we need standardized dissertations

One of the problems of graduate education in many fields is that the requirements for the dissertation are vague. Another issue is that the dissertation is a book length treatment, even in fields where articles are standard. This leads students spend years writing overly long documents that have little value. For that reason, I encourage all my students to use the “three essays” format as the default. It’s simple, it works, and they’ll get done. If they have a good reason for deviating, then we can talk about it. But most folks should really stick to “three essays.”

There is now more systematic research showing that this advice is correct. A recent AER paper authored by Wendy Stock and John Siegfried shows that economists who use the “three essays” format do better in terms of academic job placement and subsequent publication. The abstract says it all:

Dissertations in economics have changed dramatically over the past forty years, from primarily treatise-length books to sets of essays on related topics. We document trends in essay-style dissertations across several metrics, using data on dissertation format, PhD program characteristics, demographics, job market outcomes, and early career research productivity for two large samples of US PhDs graduating in 1996-1997 or 2001-2002. Students at higher ranked PhD programs, citizens outside the United States, and microeconomics students have been at the forefront of this trend. Economics PhD graduates who take jobs as academics are more likely to have written essay-style dissertations, while those who take government jobs are more likely to have written a treatise. Finally, most of the evidence suggests that essay-style dissertations enhance economists’ early career research productivity.

My take home message? We should drop the pretense of the sprawling dissertation. All departments should require or strongly encourage the three essay format as the default. If the student wants something else, they need to make the argument.

Hat tip to our evil twin, Organizations and Markets.

Adverts: From Black Power/Grad Skool Rulz 

Written by fabiorojas

May 29, 2013 at 3:35 am

20 Responses

Subscribe to comments with RSS.

  1. I only accept dissertations in CSV format with designated fields for primary method, secondary method, primary literature, secondary literature, etc. I’ve got a Perl script that does most of the work of running a defense but I’m hoping to get it to generate letters of recommendation and pipe them to LaTeX.



    May 29, 2013 at 5:00 am

  2. Because as we all know it, standardization and compliance to the market rules are the best ways to innovation and creativity. It makes me think of Raymond Aron’s interrogation about Weber’s rationales (1967). It gives something like that: “The main pattern of the world were we live in is rationalization. (…) The whole modern society tends to be “zweckrational” and the philosophical issue of the present time, the highly existential issue, is to circumscribe society domains were subsists and must subsist action of an other kind”. I thought – and I still do, actually – that science, human science, was one of these domains.



    May 29, 2013 at 10:03 am

  3. It appears the effect in the top programs is only for early adopters:
    “For 1996–1997 graduates of Tier 1 and 2 programs [top 15 programs], an essay-style dissertation is associated with roughly two more publications than a treatise. In contrast, academics in the 2001–2002 cohort from Tier 1 and 2 programs who wrote essay-style dissertations had only half as many
    publications as their counterparts who wrote treatise dissertations.”

    I would also note that this appeared in American Economic Review: Papers & Proceedings, so I would want to see the authors’ Excel spreadsheet before putting much faith the analysis.


    neal caren

    May 29, 2013 at 12:43 pm

  4. FYI: AER P&P ≠ AER



    May 29, 2013 at 1:57 pm

  5. First, what Neal said.

    Second, the econ job market and tenure requirements are different from sociology’s in a couple of ways which lead me to think that heterogeneity in sociology is probably much more functional than it would be in econ. First, econ grads are judged largely, though not exclusively, on the brilliance of their job market paper. Search committees do not expect this paper to be published, nor is the student expected to have a long cv of prior work. The other papers in the dissertation are almost after thoughts – things you plan to eventually publish after you get a job with the one really good paper. So, the three essay format is extra functional, and books make very little sense for most people. I am quite curious what the “book-style” dissertations from the Tier 1 sample even look like in recent periods – I wonder if they are not just three essay dissertations with more than the usual level of integration and front matter.

    Compare this to sociology. In soc, job candidates are more or less required to have substantial cvs before hitting the market at this point. The modal UM job candidate has a first or sole-authored AJS or ASR plus at least one other thing. You’d think that would push towards the essay-style dissertation, and I’m guessing there has been some shift in that direction, especially for quant-y folks. That said, it’s also pushed book-style people to engage in smaller empirical projects and/or spin off bits of their books as standalone publications before hitting the market. Generally, my sense (and perhaps yours is different) is that the dissertation itself has become much less important in hiring decisions, and the applicant’s existing publications have become much more so (though often the two are highly connected). Also, soc values books more than econ. One exception to the modal UM job candidate (perhaps the second, smaller mode) is a historical or qualitative scholar coming off a post-doc with a book contract from a fancy press (sometimes also with a pub based on the book). Similarly, books count heavily towards tenure in soc (at least at some departments, including many at the top of the status hierarchy); they don’t, I think, in most econ departments (nor the related fields, like Strategy or Management, that hire many economists).

    Tl;dr: sociology PhD students still write book-length dissertations because sociology’s job market cares less about the dissertation itself (in the absence of other markers of quality, like publication) than econ and because sociologists value books more.


    Dan Hirschman

    May 29, 2013 at 2:44 pm

  6. “Because as we all know it, standardization and compliance to the market rules are the best ways to innovation and creativity.”

    Actually, yes. Is your argument that a dissertation publishing practice that descends from the dictates of the academic journal market of the 16th century “the best way to innovation and creativity?” The one thing that has actually characterized markets since Weber stretched out in his armchair, is loony rates of innovation and the upending of standards.

    The idea that the academy is an ethically pure place which “circumscribes the market” was invented by Greek aristocrats and carried forward by the religious and intellectual patrons of aristocrats for thousands of years. Nothing could be less innovative and creative that repeating it over and over again.

    Put that in your third essay and smoke it.


    Graham Peterson

    May 29, 2013 at 2:48 pm

  7. I’ll just echo what everyone else said about the differences between sociology and economics. It’s a bad idea to generalize from the economics job market to what happens in sociology.

    That said, I think the three essay model is fairly common (and becoming more prevalent) in sociology, in part because the pressure to publish articles prior to graduation has encouraged students to write dissertations that help them accomplish this goal while also finishing up their dissertation requirement. Even students who are planning on writing books from their dissertations are encouraged to think of at least one chapter that will be their ASR/AJS article. Business school phds have been working under this model for years. The idea is to conceive of each chapter of your dissertation as an article that will already be published prior to graduation or be submitted shortly after landing at your first job. It makes sense given the high bar that we place on graduate students to publish in the first couple of years on the job. Young faculty want to be moving on to their second and third projects in the year after settling into their new job.


    brayden king

    May 29, 2013 at 3:26 pm

  8. There seems to be a tension in nlalld’s comment? He invokes Aron to say that science (human science) should be a domain that pushes back on rationalization? But hasn’t science–and its application across a number of fields– been an essential component of this march towards increasing rationalization in modernity?

    What’s the suggestion? We only use a systematic approach to understand the world that lies outside of the academy? So program requirements and the structure of programs within the University should remain vague and unstructured because such an approach fosters the creativity/innovation required by a graduate program?

    The issue I have with this sentiment is that the academia as an institution (whether we like it or not) has some inherently purposive/instrumental properties. We are training people for positions, both within and outside the academy. And yes, I know it’s not solely instrumental, nor am I arguing that it should be.

    To me, the problem is not with standardization per se (Gabriel’s snark definitely made me laugh), but rather when standardization becomes an end in and of itself. I think there is a middle ground. If we find that some disciplines are worse at getting students to the degree and ultimately placing students in positions that were worth the effort and cost, then I think it is imperative for the disciplines to try to understand why and how the process could be improved. Granted, these are questions that still require considerable empirical substantiation. But I think it’s fair of Fabio to attempt to identify what might help (though as Neal and others point out, the evidence for the three essay approach is still murky).

    And don’t get me wrong, I think the idea of circling the wagons around the academy and creating an insular world untouched by the realities of modern (capitalist?) world is appealing (utopian). I just wonder whether living with this idealized vision of what the academy is supposed to be sometimes makes us forget the realities of the world it operates within.


    Scott Dolan

    May 29, 2013 at 3:26 pm

  9. I agree with the sentiment that three-article dissertations make sense for sociologists working in “article” fields, and dissertations that are written as books make sense for sociologists in “book” fields. The common feature is that the dissertation is more closely tied to early career publications. Fine.

    I’m more concerned with emergence of three-article dissertations in which one or more of the articles are co-authored with the thesis chair or other faculty member. Or at least, it seems like I’ve seen more of these recently.

    Assuming this is a real shift, it could reflect (1) changing norms about coauthorship (e.g., some thesis advisers may now expect co-authorship credit for intellectual contributions that, a generation ago, advisers gave “for free”); or (2) a real increase in thesis advisers’ contributions to students’ “independent” scholarship. Both are plausible, given changing publication incentives for faculty and the push toward shorter times to PhD. Regardless, I suspect that some of the students who go this route don’t fully appreciate how co-authored dissertation chapters can hurt them, even if their job talk is on a solo-authored chapter off the same project.



    May 29, 2013 at 4:14 pm

  10. What’s the discount rate on a graduate school co- or multiple-authored paper relative to solo? 1/3? 1/5? I know there are several vectors to consider, and it’s contingent and contextual. But I’m just saying rough n’ ready, on first read of a job file. You see that second author and your expectations drop how much immediately?


    Graham Peterson

    May 29, 2013 at 4:19 pm

  11. Sorry for the perceived aggressivity. My point was not at all conservative, neither utopian. If academia has inherently instrumental properties – and sometimes for the best – to me it doesn’t have to be all about utility. But that was how Fabio’s comment sounded like.

    In fact, as a young scholar I do think that :
    – time is necessary to achieve good research
    – space is also necessary sometimes to render complexity
    – and, more generally, having some leeway in your research (on the content as well as on the form);
    For me, that’s what a PhD is for. It is a once lifetime chance to gather all these conditions. For these reasons, I think it’s worth trying to resist the rules we consider irrelevant if we can – and who better than academics within the system can do it ?



    May 29, 2013 at 4:37 pm

  12. Fascinating thread! It ties nicely to some of the other recent discussions about choosing advisors and the (appropriate) length of a soc PhD program. A couple of thoughts are provoked. (1) Don’t fool yourself about the fuzzy boundaries between the advisor’s and the student’s intellectual contributions to the monograph dissertation being substantively different from collaboration on an article. If the student has created the monograph as an autodidact, I’d be surprised. The examining committee just pretends that the work is original to the student. (2) If you want to write the three-essay form of dissertation, DO NOT work with a junior faculty advisor. You will have competing incentives for first authorship. Choose an old advisor, who can graciously cede senior authorship to a junior collaborator.



    May 29, 2013 at 4:58 pm

  13. nlalld, I don’t disagree. But, I wonder how much this perspective can be used to rationalize away issues with potentially poor outcomes.

    High dropout rates become “they just weren’t cut out for this line of work.” A program with high average years of completion becomes “this is what is required to do good research.” The lack of a coherent, sequenced, and goal-oriented structure from the very beginning of the program is legitimized as “we’re giving you freedom to use your creativity to pursue your ideas.”

    I am just trying to point out that this view about the purpose of academia can cut both ways. So the middle ground for me was to find where program requirements and clearer objectives can help solve some potential problems with the failure to complete or the length of time to complete (that is if you see this as a problem. And maybe it’s not).

    I guess I was just saying that these instrumental outcomes are certainly a component of the equation, and some institutions/disciplines have better outcomes than others. And for me, then, it is interesting to ask, what is it that they do better in meeting these outcomes? As for the three-essay argument, I don’t know where I stand.

    With respect to what you said about resisting rules you find irrelevant, you are right, it is a matter of where you are situated within the system. But just because those rules are irrelevant to someone established within an existing order doesn’t mean they are irrelevant to an up-and-comer seeking to establish themselves. In fact, this idea seems quite “natural” to me that someone who is established would find proposed changes to the system to be potentially irrelevant. I am just asking (emphasize asking) if there is any merit to competing conceptions regarding what the order should look like.


    Scott Dolan

    May 29, 2013 at 5:28 pm

  14. Graham: I think it’s too context-dependent for even a general rule-of-thumb:
    – Who is the co-author? A current or former grad student in the same program? A post doc or junior faculty at another university? A faculty member in the students’ program but not on the committee? A committee member? The thesis chair?
    – Is the co-authored paper off the same project or a closely related project as the faculty member’s other recent work? How closely related is it to the students’ other publications, if any?
    – Does it “read” like the faculty member’s other work, or use the same theory or (unusual) method?
    – If the grad student has presented the work at conferences or written about it in a research statement, did s/he “own” the material, or did it seem like s/he didn’t really understand the project or its significance? (Hint: in a job talk, do not answer a question about your method with, “because my chair told me to do it that way.”)
    – Does the senior colleague have a reputation for putting him/herself on student’s papers at the drop of a hat (yes, there are a few out there)? Conversely, is s/he known for “giving” papers to grad students even after making substantial intellectual contributions?

    A lot of this is subjective, or based on background information that isn’t on the CV. I suspect the attribution of credit is also a place where various biases, including about gender and race/ethnicity, can creep in. But, that’s a different conversation.



    May 29, 2013 at 5:44 pm

  15. I think I’ll have to roll with you on that point, krippendorf. The major criticism of my statements of purpose in graduate applications this last year, was that it was hard to disambiguate what was my contribution and what was McCloskey’s. I had to get hammered by three different people on this point before I sucked it up and said: “that is what she did — this is what *I* did.” I would urge people to be careful about interpreting how the work “reads.” People who work really closely with an adviser will pick up their methods and writing style. I have.

    If I can add a little about why Three Essays has become popular in economics, and why that will not translate into a Three Essay world in sociology.

    (1) Axiomatization. There are few economics dissertations with huge literature reviews because much of the theoretical background is already agreed. Nobody goes back and forth in the beginning of their dissertation saying, “Card interprets the term ‘income elasticity of demand’ in X way, while on the other hand Arrow defines it as Y. Before we proceed, I will resolve this definition.” One can start an economics paper and bring the reader up to speed extremely quickly, and in fact many senior economists like Hal Varian and Ariel Rubenstein note that most lit review in economics papers amounts to meaningless nodding and totem-saluting.

    (2) Mid-level theory. There is very little Big Grand Theory left in economics, and for the most part the subfields are well established with quite clear directions for research. If I like the environment, and I like auction theory, then I write three essays on cap and trade and send one of them to the EPA and two others to liberal arts colleges at the AEAs. Straightforward stuff (at least institutionally).

    (3) Agreed methods. Nobody justifies their methods in economics. It’s rather much understood what is and is not acceptable methodologically in the field, and people who do methodology are, most of them, considered cranks unless they spend the rest of their time practicing mainstream methods (Rubenstein, Arrow, etc.)

    But sociology has no unifying foundation theoretically and there will continue to be quite a bit of Big Theory and methodological debate — even debate at the middle and applied levels. This breadth of argument may be obnoxious at times, and even unproductive, but it is on balance I think a force for creative good.

    I hope sociology becomes more open to Three Essays dissertations, but arguing that it ought to become the standard ignores the institutional, theoretical, and methodological differences between economics and sociology which created the variation in the first place. The world doesn’t need more Three Meaningless Regressions About Minutiae That Have Already Been Done To Death, dissertations.


    Graham Peterson

    May 29, 2013 at 6:21 pm

  16. Scott: Then we agree with each other. I am not saying that 3 essays-dissertation are necessarily bad research and book-length dissertations good research. Actually my own dissertation is a hybrid.
    As I said, I am a young scholar, who just starts being confronted to the job market, not an incumbent resisting challengers’ disrupting institutional work [footnote]. If I chose to do research, it was certainly not to make a career, or to make money (my pizzaiolo makes twice as much as I do as a full time lecturer), but to try to do good research (the kind of research that can sometimes be wonderfully expressed through a paper, but not only).
    I try to learn the rules because I am well aware it might be useful, but knowing the rules doesn’t mean I have to take them for granted. I feel myself lucky that I am quite confident in the fact that I will find an academic position in the end. If it must not be in the most prestigious places, then it won’t. But for now, I’ll keep trying. I don’t see why people that I consider brilliant scholars would not be able to see potential – if there is one – beyond results.

    Footnote: Actually, incumbents of the academy are to me the ones, sometimes unconsciously, that implemented the publish or perish rule that led to increased standardization in research – at the expense of some diversity and richness – not the ones who resisted them. In fact, paradoxically it values them. As a colleague said to me yesterday about the director of the university were we made our PhD (which claims to be a leader in Europe): “He created a world were he himself couldn’t survive”.



    May 29, 2013 at 7:06 pm

  17. There’s sometimes a funny narrowness to these threads. Looking outside of the narrow band of job candidates who are viable the top 25 PhD programs, there seems to be a lot more variation in a successful publication record. If you move into the R2 and SLAC world, there’s even more. And then there’s some programs, like Cornell and Stanford, which are well known for not being impressed by a CV as much as the quality of the writing samples submitted.



    May 29, 2013 at 7:29 pm

  18. I’ll be done after this. And apologies in advance for long-windedness.

    nllad, we might be speaking past each other a bit. I think it is partially my fault as I shifted the discussion away from the relationship between three essay format and academic placement/subsequent publication records to a discussion of the relationship between Ph.D. program structure and program outcomes. And please don’t take any of this as aggressive, but rather just me throwing out some ideas, which might be different than your own. All in good fun (hopefully).

    I agree with you when you say we cannot/should not dictate a specific format for a dissertation. And I agree with what many others said, the format a dissertation takes is contingent upon many factors (disciplinary assumptions, theoretical approach, research question, the kind of data you want to collect to answer the question, and others). However, Fabio’s suggestion wasn’t to dictate the three essay format, but rather to use it as a baseline or starting point. And I think the reasoning was to make a relatively open-ended and sometimes hazily fleshed out task (the writing of a dissertation) more manageable and concrete.** Which I think is commendable given some of the numbers that we see in terms of the retention rates, average number of years to complete, and placement of Ph.D.’s in some disciplines/programs.

    And I think Fabio was also arguing that the three essay format might be a good option given the reward system currently in place (the necessity of journal publications) to land an academic job at an R1 type University—which I think was echoed by Brayden in his comment. I think cwalken rightfully took issue with Fabio’s narrow conception of who we are training as Ph.D students, and the reward systems that are currently in place–even within what we would call R1 schools. Though I think Fabio has been pretty clear in previous posts he suggests a Ph.D. track only to individuals who want to land academic jobs, but maybe sometimes forgets that there are many more jobs outside of the R1 arena (if I am wrong on any of this, please correct me).

    As a bit of an aside, I think it is important to ask some questions given your depiction of academia.

    If we operate with the conception of academia you laid out in a previous comment—a once in a lifetime opportunity to have the time to pursue good research, space to render complexity, leeway/independence to pursue your ideas—what are some issues that might arise from such a conception? I will just raise three. And admittedly these might not be issues to you, which is perfectly fine.

    First, I can see how this conception can create an exclusive institution. Think of all the people who don’t have the time, space, independence, and/or external support necessary in order to pursue such an endeavor. No wonder diversity in academia remains an issue. In the United States, this might be a bigger concern given the changing demographics of the student population over the next few decades.

    Second, how much time is an appropriate amount of time to devote to “good” research? I think this question is fundamentally instrumental. For example, if it takes an average of 9 years to complete a Ph.D. in Sociology, what effects does this have on the individuals who complete a Ph.D. in Sociology? Especially given the current changes in the environment: cost increases, decreases in funding lines, and shifts towards more non-tenured positions.

    Third, what the heck is meant by “good” research and who gets to define “good” research? For you, it comes down to the brilliant professors with whom you interact, and certainly it’s hard to argue with this fact. Agreed, any field differentiates between what is “good” and what is “not good enough.” And you are right to point out that it is important to be aware of the currently operating rules so that you can navigate them and be successful (I don’t deny the importance of individual skill). Too often, though, I think the process of determining whether a dissertation is “good enough” becomes a somewhat arbitrary negotiation process between a graduate student and their advisers, one that is inherently asymmetrical in terms of power. I was blessed to have incredibly supportive advisers (with whom I remain colleagues and friends today), but not everyone is as lucky as I was (or apparently as you were). I think the relatively un-standardized nature of this negotiation process (with obvious variations across disciplines and programs) has something to do with the amount of time it takes to complete the degree, where the ultimate evidence of goodness becomes a passable dissertation according to the opinions of a committee of three or four individuals.

    **I know that research is inherently complex and that there isn’t–nor should there be–a one-size fits all approach. I am not arguing that there should be universal standards. I am just asking, if 9 years on average is what it should take? And if not, what can we do to get the number down? I am asking this when most disciplines/programs set up their programs to be quicker than that. I’ve never read a graduate handbook that lays out the program and says it should take an average of 9 years. So there seems to be a disconnect between what some disciplines/programs set out to do and the results. And some of it is related to implementation of the program. And certainly there is no denying individual factors and others (life happens).


    Scott Dolan

    May 30, 2013 at 6:42 pm

  19. As someone just about to turn in a three-essays dissertation — but given a choice in the matter — I think the issue is a bit more complicated. I think that if you’re going into a field, subfield, or discipline where you will be evaluated on and get or not get tenure based on books, you should write a monograph for your dissertation. If you’re going to be evaluated by publications in journals, go for essays. In each case, the choice should be motivated by what is going to be of more direct help to you in your early career.

    With a background in sociology, I could have gone either way. That said, because my work is overwhelming more quantitative and because there is not a strong market for very quantitative sociology books, I figured that a book was probably not in my very near future and decided that going with essays made more sense. Since economics is a very quantitative discipline, I’m not at all suprised that going with essays would tend to be a wiser move for economists who overwhelming forgoe books in their early career.

    Side Point: As at least one other person pointed out, AER Papers and Proceeedings is not the AER. Having such a similar name for such a different publication with different standards is a pretty confusing them for them to do. That said, they are different publications with very different standards.


    Benjamin Mako Hill

    May 30, 2013 at 7:17 pm

  20. […] les thèses de plus de 350 pages ! première partie et deuxième partie sur […]


Comments are closed.

%d bloggers like this: