Junk Science and Logical Fallacies

Rochelle Terman
Aug 9, 2017 · 6 min read

The distribution of preferences and abilities of men and women differ in part due to biological causes and that these differences may explain why we don’t see equal representation of women in tech and leadership. ~James Damore

Damore’s thesis relies on three interconnected claims:

  1. The populations of men and women exhibit significant differences in their distributions of personality traits, including interests, skills, and proclivities. We’ll call this the descriptive claim.
  2. The descriptive differences we observe are due to biological causes, and are therefore fixed and universal. This is the causal claim.
  3. The above two facts explains the gender gap in tech, which undermines the wisdom of certain policy initiatives, such as affirmative action and diversity hiring. We’ll call this the normative claim.

I’m going to argue that each of these claims are either scientifically wrong or fatally misleading.

Descriptive Claims

Damore makes a series of claims about personality differences between men and women. Women, on average, have more “openness directed towards feelings and aesthetics,” “extraversion,” “agreeableness,” and “neuroticism.” When talking about science, it is important to not to confuse the model for the reality. Damore here is relying on a particular model of personality that relies on strong assumptions. Ignoring such assumptions leads to highly misleading results.

Take, for instance, the claim about neuroticism. Neuroticism is one of the “Big 5” personality traits developed in psychology. The measure is not a theory-driven construct but an empirical artifact derived from personality data. In brief, researchers collect a database of words used to described people (e.g. “messy,” “neurotic,” “conscientious”) and then apply factor analysis to “group” these words into 5 clusters or dimensions. The labels applied to these dimensions — not to mention their normative connotations — are subject to considerable interpretation. For instance, the dimension commonly labelled neurotic may just as easily be called dynamic, sensitive, or inspiring.

In addition to the interpretation problem, the Big 5 has come under heavy scrutiny for methodological failings. Since the model is derived completely from lexical descriptors, it tells us more about how people used language than the objective reality of personality. In other words, Damore’s “finding” that women are more neurotic than men is really a refection of the fact that people describe women as more neurotic than men.

Damore then goes on to link these highly interpretable personality differences to gender gaps in programming. Coding, he argues, is more suited to those with personality traits that are disproportionately exhibited in men. Yonatan Zunger has already debunked Damore’s spurious claims about software engineering, and I encourage you to go read his piece. Other have noted that Damore’s reference to population-level findings are ill-suited to environments like google, where individuals are clearly not “average.” In general, we cannot easily apply vague, interpretable population-level differences to specific problems and environments. Doing so is highly misleading at best, and malicious at worst.

Causal Claims (i.e. Biological Determinism)

Let’s say, for the sake of argument, that Damore is correct in his descriptive claims (which he isn’t) about average differences between men and women. That prompts serious question: what explains these differences?

Damore frames the debate as a choice between two causes or some combination between them: 1) malicious bias, misogyny, sexism, or 2) innate, biological mechanisms. He argues that we have failed to adequately appreciate the later explanation.

Damore’s causal claims are wrong in two respects: First, he severely overstates the scientific evidence linking biological mechanisms to personality differences. Second, he fails to address or even acknowledge a third explanation for gender differences: socialization.

On the first issue, Damore argues there are “clear biological causes and links to prenatal testosterone” that explain gender differences. This constitutes a clear instance of pseudo-science, over-simplifying and cherry-picking evidence to suit one’s spurious argument. Any biologist worth his or her salt will tell you any link between biological mechanisms and personality traits is extremely complex. There is no “gene” for neuroticism. There is no “hormone” for having an interest in people over things. While biological mechanisms may affect brain development in males and females, the know very little about how these differences manifest in personality or learning. Anyone who tells you science has incontrovertibly shown a biological bases for the gender gap in tech is either lying or a TedTalk speaker.

Damore further attempts to substantiate his claims by saying they’re “exactly what we would predict from an evolutionary psychology perspective.” The problem is that evolutionary psychology can be used to predict almost anything you want it to. Imagine a world, for instance, where men exhibited more “sociable” and “agreeable” traits than women. We could very easily concoct a story ascribing this difference to the evolutionary necessity of men to hunt and fight together, lending a more pro-social disposition. It is this ad-hoc flexibility that leads many critics to call evolutionary psychology unfalsifiable, and thus unscientific.

Finally, Damore makes the (pretty outrageous) claim that differences between men and women are “universal across human cultures.” No doubt he relies once again social-psychological findings based on Big 5 personality metrics. But there is another field that is uniquely qualified to assess cultural diversity in gender norms: cultural anthropology. And it is in fact the consensus in that field that gender and sexuality norms are highly mediated by social and environmental factors.

In fact, while Damore seems at least superficially acquainted with the biological literature, he appears completely ignorant of social science. This is evident by his perpetual reliance on the spurious dichotomy separating “bias” from “biology.” In fact, there is a third, and altogether more convincing, explanation for the gender differences we observe: socialization.

From the time we are infants, human beings are exposed to a complex set of intersubjective norms and rules demarcating proper and improper behavior. These rules constitute our notions about gender, from “boys like blue, girls like pink” to “men build and women nurture.” Children internalize these rules into their personalities, which manifest as interests and proclivities. In other words, girls do not necessarily have to be victims of overt sexism or direct bias to be influenced by a culture of gender roles and norms that shape their lives.

Even if the gender gap if driven by differences in interests and personality, it is far more likely these differences are determined by environmental and social factors than crude biological causes. This is especially relevant in STEM, where social influences shape women’s careers from the time their are children to CEOs.

Normative Claims

Finally, Damore writes that these “biological” differences in personality traits should make us question policies such as diversity initiatives in the workplace. Again, I’m going to be charitable to Damore and assume he is correct (which he isn’t) about the biological roots of the gender gap. So what? Biological differences alone tell us nothing about how we should order society. In other words, there is no logical reason why the existence of distributional differences should negate our desire for more diversity or equity in the workplace.

Damore falls into the “appeal to nature” fallacy to justify his normative preference on policy. But computers are not natural. Email is not natural. So why would our hiring practices strive to be natural? Our desire for more diversity in STEM in rooted social values and ideals, not in an misguided attempt to reproduce the “natural order” of things. Diversity may generate other goods we value, such as justice, equity, and innovation.

I want to close by addressing the debate this memo has provoked. Many people have praised Damore for opening a conversation that they believe has been stifled by “political correctness.” But people have been debating the the biological roots of gender (and race) inequality for hundreds of years. In addition to taking a sociology or anthropology class, I suggest Damore become more familiar with history, especially the history of scientific racism and biological determinism. Perhaps there he will gleam some insight into why his memo has provoked such passion.