Sex Differences
A new study, covered by CNN and other news organizations this week, shows that women benefit more than men do from early Alzheimer's interventions.
At the end of this newsletter I'll discuss the pros and cons of this study, because it provides a nice example of my broader theme: Statistics has been both enormously helpful, and deeply damaging, via the study of sex differences over the past century.
(This particular newsletter doesn't pertain to gender, or to cases in which a person's biological sex is ambiguous. I'll cover these topics in other newsletters. Here my focus is on people whose chromosomes, hormones, and anatomy all distinguish them, physically, as male or female.)
Statistics and sex differences: Some upsides
Some sex differences are obvious. Men, on average, are taller than women. Elementary school teachers are more likely to be women, while CEOs of Fortune 500 companies are more likely to be men. Although statistics add precision, we don't need stats to recognize these differences.
Other differences aren't so obvious. For example, women experiencing heart attacks often report different symptoms than men do (e.g., shortness of breath, pain in the abdomen, and pressure in the upper back rather than in the chest). As I noted in a recent newsletter, these differences weren't widely known among medical professionals until recently, because males were traditionally the main participants in studies on cardiovascular health. Thus, women who experience heart attacks have sometimes been misdiagnosed.
As this example illustrates, women are often disadvantaged by sex differences that aren't immediately obvious. Statistics has played an essential role in correcting some of these disadvantages. For example, simple frequency data revealed differences between men and women in the kinds of heart attack symptoms they experience. Another example: data on mean or median salaries show that in given professions, women earn less than men do. These kinds of statistics are useful because they're easy to understand, and they don't force us to rely on anecdotal evidence of gender inequities.
Sometimes the statistics get a little more complex. For example, some people (usually men) have argued that in certain lines of work, women earn less than men do because the women are less experienced, or have less training, or whatever. This argument is typically dismantled by statistical analyses showing that even if we take into account differences in education, experience, training, and/or seniority within a company, women still earn less than men do.
Further analyses show that the gender wage gap is greater among high-paying jobs, in part because the federal minimum wage creates what's known as a floor effect: Men and women working at minimum-wage jobs earn the same amount of money per hour, whereas at the executive level, we find more variability in wages (favoring men). Statistics also tell us that this gender wage gap has persisted, largely unchanged, for the past 15 years. Bottom line: men tend to make more than women do after controlling for all the key variables. (What it means to "take into account" or "control for" these variables means different things, mathematically speaking, depending on what type of statistics are being run, but in some cases it's as simple as making apples-to-apples comparisons of salaries among men and women with comparable levels of education, experience, etc.)
Many more examples could be cited. The main point is that relatively simple statistical procedures have been used to identify gender inequities and other differences that might not be immediately obvious.
Statistics and sex differences: An infamous error
Statistics have also, in many cases, created or reinforced gender inequities by misrepresenting sex differences.
One of the most enduring examples is the idea that women are less intelligent than men because their brains are smaller. This idea has been kicked around for centuries. Initially, the statistics involved were pretty simple – measurements of living people showed that men's heads were about 10% larger. Filling empty skulls with buckshot or bird seed showed that women's brains were about 5 ounces lighter. These stats represent averages that scientists calculated, and differences within each sex – what we would call standard deviations – were generally ignored (which is unfortunate, because variability in head and skull size within each sex is greater than variability between sexes.)
The methods and analyses used to compare brain sizes got a bit more sophisticated in the late 19th century, but the conclusions remained unchanged. For example, James Crichton-Browne, a prominent neurologist and founder of the still-renowned journal Brain, conducted anatomical studies of 400 brains and concluded that the smaller brains of women were responsible for their intellectual inferiority. Others inferred that because Crichton-Browne also found that the brains of "lunatics" were lighter than those of "normal people", the relative lightness of women's brains explains why they're more irrational and morally weak than men are. (Back then, most experts assumed that the physical features of the brain reveal psychological ones, such as intelligence, personality, and even moral character.)
This is nasty stuff – and not very sensible. Some 19th century scholars objected that if intelligence is simply a function of brain size, then larger people (regardless of sex) would be smarter than smaller ones. Surely this cannot be the case. If it were, then Donald Trump (6 foot 3, 243 pounds) would be smarter than Albert Einstein (5 foot 7, 154 pounds).
Neurologists who felt threatened by this objection quickly replied that what predicts intelligence isn't brain size per se, but rather brain-body ratio. Those with larger brains, relative to their body size, will be more intelligent than those with smaller brain-body ratios. Indeed, by the late 19th century, scientists agreed that the brain-body ratio tended to be larger for men than for women, and so well into the early decades of the 20th century, many neurologists still assumed that women's supposed intellectual inferiority reflects something about the size of their brains.
What we see here is statistics being complicit in misogyny. The neurologists in question (all men) used various statistical procedures to calculate brain-body ratios, and then asserted that because these ratios were larger for men (an objective fact), male superiority in intelligence (not an objective fact) is causally linked to sex differences in those ratios (a speculation).
Statistics and sex differences: Correcting an error
Now, a century later, our views on sex differences in brain size and intelligence are quite different, thanks in part to increasingly sophisticated statistics:
1. Larger brain-body ratios don't predict greater intelligence. The evidence for this among humans is clear, but my favorite illustration is the so-called chihuahua paradox. Chihuahuas have larger brain-body ratios than any other breed of dog, but there's no basis for saying they're the most intelligent among their species. (Anecdotally, I would say that they're on the lower end of doggie IQ. If you happen to own a chihuahua, my apologies to you and your tiny, surely adorable creature.)
2. Overall brain size is unrelated to intelligence. Long story short, one of the most entrenched assumptions of pre-20th century neurology has been clearly disproven by research on people (and across species, assuming that you trust the ways researchers have imputed intelligence to other species). If you have a bigger brain, it means you need to wear a bigger hat. Period.
3. Although scientists have talked a lot about "male brains" and "female brains", and although there are many anatomical and physiological differences between them (including differences in the sizes of certain parts), it's generally unclear how these differences translate into psychological characteristics such as intelligence. This is a controversial topic that I'll return to shortly, and address in a separate newsletter.
Statistics and sex differences: Recent progress
So far, I've touched on the use of statistics to identify important sex differences (in heart attack symptomatology and in wages). I've also noted how statistics were once used to support a misogynistic belief (women have lower IQs owing to smaller brains are smaller), and how statistics have helped correct that belief.
In my opinion, the biggest change that occurred over the past century is that statistics are used now to address much more specific questions about sex differences. Reputable scientists aren't asking whether men and women differ in "intelligence". They're asking, for example, how stereotypes and other environmental influences affect performance on highly specific tasks (e.g., spatial cognition). Reputable scientists aren't asking whether women are the "weaker sex" – a point of vigorous debate in the early 20th century. Rather, they're focusing on more specific topics that might, if you wish, be framed in terms of strength, or hardiness, or whatever term you deem suitable. For example, although men tend to be physically stronger than women, studies also show that:
—women live longer on average
—women tend to have greater resistance to the effects of famine and epidemics
—women have a higher tolerance for pain
—women tend to have better impulse control
—women have higher college attendance and graduation rates
In short, there is no "weaker sex". Each sex has particular "strengths". Even then, to say that in some particular sense one sex is stronger, or more persistent, or whatever, we're generalizing across large and highly diverse groups of people. Research on sex differences typically reveals more variation within each group (men or women) than between the groups. This is illustrated by the figure below:
The x-axis of this figure is height in millimeters. The y-axis is the number of people observed at each height. The figure shows two overlapping curves, the green one on the left for women, the violet one on the right for men. This figure tells us two things: (a) The average for men will be higher than the average for women. (b) There's a lot of overlap. In other words, a substantial number of women will be taller than a substantial number of men.
Curves like this are the norm in studies on sex differences. This tells us that "sex differences" are often averages with many exceptions at the individual level.
Neurosexism
Thanks to evolving social attitudes as well as better statistics, the study of sex differences is vastly less biased than it was a century ago. However, biases still exist. In critiques of brain research, these biases are referred to as "neurosexism."
For example, Simon Baron-Cohen (a famous psychologist whose brother achieved even greater fame as Borat) has proposed a widely-discussed, controversial theory of autism which holds that human brains vary on a continuum ranging from highly systematic to highly empathetic. According to this theory, people with autism have brains that represent one extreme of this continuum: unusually strong systemizing tendencies, coupled with extremely impaired empathy. Integral to Baron-Cohen's theory is the idea that systemizing brains are male brains, and empathizing brains are female brains. This is why, as he once put it, "people with the female brain make the most wonderful counselors, primary-school teachers, nurses..while... those with male brains make the most wonderful scientists, engineers, mechanics..." As it turns out, statistics play a key role in some of the technical aspects of this theory.
Baron-Cohen's theory seems like an obvious example of neurosexism (unless you're among the small minority of experts who wholeheartedly embrace it). Less obviously, but also perilous, are studies that misrepresent gender differences, whether or not sexism plays a role behind the scenes. This brings me to an Alzheimer's prevention study published on Tuesday.
Alzheimer study background & methods
Alzheimer's disease develops slowly – changes in the brain begin to occur at least 15 to 20 years before the onset of symptoms – and so researchers have begun to focus on lifestyle changes that can delay, if not prevent, these symptoms from emerging.
The study I'll be discussing, published in the Journal of Prevention of Alzheimer's Disease, was conducted by neurologists at Cornell and nine other institutions. Participants consisted of 80 adults (mean age 60.4 years) with a family history of Alzheimer's disease (AD) who were enrolled in an AD prevention program at Florida Atlantic University. This program provided each participant individually-tailored, evidence-based recommendations for improvements in the areas of diet, insulin resistance, physical activity, sleep, stress reduction, social interaction, and choice of leisure activities. (I was happily surprised to learn that good choices in each of these areas reduces the risk of, or at least delays the onset of, AD.)
Participants were tracked for 18 months. What the researchers called the primary outcome of interest was change in performance on a test of cognitive functioning called the modified-Alzheimer’s Prevention Initiative Cognitive Composite (m-APCC), which tests low level cognitive skills like perceptual speed, working memory, pattern analysis, etc. For example, a person might view the figure below and then choose which of the five symbols (A through E) goes in the empty box in the lower right.
The researchers also looked at what they called secondary outcomes of interest, including scores on other cognitive tests, as well as various biomarkers obtained through blood tests.
(The answer to the puzzle above is C, by the way. If you're an educator, you may recognize this as a type of question from the Raven's Progressive Matrices, one source for the m-APCC.)
Alzheimer study findings
One of the main findings – and good news for everyone – is that regardless of whether male or female, all participants who were moderately compliant with the program's recommendations (i.e., by following at least 12 of 21 lifestyle suggestions) showed small but significant improvements on the m-APCC and other cognitive tests. Whether or not these individuals had begun to develop AD, their cognitive functioning improved after 18 months through easily implemented lifestyle changes (eating more healthy food, exercising more, engaging in more cognitively stimulating activities, etc.)
No sex differences were found in improvement on the cognitive tasks. However, women improved more than men did on several biomarkers. Women who had begun the study with no cognitive symptoms of AD showed more improvement than men did on a global indicator of cardiovascular risk, as well as in HDL levels (high-density lipoprotein – aka the "good cholesterol"). Women who had begun the study already experiencing mild symptoms of AD showed similar improvements, as well as favorable changes in blood sugar levels. In all, very encouraging findings.
Statistical concerns
The results are good news for everyone but especially pertinent to women, since they constitute roughly two-thirds of AD cases. However…the study consisted of just 39 men and 41 women. Of those 41 women, 33 began the study with no cognitive symptoms of AD, and 8 began the study with mild symptoms.
You can probably guess my main concern: The sample size is much too small to make any reliable generalizations about sex differences in responses to AD prevention strategies. Statistically speaking, there's just too much variability among women (and men) in the kinds of variables that were studied.
Another concern, albeit a more narrow one, is that the study didn't show any impact on AD per se. What it shows is that everyone benefits physically from healthy lifestyle changes – which we knew already – and that a small group of women benefitted more than a small group of men did. The study is ongoing, and so it's possible that benefits will extend beyond the 18-month mark and come to include performance on cognitive tasks. That would be great news, but the sample still seems much too small to be trusted.
Context for concerns
Why bother picking on a study for having a small sample? There are lots of studies like that.
What concerns me is that the results are pitched by the researchers (and in news reports) as suggesting better outcomes for women than for men. There's a history of overconfidence about the efficacy of women's health care. For example, as I mentioned in a recent newsletter, a number of studies show that even now, some medical professionals consider women to be largely protected against heart disease, an assumption that leads to under-treatment, among other things.
In short, I'm concerned about a study with such a tiny sample creating the impression that women are especially likely to benefit from AD intervention programs. Program administrators who hear about this study might pay less attention to women who participate (or pay more attention to the men). The results don't justify doing so.
According to this study, and others, engaging in cognitively stimulating activities prevents – and may even reverse – the cognitive declines that occur as we age. So, if you've read this far….you're welcome! I'm glad to have helped keep you sharp.