The utility of the ASR

The small sociological blogosophere has been stirred up lately over the Journal impact factors.¹ [I can hear the collective yawn of those readers who are not sociologists; I'm afraid to say this will not get any more interesting to you]. Within Sociology, three journals have historically been considered the big three.

1. The American Sociological Review.
2. The American Journal of Sociology.
3. Social Forces.

Jeremy kicked off a bruhaha by calling attention to the relatively low factor assigned to Social Forces. Now I could care less about these impact factors; generally these journals don't publish the kinds of work that I do. But in the the long thread of comments to these posts several people put forth assertions concerning what constitutes "good sociology" that has me thinking. The ideas to follow are provisional and incomplete, but I think there is something important to be said about methodological discourse within Sociology. I'm starting to write a working-paper on this, so I welcome constructive comments.²

In this post, I wish to focus on one key question:
Does ASR truly reflect the cutting edge of sociology?

Several comments in Jeremy's thread got me thinking about the value of ASR. First, the infamous assertion made by anonymous at 2:37 a.m.

they like to publish boring articles that only other sociologists care about ...public has no use for most of the rubbish printed in "the big three", but every once in a while a study relevant to public will be printed.... Minimal publication of qualitative findings and impression management that sociology is a "hard science" by printing numerous formulas that researcher got from the latest (spss, stata, sas) manual.

This commenter (correctly in my view) was taken to task for sweeping generalizations. However nestled inside the invective is a point worth considering... is the ASR relevant? A subsequent comment on thread suggests it is ignorant to even ask such a question. Well, in my view, the best way to pierce ignorance is through empirical investigation... so let the examination begin.

I have to admit, my knee jerk reaction to that question is no, the ASR has not been relevant (to general public OR general sociology) for a long time. I polled my colleagues and found only one reads the ASR with any regularity and that person said that s/he doesn't find much of practical use for his/her own work.³ A senior colleague told me that s/he toys with the idea of running a call for papers on "When did the ASR become unreadable and what should we do about it".

However, I had to acknowledge that my knee-jerk reaction is also anchored in ignorance. As a tenure-track person, my reading time is spent on stuff directly related to my research program. And though I work in a mainstream area of Sociology (Criminology and Social Control) this area of work is not proportionally represented in ASR (or so I assumed).⁴ I started going through this calendar year's issues of ASR. I brought two with me (the top two on my stack): (vol 71, Number 5, October 2006 & vol 71, Number 1, February 2006).
Here are the titles from Feburary's issue {71(1)}

1. Comparative Perspectives and Competing Explanations: Taking on the Newly Configured Reductionist Challenge to Society, by Troy Duster, NYU (2005 Presidential Address)
   
2. Widowhood and Race by Felix Elwert & Nicholas A. Christakis, Harvard
   
3. Networks, Race, and Hiring by Roberto M. Fernandez (MIT Sloan School of Management) & Isabel Fernandez-Mateo (London Business School)
   
4. Wealth, Race, and Inter-Neighborhod Migration by Kyle Crowder (Western Washington University), Scott J. South (SUNY Albany), & Erick Chavez (SUNY Albany).
   
5. Deterring Delinquents: A Rational Choice Model of Theft and Violence By Ross L. Matsueda (University of Washington), Derek A. Kreager (University of Washington) and David Huizinga (University of Colorado)
   
6. liBlack and White Control of Numbers Gambling: A Cultural Assets-Social Capital View, by Darrell Steffensmeir & Jeffery T. Ulmer, Penn State. ** Followed by a comment and reply which I don't address here.

So among the empirical papers in the first issue of 2006 (not counting Duster) we see two papers addressing demographic concerns (Elwert & Christakis; Crowder et al), two papers on networks and opportunity structures (Fernandez & Fernandez-Mateo; Steffensmeier & Ulmter) and a paper addressing rational-choice behavior (Matsueda et al).

Are these papers important and relevant?

Having only read the papers superficially, I answer in the affirmative.

Elwert & Christakis explore the widowhood effect (that one's probability of death increases significantly upon the death of one's spouse). Using an enormous dataset (n=410,272) they demonstrate that the bereavement effect varies by race.

"Whites married to whites suffer a large and enduring widowhood effect. By Contrast, blacks married to blacks do not suffer a detectable widowhood effect, possibly because they manage to extend the survival advantage of marriage into widowhood (pg 16)."

This on its face strikes me as incredibly useful and relevant. Unfortunately, as I read the article, I'm lost on the take-home point. (e.g., what would make this paper relevant for non-sociologists). The bulk of the article focuses on measurement and modeling. They offer a useful conjectures in the discussion that the lack of a bereavement effect among b/b couples:

On one hand, blacks may not experience a widowhood effect because, unlike whites, they never gained survival advantages from marriage in the first place. On the other hand, blacks may have gained survival advantages from marriage, but unlike whites, manage to extend this marital survival advantage into widowhood (pg 36)."

They lack the data to directly determine which alternative is correct, but deduce the latter. In this regard, the paper is limited by logico-deductive model. That is, while they make a compelling argument through reason and deduction, at the end of the day their conclusion is essentially an educated guess (conjecture) which requires further refutation and testing (falsification). [Which is not to say its wrong... but the problem could be approached inductively, to paraphrase Mitch Dunier...to use shoe leather... to go and talk to people who have lost their spouses and see what the common (or uncommon) coping mechanisms are (not that I want to get into the Dunier can of worms right now, I'll post on that squabble later)].

Crowder et al analyze Panel Study of Income Dynamics data merged with census track data to estimate "modest effects of wealth on these patterns of inter-neighborhood migration (pg 72)." The topic in and of itself is interesting and relevant (what factors perpetuate segregation or facilitate migration). But unlike the previous article, I don't find this paper terribly interesting or useful.... Whereas Elwert & Christakis got me thinking about their conclusions, after skimming this paper, my reaction was... that's it? I'm not sure what's really new here. [The authors would say that they've uncovered some nuances in the effects of wealth by race....great, I'm just not convinced that this is something that I should encourage my students to read. Having just criticized it, I do think this paper is refreshing for an ASR empirical piece in that its methods are generally comprehendable to anyone with a basic grasp of regression analysis (OLS and Logistic models). In contrast Elwert & Christakis develop a statistical argument that is much harder to follow.

I'm going to close out this post for now without really answering my own question. The short answer is that yes, the ASR has utility. But the new question is: utility for what? Provisionally, I think the ASR is useful in that it brings together papers from across the discipline that (are supposed to) say something we all (sociologists) need to think about. In the two paper that I reviewed in this post (not that I read either really carefully... apologies in advance if one of the authors stumbles by) one got me thinking while the other didn't. Thus, I need to revise my kneejerk reaction to the ASR. But these thoughts are still forming.

Perhaps I'll start a series of posts on ASR articles, if nothing else to think about the arguments and evidence more carefully.

===
1. The impact factor is a proprietary measure from ISI web of science. It is designed to measure how much influence (through citation networks) a particular journal has. For the interested, see Dan Myers' comment on a long (and somewhat contentious) on Jeremy's initial thread.

2. Not that I get the kind of traffic that Jeremy gets, but to anyone who finds me through his sidebar, I'd really like to keep the commentary constructive on this topic. If you think my argument is off, tell me so! But please explain why and share an alternative.

3. Sorry for the clunky his/her anonymity. Since I publicly identify who I am and where I work and haven't asked these people if they want their views known, I am covering their identities.

4. I believe this claim can be empirically validated. However, given that two papers in the first ASR I canvass below are related to Criminology, I will reserve the right to correct myself at a later date.