The full head of this NBC report was Not smart enough? Men overestimate intelligence in science class. A subhead read, “Even when grades show different[ly], men overestimated their class ranking.”
The gist of the article was that men thought that they were in the 66th percentile of their class, whereas women thought that they were in the 54th percentile, which is considerably closer to the truth. In addition, women in Stem courses are often belittled and think they are inadequate.
The first finding, that people overestimate themselves, will come as no surprise to those familiar with the Dunning-Kruger effect, which we discussed here several years ago.
I was somewhat curious about the women, though, so I performed an experiment of my own. I promise you that it has no statistical significance whatsoever. But first it is probably necessary to set the stage: Between 1999 and 2014, I taught a course called Epics at the Colorado School of Mines, primarily to first-year students. Epics is an acronym whose etymology I promise you, you do not want to know. It is a project course in which the students are divided into teams of five, design something, and often build a prototype for an outside client. For example, we have designed wheelchair-accessible cattle guards, chalk-making machines and distillation equipment for certain cottage industries in the developing world, a student health center, examining rooms for physicians, a pedestrian bridge or tunnel across a major highway, and sensors and other equipment for balloon experiments undertaken by NASA.
I undertook an archaeological dig onto my hard drive and found my grade books for that 15-year period. I chose three terms, not entirely randomly, because I wanted to span the period and look at the grades and evaluations of males and females. I had to search a little bit, because many first names were ambiguous or non-European, and I could not always remember who was female and who was not. Eventually, I ended up with three terms: Fall 2002, Fall 2010, and Spring 2013.
One of the exercises at the end of the year was a peer evaluation, in which the students evaluated themselves and the other students on their teams. At the beginning, many individuals, invariably male, gave themselves and their teammates 5 out of 5 on each of eight criteria. By Fall 2002, I had prohibited that practice, so my data are not biased by such inflated scores. With minor exception, the peer evaluations were approximately constant across the three terms.
Additionally, I assigned a mentor grade. Initially the mentor grade was largely subjective, but later on included certain objective criteria based, for example, on in-class assignments. In 2002, the grade was subjective; in 2010 and 2013, it was largely objective.
Finally, I looked at final grades.
In total, the three classes had 69 students (which is roughly half the number on which Kruger and Dunning based their conclusion); 18 of those students, or 26 %, were female. I segregated the three grades – peer evaluation, mentor grade, and final grade – by sex and calculated the averages of each for the whole class, for males, and for females. If this were a proper paper, I suppose I would post something called Supplementary Material online in some kind of cobweb site where no one would look at it. Instead, I will just tell the results.
(1) To 2 significant digits, the results across the three classes are fairly consistent, even though the criteria for the mentor grade evolved considerably.
(2) I had a vague suspicion that the women in the class rated themselves below the class average. Not so in these three classes: in two of the classes, the women rated themselves approximately the same as the class average, and in one class measurably higher.
(3) The mentor (that is, I) rated women slightly higher than the class average in two of the three terms and the same in one of the terms, again, to two significant digits. Anecdotally, it seems to me that many times when there was a woman on a team, that woman took over the team and became team leader. Unfortunately, I did not keep notes on this phenomenon.
(4) In two of the classes, the women’s final grades were equal to or higher than the class average, and in one class, a bit below. The final grade is not an independent variable, because it includes the peer and mentor evaluations.
Everything we think we know in psychology seems to be based on the behavior of undergraduates, so I will be careful how I phrase this: I can find no evidence that women who enter the Colorado School of Mines as freshmen and complete one or two semesters are less competent than the men, are seen as less competent by the men, or see themselves as less competent.