8.7 Article Lengths

This isn’t specifically to do with the model I built, but it’s an interesting finding. Articles have been getting longer, much longer, over time. Some of this is to do with the journals getting rid of things like abstracts of APA or PSA papers. But it’s also just a fact that articles have been getting longer. Here is a helpful graph that shows the patterns.

A scatterplot (with trend lines) showing the number of pages of articles from roughly 1880 to 2000, sorted into five deciles (10 percent, 30 percent, 50 percent, 70 percent, and 90 percent). Across all categories, the median number of pages has a peak in the early 1900s, drops until the middle of the centrury, and begins to rise sharply around 1980.

Figure 8.7: Measures of article length in each year.

For each year, I’ve sorted the articles by length, then plotted the lengths of the articles at five decile markers. The red curve is the length of the article that is 10 percent of the way up the length table, the olive line is the article that is 30 percent of the way up the length chart, the green line is the length of the median article (by length), and so on.

I’m using medians rather than means because the outliers here are really significant. I don’t want the numbers to be thrown off by the fact that a journal publishes a single ninety-page article. But I also want to be able to see on the graph how much impact the one-page articles are having.

The latter turns out not to be too significant. Even when the articles are at their shortest in the early 1960s, the olive line only gets down to five pages. So even then, 70 percent or so of the articles are five or more pages. There are more abstracts and discussion notes being posted then than there are now, but not enough to explain all of what’s happening. The red line creeps up very slowly as first the regular journals start abolishing short articles and then Analysis starts increasing its average page length as well.

But here’s the really striking feature of the graph: the median article in the 2010s is as long as the ninetieth percentile article from the 1950s and 1960s. For a while there, articles over twenty pages were real outliers. Now they are the norm. The outliers are now over thirty-five pages. This feels like a bad thing; articles are getting bloated, and we need to find a way to get them back to a reasonable length.