Prestigious journal publishes sexy selfie study

Stephen Oliver writes:

Not really worth blogging about and a likely candidate for multiverse analysis, but the beginning of the first sentence in the 2nd paragraph made me laugh:

In the study – published in prestigious journal PNAS . . .

The researchers get extra points for this quote from the press release:

The researchers say that the findings make sense from an evolutionary point of view.

In evolutionary terms, these kinds of behaviours are completely rational, even adaptive. The basic idea is that the way people compete for mates, and the things they do to put themselves at the top of the hierarchy are really important. This is where this research fits in – it’s all about how women are competing and why they’re competing.

All right, then.

8 thoughts on “Prestigious journal publishes sexy selfie study

  1. They seem to have buried the lede. If they’ve discovered that evolution is “rational” that’s quite a finding right there.

    P.S. For my part, I’ll need some convincing that publishing in PNAS is rational at this point in time.

    • Found this quote yesterday:
      “Aristotle once defined human beings as rational animals. They are, at least, rationalizing animals.”

      (Michael Gerson, Washington Post)

      That would seem to apply quite nicely to that second sentence.

    • They don’t exactly say “evolution is rational.” Rather they are saying that the behavior that is observer is consistent with evolutionary theory. That seems reasonable to me.

      Of course, they also say, “Think of her as a strategic player in a complex social and evolutionary game. She’s out to maximize her lot in life, just like everyone.”

      That’s not what I understand evolution to predict. Rather it predicts that in some long-term average sense living things behave to maximize evolutionary fitness. Thus grandparents dote on their grandchildren and are happy to babysit them and otherwise improve the grandchildren’s lot in life. See also John 15:13.

      Bob

  2. I skimmed through the paper. Beyond the, um, interesting hypotheses and the many (many!) pages of stata calls in the supporting information, I can’t find how they define “sexy selfies”. Has anyone figured this out?

    • From the press release: her team tracked posts where people had taken selfies and then noted that they were tagged sexy, hot or similar.

      so it sounds like they used *other people’s tags* to decide whether people responded to it as sexy.

      but whatever, I don’t care enough about this paper to read it to figure out what they did really.

  3. Does PNAS have no standards for data release?

    The osf website linked to in the paper has csv files but no code files *at all*. That is what I call a passive aggressive data release—meets the technical requirement of a data release but it is not in the spirit of reproducibility. It was the same with another PNAS study I looked at recently.

    And this is what people often give me when I ask for their data: just the data files, no information about the columns, no code, nothing else. Often I get emails like, the columns are self-explanatory. Is fpr first-pass regression or first-pass reading time? You can guess by looking at what the data look like, but it is just a series of guesses. When you ask them for the code, some actually release it right away; so they had it all the time, but they didn’t send it to their correspondent? What’s up with that?

    To be fair, the authors of the PNAS paper release a 160+ page pdf file (!) called supplementary material or the like, and it actually contains code. Code in a pdf file only makes sense when it’s in an appendix of a paper; the publicly available code and data should provide a reproducible workflow, an R or Rmd file at the very least.

    What’s happening now is that people are releasing data with their papers to meet the requirements of journals, but the journals are not setting any standards for what counts as data release. They wouldn’t accept a journal article submission that doesn’t follow APA or whatever field-specific standards. But they will accept pretty much anything that passes for a data+code release, even a pdf file containing the code behind the paper.

    I’m always surprised that the people writing these papers can be so inconsiderate. Don’t they have the imagination to ask themselves, what would a code+data release look like that’s actually usable by someone without putting in dramatic amounts of detective work? If I were a reader, what kind of code+data would help me understand what lies behind the paper? Surely they ask themselves that question when writing the paper: what does the reader need to know to understand my work?

    Incidentally, I noticed some asterisks in the paper which I doubt will hold up to closer scrutiny. One of the tables shows two coefficient estimates for inequality in two different models, both were 0.28 with a standard error of 0.14, and each has one asterisk next to it, meaning less than 0.05. That SE brings the coefficient remarkably close to 0. Probably people using statistics at this level are better off using a Bonferroni correction.

    In their table 1, there are some 51 t-tests shown from multiple linear regressions. Doing a Bonferroni correction means only a few will will come out significant at a revised alpha of 0.00098 (I can’t tell how many because they don’t give exact p-values but only stars). And these are only the tests they reported. If you feel that a Bonferroni will be too conservative, do a Sidak correction—you will find that with this number of comparisons it just doesn’t matter which correction you use.

    • “I’m always surprised that the people writing these papers can be so inconsiderate. Don’t they have the imagination to ask themselves, what would a code+data release look like that’s actually usable by someone without putting in dramatic amounts of detective work? If I were a reader, what kind of code+data would help me understand what lies behind the paper? Surely they ask themselves that question when writing the paper: what does the reader need to know to understand my work?”

      My guess is that it’s ignorance rather than knowingly being inconsiderate.

Leave a Reply

Your email address will not be published. Required fields are marked *