Response to criticism on the death of academic -isms

Share on FacebookShare on Google+Email this to someoneTweet about this on Twitter

My first post detailed the demise of wooly-headed theories in academia. In this post, I’ll also address some common criticisms that have come up so far. In the third post, just above this one, I will look at a rival class of theories, namely the scientific and in particular biological approaches to studying humanity. The take-home message is that, while the Blank Slate theories are slowly being driven out of academia, new ones based on the biological sciences are becoming ever more popular. But let’s start with the criticisms:

1) You’re confusing popularity with accuracy, truth, etc.

I never said anything to this effect. I am just interested in whether certain theories are becoming more or less prevalent. Now, I happen to believe that in the case of, say, psychoanalysis or Marxism, the theories are becoming less popular because people realize that they’re not very insightful. And certainly what I think is a great theory could become unfashionable for whatever reason. Whether you’re celebrating or mourning the death of some theory, I don’t care — I just want to show whether it is or is not dying.

2) You didn’t account for the lag between when an article is published and when it is archived in JSTOR.

I did do that, but I was only explicit about it in the comments to the first post. Journals in JSTOR have a “moving wall” between original and archived dates, with most having a lag of 3 to 5 years. Here is the distribution of lag times. By excluding data from 2003 onward, I’ve taken care of 88% of journals. And I don’t want to hear a non-quantitative objection that “the remainder could be affecting the results” — tell me what you think the data-point should be for, say 2001, and then derive how large of an effect the 12% of journals would have to have in order to get that value. We’ll see how reasonable that sounds. Moreover, since no moving wall is greater than 10 years, any decline that started before 1998 is not subject to even this vague objection — for example, Marxism, feminism, and psychoanalysis.

3) You don’t have a neutral control case to show that Marxism is “really” decreasing in popularity.

I did admit in the first post that ideally we’d have the total number of articles that JSTOR has for a given year, and that we’d divide the number of articles with Marxism by the total number of articles to get a frequency or prevalence. We can estimate the total by searching for articles with some highly frequent word, such as “the”, so that the number returned is very close to the total. For “the”, this approach is almost guaranteed to work, since almost no article would slip through the net.

However, JSTOR has a list of highly frequent words that it doesn’t allow. Still, not all common words are blocked. I consulted a frequency list compiled by Oxford Online, and chose the highest-ranking words there which are not blocked by JSTOR, though I excluded the personal pronouns and “people,” since I don’t expect those to show up much in hard science or social science journals. This gives the variants of “time,” “know,” “good,” and “look.”

So, I’ve estimated the total number of articles for a year by searching for “time” OR “know*” OR “good*” OR “look*”, where the asterisk means the word-ending can vary. How closely this estimates the true total is not of interest — the point is that it serves as a common, neutral yardstick to measure the change from one year to the next.

Interestingly, using this control has almost no effect on the shape of the graphs from the first post. That is because the increase in the total number of articles increases only linearly from about 1940 onward, whereas the articles on postmodernism increase or decrease exponentially — and an exponential divided by a linear is still growing or dying very fast. I’ve redrawn the original graphs and posted them here because it’s easier for me; sometime soon, I’ll substitute them into the first post for the record.

The only change I make to my original observations is that social constructionism is not so obviously declining anymore, although it is plateauing and apparently declining since 1998. If I had to guess about its behavior after 2002, I would say it’s downward simply because none of the other theories plateaued for very long — they quickly hit a peak and declined, so a steady high value does not appear to be stable for such theories.

4) You’re mistaking a decline in usage with a decline in belief — once the idea becomes taken for granted, practitioners stop referring to it explicitly.

Just on an intuitive level, we know this is horseshit — do physicists not use the words “gravity” or “electricity” anymore, or no longer refer to Newton? This objection exemplifies the problem with the average arts and humanities major: he is content to build a logically coherent argument without doing a quick reality check for its explanatory plausibility. I guess that’s why they end up in law firms.

But to provide evidence that usage tracks belief, here are some graphs for hard science keywords. In the case of Darwin, I excluded articles on “social Darwinism,” which appears to be a, er, social construction in academia. See here. I have data on academic scarewords like “biological determinism,” and perhaps in a future post I’ll show those. Right now, I want to focus on articles that are at least somewhat level-headed. For ease of inspection, I’ve given each graph a simple title, and list the search terms at the end of this post in an Appendix.

As the disciplines of population genetics and sociobiology have become staples of biology, mentioning them by name has not declined — just the opposite. Because they are such thriving fields, writing about them explicitly has shot up. Darwin’s thoughts were immensely popular in Victorian times, but they languished because no one could tell how to unite them with the study of heredity. That was, until the modern evolutionary synthesis, which began in the late 1930s — since then, interest has exploded. The same goes for Mendel’s thoughts — no one knew what the physical basis for his “gene” idea was, until the relationship between the genetic code and DNA was laid out in the late 1950s.

This shows that even hard science ideas can rise and fall and rise again, in these cases probably because some key aspect was found unsatisfying, until a later discovery fixed the problem, allowing the idea to become popular again. So there’s hope for the unemployed psychoanalyst yet, assuming he can stick around for a half-century.


Here are the search terms I used:

“population genetic” OR “population genetics” OR “genetics of populations”


“darwin*” NOT “social darwinism” NOT “social darwinist” NOT “social darwinists”

“mendel” OR “mendelian*” OR “mendelis*” NOT “mendelss*”

I put the last restriction on the Mendel search because I got a lot of results about the composer Felix Mendelssohn.

As with the first post, I searched the full text for both articles and reviews.

Labels: ,


  1. I’m a bit confused about the scales you are using on your graphs: in your original post, word counts were in the thousands; here they are expressed in decimal points. Have you changed your scale, or are the words invoked above really so rare that they only have .2 or so occurrences in relation to the total amount of text in a specific article? I’m a bit curious as to how you’re justifying the trend here against a scale that is astronomically larger amongst the other words. As I am writing a post in response to your articles I would appreciate you sending me a better explanation of your methodology via e-mail. Your points are compelling, but “population genetics” seems to be used the same amount as “deism” was probably used in 1791; I don’t think that anyone back then would agree that in this perceived rise lies the seeds of a ubiquitous theory.

  2. Walker, this is probably because he has log-scaled the Y-axis so it is easier to see the trends. This would imply that there were a much greater number of articles compared to the ones thrown up in the previous search for humanities articles in which he didn’t need to tinker with the Y axis to get a decent graph.

  3. ok scratch that, looking at it again the antilog would result in 1 paper. He/she has transformed the y-axis to make the trends easier to see.

  4. Maybe my wording was a bit vague, but the data-points are: 
    # of articles with keyword / # of articles containing a neutral word 
    It’s like how many people are sick, vs. what fraction of the whole population is sick. I’m estimating total population size by getting all articles containing variants of 4 highly frequent words — admittedly that’s not perfect, although if I could search for “the”, I would miss no articles, because “the” is so frequent. 
    The important thing is that I’m comparing the change over time of keywords to change over time in neutral words, so I’m picking out only the changes over time that are unique to keywords, rather than those that apply to words in general.

  5. Agnostic, instead of going into these discussions on the graph technics, I am more interested in getting to know the origin of your frustration that makes you put so much effort in this. I don’t feel the need to defend this loose group of academics you are criticising, neither myself (that is: I don’t count myself among them), but I wonder if you have actually read through some of these (recent) articles, instead of only counting them? (I guess your own article would also be counted among those that quote these terms). 
    I mean: a lot of people (maybe not enough yet) know that words such as postmodernism etc. do not refer (anymore) to a theory, but to a collection of ideas that aim(ed) at describing social, cultural, political, … dynamics, including historical interpretation. If you are really serious about reflecting on how we can engage in critical discourse ‘for the better of society’, you should accept that we need some kind of vocabulary to understand each other, and that, for the very reason that they have been used before, these terms are part of it. The next step is using them in a cautious, reflexive way. The articles that take this approach also exist. Maybe you should try to extract them out of your data. Only then the exercise would become interesting.  

  6. This is interesting, if not conclusive. We readers should be grateful he/she was willing to put this much effort into it.  
    That said, I really wish you would try to be and to sound more objective.  
    IMHO, feminism is very different than the other “isms” you mention here.