seems rather incoherent, except the sample size complaint.

Heckman complains that it’s an audit study rather than an experiment. Devah sent out matched pairs of resumes to employers to see what they would do. But this isn’t quite the same as a classic controlled double-blind study, which would be extremely difficult to do in this case, particularly with anything resembling a large sample. Heckman is basically saying that while this is a really well-executed study on labor market discrimination, his _a priori_ belief in unbiased and efficient labor markets is so strong that it would take something more — the second coming of Jesus Christ, perhaps — to convince him to change his mind.

6

drapetomaniac 03.20.04 at 10:00 pm

But in a way, the conclusions of her study don’t really strike me as that “new” or “shocking.”

oppression is not an entertainment, and is generally sadly lacking in novelty or surprise value. it is worth remarking on until it is no longer true.

7

drapetomaniac 03.20.04 at 10:08 pm

I’m going to claim a small measure of credit for this,

Start collecting props, connecting plots, Networking like a conference, because the nonsense has yet to stop… :-)

8

DJW 03.21.04 at 1:35 am

Thanks Kieran.

9

John Quiggin 03.21.04 at 1:53 am

Kieran, this is interesting, since it reverses the conventional wisdom (to which I have contributed) that the ecological relationship of blog to old media is that of parasite or predator – blogs sitting at the top of the food chain, rather than, in this case at the bottom.

In relation to academic research, this relationship is still present (for example, my post on Cavanagh), but examples like yours are likely to predominate in future. The same will probably be true in lots of areas where there is a substantial group of bloggers with specialist knowledge.

10

Jacob T. Levy 03.21.04 at 2:25 am

Steven Krause wrote: But in a way, the conclusions of her study don’t really strike me as that “new” or “shocking.” I recall seeing a piece […]where two groups of young men with identical references and resumes went about to get an apartment, a job, etc. One group was white and one group was African-American.

It’s worth emphasizing that what’s new isn’t documenting race effects in job market discrimination. It’s documenting that those effects are larger than the effects of a felony conviction.

This finding also, I should note, does serious damage to the ‘rational racism’ hypothesis, i.e. that whites are amateur racial-profilers who use race as a (reasonable approximation of a) proxy for other, genuinely undesirable, traits, especially a proclivity toward crime. It’s absurd to think that race is a better proxy for likelihood-to-commit-a-crime than is a felony conviction.

On the other hand, I wonder whether the results would have been any different using a felony other than coke possession– embezzlement, larceny, auto theft. I doubt it, but I wonder.

11

Alexander Crawford 03.21.04 at 10:50 am

Kieran…

You should restrict yourself when you attribute claims or propositions to her research to only her own explicitly stated conclusions. The labor market samples are time-specific means across broad occupational categories and only lend themselves to heavily qualified and general indentifications of trends… which she’s careful to note.

Furthermore, without a great deal of secondary reading there’s no way to test those systemic models and conclusions of other scholars that are used as initial premises; their publication dates are far apart (some older than I am) and the geograpic spread irregular (for example, D., is the model from the UK really translatable?). And the code used to derive the intersection angles slopes should be included (diatonic shouldn’t be used on selected % of samples without showing the super-set as well).

Yeah, I’m nitpicking. I don’t have any serious objections, but would suggest the occupation piece led to some hasty generalizations from poltical hacks.

12

Barry 03.21.04 at 12:49 pm

Kieran:

“But this isn’t quite the same as a classic controlled double-blind study, which would be extremely difficult to do in this case, particularly with anything resembling a large sample.”

I’d guess impossible, since that would mean that the subjects would have to be randomized to race, and left unaware of which race they were.

Of course, if Heckman has that level of objection, then perhaps he should leave his chosen field of economics, and go into one where the most common method is not running models on aggregated observational data.

Comments on this entry are closed.