Not to kick a dead horse down the stairs, but I thought I would do an HPE version of the recent XKCD comic “12 papers” spinoffs (with help from co-editors Scott Gehlbach, Jared Rubin, and Pavi Suryanarayan). To be clear, I’m poking fun, and there will be no paper shaming in this post. But these Twitter shenanigans relate to a larger topic, which is the strategy for writing HPE papers.
When I was in graduate school at Columbia, I took an excellent class from Professor Tim Frye on “Comparing Institutions.” And I remember the syllabus from this course — in addition to listing the required readings, it had a section that was amazingly entitled “strategies for writing papers and becoming famous”. There was a list of strategies that I have continued to consult over the years, with corresponding academic examples (that I’ll omit here); but today, I have taken some liberties, paraphrased, and listed it here:
New Data for an Old Problem
New Theory for an Old Problem
New Method for an Old Problem
Take on a Big Fish (This was Skocpol’s critique of Barrington Moore, for the record)
Attack from Below – Give Microfoundations to Macroarguments
New “Old” Data for an old Problem
Use a Formal Model to Get Counterintuitive Prediction
Show the Received History is Wrong
Apply an Outside Theory to a Political Question
No One Has Studied X
More generally, the idea of how to find research ideas is not new — every political science PhD student should read KKV’s Designing Social Inquiry, and I found a recent article by Gustafsson and Hagström (2018) helpful in discussing how articles are often judged by three criteria: filling a gap, addressing a real-word problem, and/or being methodologically rigorous.
While Tim’s advice was designed for comparative politics and more generally political science students, I think it’s interesting to think about how these apply to historical work.
Clearly HPE scholars always use “New Old Data” for an old problem — advances in digitization means we can access more data from across the globe, and the field of HPE provides resources and scholars who excel at this type of work. Though the trick here is justifying why and how this new data tells us something that we didn’t know before; otherwise it’s a replication exercise using a historical case (which might have value, but is a different approach).
We also often “Attack from Below” by using micro-level data to explore institutions and behavior. This could be fine-grained administrative data or primary source records, but since collecting historical data is a massive investment anyway, we can often take the opportunity to collect more detailed measures. And thanks to the rise of design-based inference and natural experiments, there are plenty of “New Methods” to apply to old data.
Sometimes HPE scholars even use “New Old Data” to “Attack from Below” to show the “Received History is Wrong”. There’s a great new working paper I saw at a workshop recently that does this — Jack Paine, Xiaoyan Qiu and Joan Ricart-Huguet use new historical and spatial data on precolonial borders in Africa, to show that (contrary to the existing wisdom) colonial powers followed natural borders. (This is a work in progress, but keep a weather eye on the horizon for this paper!)
But back to the XKCD-inspired HPE comic – it’s not enough to just collect historical data, any strategy that involves “everyone should love my historical case” needs to be thoroughly justified. And let’s consider the last item in the list — “No One Has Studied X.” This strategy is more difficult to pull off, especially with historical cases, but that doesn’t prevent folks from trying — and one of the most frustrating occurrences in an interdisciplinary area like HPE is getting papers to review from scholars who clearly haven’t done a literature review across fields.
How does one prevent this? Luckily, there is a new tool to help us find networks in research! Cheers to Dan Smith for this tip, whose post I saw on Twitter. I’m fairly certain everyone should try https://www.connectedpapers.com. Here I’ve plotted one of my/coauthor’s papers, which uses 19th century French data to speak to a longstanding literature on the effects of committee service. Sure enough, we link with American Politics papers, but also branch out to link with related historical work (19th century Austria, for example).
The Connected Papers network graph is be particularly useful if your work crosses topics or fields — there might be a cluster of scholars previously unknown to you, but essential for the ultimate research project. This also prevents you from falling into the biggest hole there is — writing “no one has studied X” when they have.
Now if you’ll excuse me, I’m going to go write papers and become famous.