By Anne M. Burton, Sumedha Gupta, and Barton Willage
What makes a good conference abstract? In anticipation of the ASHEcon submission deadline coming up this fall, we compiled our thoughts and advice on the answer to this question. Sumedha Gupta is an Associate Professor of Economics at Indiana University Purdue University Indianapolis (IUPUI) and one of two new ASHEcon newsletter coeditors. Barton Willage is an Assistant Professor of Economics at University of Colorado—Denver, a Faculty Research Fellow at the National Bureau of Economic Research and the other new ASHEcon newsletter coeditor. Anne M. Burton is an Assistant Professor of Economics at the University of Texas at Dallas and the editor of the ASHEcon newsletter. Between the three of us, we have served on the ASHEcon Scientific Review Committee (the group that reads and evaluates every abstract submitted to ASHEcon) for several years, served as Program Chairs for the 2023 ASHEcon conference, and served as a grant reviewer for the NSF and RWJF. We have also collectively submitted numerous abstracts to conferences over the years with varying degrees of success.
What are the key components of an abstract or extended abstract and what is the difference between the two?
AB: There are 5 key components to an abstract/extended abstract for a conference submission for an applied micro/health policy project: 1) research question, 2) motivation/framing, 3) description of data and methods, 4) results (if ready; can be preliminary), and 5) policy implications. The first and second components can be done in either order, but they are both necessary: clearly and concisely state your research question near the beginning of the abstract and motivate why it is an interesting and important question. Then go into how you are answering the research question: what data and methods are you using? If you have results, provide an overview of what they are. And don’t forget a conclusion/policy implications!
The main differences between an abstract and an extended abstract are the word count and expected level of detail. Abstracts are typically 100-250 words and similar to the level of detail you’d find in the abstract of a published paper. For an extended abstract, sometimes conferences will specify a word limit (500, 800, etc.). That doesn’t mean you need to be exactly at the word limit, but you should take the word limit as a signal for the expected length and level of detail. Go into more detail on the data sources/key variables, the method(s) you’re using and key assumptions of those methods/brief mentions of ways you test whether those assumptions are likely to be satisfied, and some more detail on the results.
SG: I think the key components of an abstract or extended abstract include – a brief description of a problem and why it is important (e.g., how many lives it may impact, or costs it may impose on individuals or society), the gap in the literature/understanding that prevents a full or efficient dissolution of the problem, data used, methods, results, discussion, and conclusion. However, economics journal abstracts are generally shorter (sometimes just ≤100 words) and focus much more on data and identification strategy used for causal inference, whereas extended abstracts may be longer with separate headings for “Importance”, “Objective”, “Design, Setting and Participants”, “Exposures”, “Outcome Measures”, “Results” and “Conclusion and Relevance”, quite like the abstracts of medical/health services research journal articles but with additional methodological description (e.g., causal inference design details). Since extended abstracts force the authors to pull together key study information in a concise and clear way, they may be helpful to conference reviewers to make informed selections. For the same reasons, in early stages of manuscript development, extended abstracts may also serve as helpful roadmaps, even for articles being written mostly for an economics audience with eventually much shorter abstracts.
Of course, it is always important to confirm if the conference you’re submitting to has any prescribed guidelines for abstract format or length as non-adherence may delay or exclude the acceptance of your submission. But generally, economics conferences are more flexible.
BW: The key components for a conference abstract are similar to a paper abstract, with some important differences. You need to state the research question and explain why it is important, clearly explain how you answer that question (method), briefly cover the data sources and results with explicitly stated outcomes, and state the key takeaways and possible policy implications.
For ASHEcon, one difference from a paper abstract is length. Submission abstracts can be up to 500 words, so if you are just copying and pasting the abstract from your paper, you’re missing an opportunity to give conference reviewers as much information as possible. Other conferences have even higher limits: AEFP’s (Association for Education Finance and Policy) “extended abstract” can be up to 1,000 words.
While there is likely no bonus for using the maximum number of words nor is there a minimum word requirement (this is not a 10th grade writing assignment), using a small fraction of the allotted space might let you know that you are not providing enough detail, or you might inadvertently indicate to reviewers that you did not take the submission seriously.
There is a third category of submissions required for some conferences, and that is the full paper. I would not recommend sending anything less than a paper you would share with others for feedback, because I think reviewers can tell the difference between an actual paper draft and something hastily put together for a conference submission. However, the draft need not be so far along as to be ready for a journal submission.
What is a good way to motivate your project?
SG: You must describe clearly to the reader the problem and convince them why they should be worried about it and interested in it too. Description of the problem and its ‘significance’ is typically followed by what we have learnt about the problem (and any policy/other solutions) from prior literature, and what critical gaps in understanding remain without which the problem may not be fully understood or satisfactorily resolved. Thereafter, the key “Objective” of the study is to fill this ‘important gap’ in the literature.
The motivation of the study is crucial as it sets the stage for the entire paper, and thus it is important that this section be as methodical and compelling as possible to pique the interest of readers, to ensure that they read on.
AB: Why are you working on this project? What drew you to this question in the first place? By answering your research question, what do we learn about the world, about human behavior, about basic economic concepts, about the intended or unintended consequences of a specific policy or regulation? The answers to questions like that are all good ways to motivate your project.
Do you tailor the framing/motivation for specific conferences, and if so, how? e.g., if you submit the same project to ASHEcon and AEAs/ASSAs, is your abstract the same or different? What about ASHEcon and AcademyHealth?
BW: I think it makes total sense to submit the same project to multiple conferences, but make sure the project is a good fit for the conference. Almost all projects are a good fit for a general interest conference such as AEA/ASSA and regional economic associations’ conferences (e.g., Southern Economic Association). However, many projects will not be a good fit for both ASHEcon and APPAM or ASHEcon and AEFP. I have not traditionally changed the framing of my abstract for specific conferences, but I’ve also been rejected from my fair share of conferences.
The main thing I want to discuss here is regarding AEAs/ASSAs. The conference that we all know as the AEAs actually allocates several sessions to other organizations (collectively known as ASSA). ASHEcon has one session that it creates at the AEAs/ASSAs. iHEA and HERO also organize health-related sessions for this conference. Keep an eye out for the calls from these organizations’ ASSA sessions, which might be relatively more interested in health economics research than the general AEA call for papers.
AB: I don’t usually tailor the framing/motivation for specific conferences but it’s not a bad idea, so I’ll probably try this in the future. Think about who the audience is for a specific conference. e.g., ASHEcon’s audience is comprised primarily of health economists, health policy researchers, and health services researchers. So they are going to have a general understanding of popular topics like Medicaid, Medicare, opioids, etc. and you don’t need to go into a lot of background detail in an abstract. For a conference like AEAs, however, the audience is all sorts of economists, so it may be helpful to provide a bit more context or explain how your research question and results are interesting to a broader audience.
SG: I typically submit to the ASHEcon/iHEA/HERO sessions in the AEAs/ASSAs so my audience remains other health economists, and I wouldn’t necessarily change my abstract. However, the timing of the call for submissions is somewhat different for these conferences, and if the analyses has come further along, an upcoming conference submission deadline is a great opportunity to update the manuscript draft prior to submission. However, the ASHEcon and AcademyHealth audiences may differ, the latter being more multidisciplinary; a more health policy or descriptive paper, likely to be submitted later to a health services research or medical journal, may be of greater interest to the multidisciplinary AcademyHealth audience.
How do you clearly and concisely describe your data/methods?
SG: In addition to listing the data sources and observation windows, walking the reader through the construction of the key outcome and independent variables of interest is helpful. When describing methods, along with describing the research design, and providing the exact regression set up, I find it useful to lay out for the reader the assumptions underlying the identification strategy, and the types of threats to identification. By doing so, it becomes clearer if the threats to identification are severe or in fact unlikely. It also sets the stage for subsequent sensitivity analyses to test for violations and/or robustness of results.
AB: Data: provide the name of the dataset. Highlight the key variables you are using in your analysis (outcomes, treatment, essential controls).
Method: if you are doing reduced-form work you are likely using one of the following 4 methods: 1) difference-in-differences, 2) instrumental variables, 3) regression discontinuity, 4) randomized controlled trial (lab or field experiment). Explicitly state which method you are or will be using. If you have the space, briefly note the key assumptions (e.g., for a difference-in-differences project, parallel trends, no concurrent shocks conditional on the controls, and a constant treatment effect). Mention how you test whether these assumptions are likely to be satisfied.
The same general principles probably apply with qualitative research, structural estimation, or theoretical models. You could look at the abstracts or introductions of published papers or projects on a conference program to see how others describe these methods.
Do you need results to submit an abstract? Does this depend on your career stage (e.g., Ph.D. student, tenured professor)?
SG: An extended abstract with some results, even if just descriptive, may improve the chances of acceptance for conference presentation for at least 2 reasons – First, a study with some results is evidence of the study’s feasibility, which may reduce the risk that authors may later decline to present in case the study did not work out for any reason. Second, an extended abstract with some (preliminary) results is more likely to pique the reviewer’s interest. But if the data is in hand and results are in the works, I think it does not hurt to be aspirational and still submit an abstract to the conference, because a conference can serve as a natural deadline to accelerate research productivity.
AB: You don’t necessarily need results, but if you are an early-career researcher, haven’t presented at many conferences, or don’t have a good idea of how long it takes to go from acquiring the data to having conference-ready results, I’d recommend at least having some preliminary results. Run some regressions and check whether you have a sufficient sample size, whether the variables you have are measuring what you thought they were measuring, etc. Having preliminary results can reduce potential stress around presenting at a conference. Speaking from experience, you can have a good research question and what you thought was an appropriate dataset to answer said research question, but that turns out not to be the case, and if you’re finding that out 3 weeks before a conference, that’s not a fun place to be in!
BW: The need for results depends on the conference. For instance, both the CFPB Research Conference and the FDIC Research Symposium require at least preliminary results. However, ASHEcon and AEFP are more open to forthcoming results; in the latter cases, let the reviewers know that you have access to data and a clear plan for analysis. It is okay if your results or interpretation change before the actual conference; it is very acceptable for results to be preliminary for almost any conference. I do not think career stage matters much here. Reviewers and program chairs are working with many abstract submissions, and they are unlikely to know the career stage of most submitters. Additionally, based on my experience, there is relatively little leeway and subjectivity in the scoring of abstract submissions compared to journal submissions. Either results are required or not; and there are results in the abstract or not.
How do you write a conclusion with policy implications if you don’t have results yet?
BW: There are two options that I see here. First, you can say what the implications are for a positive, negative, and null finding separately. Each of these likely has their own implications, even if we don’t know what the sign of the estimate is going to be yet. Second, just broadly discuss the area of policy that the empirical finding is related to. For instance, if you are studying the impact of cigarette taxes on smoking, this has implications for tax revenue and how policymakers work toward reducing smoking regardless of if there is a statistically and economically meaningful effect or not.
SG: It is more challenging to write conclusion/policy implications without any results in hand, of course. But, in such instances reiterating the hypothesis being tested and discussing the implications of both significant or null findings and reminding the reader of the relevance of the research question in either case is helpful. In some ways, submitting an extended abstract without results, is quite like submitting a grant proposal, where all the other sections – hypothesis, data, research design and policy implications – besides the results, are clearly laid out and well developed, in an effort to convince the reviewers that the research question is important enough and the proposed data/methods are sound enough that irrespective of the outcome of the analyses the findings would be of interest.
AB: Echoing what Barton and Sumedha said: it’s important to show the reviewers you’ve thought about what the policy implications could be.
Anything else to consider?
BW: One other consideration for getting conference submissions accepted is individual abstracts vs. organized sessions. Every year, I receive a few emails about how the acceptance rate of organized sessions is higher than that of individual submissions. Part of this correlation is endogenous; those who can create organized sessions have networks of researchers in the same research areas and are more experienced. However, one benefit of organized sessions is that conference organizers do not have to find other accepted individual abstracts to combine into a session. Sessions typically have a common theme, and there might not be sufficient accepted abstracts in the same area. So, feel free to reach out to people you might not know to see if they would like to join a session you are organizing. They can be at any stage of their career, and it’s a great networking opportunity.
Don’t take it too hard if your submission isn’t accepted; it doesn’t necessarily reflect the quality of the research. I know of at least a few instances where papers were rejected from conferences but were accepted at very highly ranked journals (e.g., AEJ or REStat). Just keep on submitting.
How does ASHEcon evaluate abstracts?
Here is the rubric that the ASHEcon Scientific Review Committee uses each year to evaluate individual abstracts submitted for the annual conference:
Criterion | 5 | 4 | 3 | 2 | 1 |
What best categorizes the purpose and objectives of the research? | The main purpose of the research is clear and concise | The main purpose of the research is identified but not clearly stated | The main purpose of the research is not well defined and appears to be incomplete and/or unfocused | The main purpose of the research is not supported by the findings | The main purpose of the research is absent |
What best categorizes the methods described in the abstract? | The methods are clear and concise | The methods are almost complete and seem valid | The methods are only partially described or valid | The methods are not well described, or not very appropriate | The methods are not described or completely flawed |
What best categorizes the research outcomes (preliminary results, or a plan for collecting results, are acceptable)? | The results or a plan for collecting results is well balanced and described | The results or a plan for collecting results is almost complete, with supportive quantitative or qualitative data | The results or a plan for collecting results is only partially described and provides limited quantitative or qualitative data | The results or a plan for collecting results is barely described and there is no quantitative or qualitative data to support the findings | The results or a plan for collecting results is not described or completely flawed |
What best categorizes the conclusion or implications? | The conclusion or implications are discussed and clearly outlined | The conclusion or implications are valid but need more nuance | The conclusion or implications partially reflects the results | The conclusion or implications do not reflect the results | No clear conclusion is drawn |