My colleague Jim Moody sent along some interesting data this morning. Using the Web of Science database, he took the most-cited papers in Sociology and produced a Top 10 list for each decade going back to the 1950s. Not a table of which papers were most popular in those decades, but a table of which papers are now the most-cited from those decades. Note that the 1950s category is really “1950s and before”. The universe of citing papers is all of WoS. I absolve Jim of any responsibility for the figure I made from his data, or my comments below. Here’s a dotplot of the results:
(You can get a PDF version of this figure if you want a closer look.)
Again, we’re looking at the Top 10 most-cited articles that were published in the 1950s, 1960s, and so on. This means, first, while the eleventh most-cited article from the 1980s might outscore the fourth most-cited from the 1950s in terms of cumulative citations, the former does not appear here whereas the latter does. Second, books are not included, even though it’s clear from other work that the most important books were and remain central to the field. Bearing those caveats in mind, there are some striking patterns. One thing to notice is the rise of articles from the Annual Review of Sociology in the 2000s. Another is the increasing heterogeneity of outlets. Of the top ten papers written in the 1950s or before, seven appear in the American Sociological Review, two in the American Journal of Sociology, and one in Social Forces. (That is SF’s only entry in the list, as it happens.) ASR and AJS rule the 1960s, too. After that, though, there’s more variety. Strikingly, for the 2000s only one of the ten most-cited articles is from ASR and none is from AJS—a complete reversal of the pattern of the ’50s and ’60s. You can also see the long shadow of post-war university expansion and “Boomer Sociology”. The most-cited work from before 1970 is not nearly as widely cited as the most-cited work from the ’70s and ’80s, despite having been around longer. The drop-off in citation numbers in the Top 10s from the ’90s and ’00s is to be expected as those papers are younger. American dominance—or insularity—is also evident, as the only non-U.S. journal to make any of the lists is Sociology, and that was in the 1970s.
Turning to the subject matter of the papers, I think you can see the importance of articles whose main contribution is either a methodological technique or a big idea. There are fewer papers where a specific empirical finding is the main contribution. If you want to hang in there as one of the most-remembered papers from your decade, it seems, give people a good concept to work with or a powerful tool to use. Of course, it’s also true that people tend to have a lot of unread books lying around the house and unused drill attachments in the garage.
It is tempting to connect these two patterns in the data. To speculate: ASR and AJS remain amongst the journals with the very highest impact factors in the discipline. Publishing in them has become more important than ever to people’s careers. Yet the most-cited papers of the last two decades appeared elsewhere. These journals demand the papers they publish meet high standards in methods and ideally also innovate theoretically, along with making an empirical contribution to knowledge. That, together with a more competitive and professionalized labor market, produces very high-quality papers. But perhaps it also makes these journals less likely than in the past to publish purely technical or purely theoretical pieces, even though some papers of that sort will in the end have the most influence on the field.
Outlets like Sociological Methods and Research and Sociological Methodology now publish articles that might in the past have appeared in more general journals. Similarly, big-idea pieces that might once have gotten in at ASR or AJS may now be more likely to find a home at places like Theory and Society or Gender and Society. At the same time—perhaps because the state of theory in the field is more confused than that of methods—theoretical papers may also have been partially displaced by ARS articles that make an argument for some idea or approach, but under the shield of a topical empirical literature review. In a relatively fragmented field, it’s also easier for methodological papers to be more widely cited across a range of substantive areas than it is for a theory paper to do the same.
One would need rather more data than the figure above to confirm that sort of conjecture, but I think it’s at least plausible.