The central point of Evan Lieberman’s essay is that the discipline of political science should recognize more explicitly that any given piece of research might contribute only to a single stage of the research cycle (description/identification of correlates/identification of causes). In particular, Lieberman wants work falling in the early stages of that cycle to be accorded more respect, especially by journal editors and reviewers. He notes that highly regarded biomedical journals such as the New England Journal of Medicine routinely publish descriptive work but that, from his perspective, most leading political science journals are fixated on random controlled experiments that make it possible to identify causes. Lieberman asserts generally that appreciating the way other disciplines operate would benefit political science and specifically that more journal space should be allotted to description, rumination, and suggestive case studies.

I agree with much of what Lieberman has to say…with one important exception. He makes reference to the fact that descriptive analysis can either be “an end in itself” or a gateway to research at later stages of the cycle. His entire essay, however, is predicated on the value of the research cycle. If descriptive analysis is an end in itself, that means it need not generate testable hypotheses for later associational and causal analyses; yet, none of his examples of useful, descriptive biomedical research was an “end in itself.” Instead, they involved the identification of new symptoms or conditions that would then be inspected for possible causes and cures. It seems to me that promoting description as an end in itself is completely at odds with the rest of his essay and would reduce political scientists to little more than “slow journalists.” I am all for description, but only when it is part of the research cycle that Lieberman so skillfully describes.

Lieberman focuses only on the biomedical model and this may be too limiting. In the spirit of “appreciating the way other disciplines operate,” let me suggest another model drawn not from biomedical journals but rather from journals with which I am more familiar. Whereas Lieberman encourages each political science journal to be eclectic and to publish contributions at various stages of the research cycle, there could be some advantage to a division of labor across journals in the fashion of general science journals. Science, Nature, and PNAS may not publish only “causal” research but they do limit articles to just a few pages in length, meaning that exclusively descriptive pieces are a poor fit and quite rare. Instead, research that sets the stage for later analyses by, as Lieberman puts it, detailing “what has come previously within that substantive area” is typically directed toward journals explicitly designed for such purposes. The clearest examples of these are the “Nature Reviews …” journals such as Nature Reviews Neuroscience and Nature Reviews Genetics. These journals are widely-cited (impact factors are often well into the 30s) and contain superb literature reviews and theoretical discussions but no new empirical analyses. A still different style of journal but one that would be valuable in political science is illustrated by Behavioral and Brain Sciences. This journal publishes lengthy, synthetic target essays that are then followed by carefully reviewed responses or critiques from dozens of diverse research teams. Perhaps this type of exchange suggests a distinct and final stage of the research cycle in which numerous causal (and other) studies are integrated and interpreted in order to identify useful large themes.

Stepping back for the big picture, I see three options for dealing with the problems emanating from the existence of research at distinct stages of the research cycle. The first is the one political science journals seem to have adopted and that is requiring each article to do everything: start off with an extended statement of the underlying theoretical question, review the literature on the topic, generate specific hypotheses from the preceding theory and literature review, perform correlational analyses bearing on the hypotheses, and then conclude with controlled experiments addressing possible causes. The second is the one Lieberman prefers and that is to have each and every journal publish research that falls at any stage of the research cycle, with no expectation that multiple stages would be addressed by any particular work. Thus, a given issue of the APSR might contain an article describing in general terms the societal forces relevant to the rise of a candidate like Donald Trump but also an article that reports the results of a randomized controlled experiment testing whether support for someone like Trump can be induced by certain manipulations. And the third is the option prevalent in the natural sciences and that is assigning specific journals to address particular stages of the research cycle. To illustrate the segmentation that characterizes many science journals, when my colleagues and I once submitted an article to a leading general science journal the reviewers unanimously concluded that a section describing the theory behind our hypotheses did not belong in a “results-oriented” journal. One reviewer even expressed discomfort with his/her sense that “the authors of this paper seem to know what they expect to find before they conducted the analyses.” Findings are typically not expected to be accompanied by theory in these journals.

I do not have particularly strong feelings as to which of these three is preferable for political science though I might lean toward the general science model in which journals have distinct missions. (As Lieberman notes, it could be argued that Perspectives on Politics constitutes an initial step in this direction.) I agree with Lieberman that the “do it all in one article” approach is problematic. Forcing authors to include a theoretical discussion that is often transparently post hoc is as silly as forcing authors to test rigorously every single idea that flows from their theoretical presentation. On the other hand, I worry that the biomedical solution proposed by Lieberman runs the risk of turning political science journals into overly compartmentalized hodge-podges.