Bias in Science: Further Thoughts from Skeptech
This weekend at Skeptech there were many great talks, but one in particular stood out to me as something that could use a bit of fleshing out.”The Other Scientific Method” looked at ways that bias can influence science. The panel itself focused on things like discrimination in science and the personal experiences of those on the panel. These are incredibly important topics, but it missed one important thing: the conclusions that we reach through the scientific method can be bad if we begin from a place of bias. Bias can affect what gets funding to be studied at all, it can affect the way results are interpreted, it can affect things like diagnosis of illness, and it can affect the results of studies.
I can only get into a few examples here, as sexism in science is everpresent and affects nearly everything we do, but I will try to point out a few representative samples of different ways that bias impacts science other than keeping minorities out of research and science-based careers.
One of the most obvious ways that bias affects the results of science is through confirmation bias. This can affect what is reported, what is considered significant, which studies are rated more highly in peer review, and even whether a researcher notices something in their research. It can lead to resistance of new advances, including things like evidence that both men and women can do science and math, or research on intersexuality. Because we tend to discount things that don’t agree with our preconceived notions, science is not the objective self-correcting mechanism we like to imagine it as: sometimes our research is flawed because the people doing it have a subjective perspective that affects how they interpret their data and what data appears important, relevant, and meaningful.
One great example of this is neurosexism, which is the assumption that man brains and woman brains are inherently different. We do see that men and women tend to excel at different things and that their brains are formed slightly differently. However confirmation bias pushes researchers to assume that this is a natural, biological tendency rather than looking at potential social influences. This is ridiculously obvious if we look at some of the explanations for why women aren’t more common in STEM fields.
Beyond confirmation bias, other kinds of bias can also affect which things get funding or interest for research at all. One great example of this is the fact that many people are researching “treatments” for obesity, despite the fact that obesity itself is not really a disease and their time might be better spent researching the diseases that tend to accompany obesity. This plays directly into fatphobia and is part of the internalized bias many of these researchers have. Similarly, women’s reproductive systems went misunderstood for many years and are only just starting to get researched because they weren’t really considered important enough for quite some time. We only just learned what the clitoris looks like! This clearly reflects the fact that women’s pleasure and women’s bodies were considered fairly unimportant for quite some time (and in many ways still are). Let’s not even get started on the complete erasure of intersex bodies from the medical literature except as “freaks” or oddities.
People also just straight out misinterpret or forget to read all of scientific studies so that they can confirm their biased opinions.
A final element of this is who actually is involved in studies. Studies tend to be held on college campuses, where white, upper class people are disproportionately represented. Oftentimes studies are then generalized to the whole population without thought for the fact that maybe people who aren’t college psych freshman behave differently. There needs to be a concerted effort to include diverse populations in our studies and experiments so that we have a more accurate representation of reality.
Additionally, bias affects our education efforts and what information seeps into public awareness. One classic example of this is heart attacks. Most people think that they know the symptoms of a heart attack, but in reality they know the symptoms of a heart attack for a man. The symptoms in women are quite different.But because male is generally assumed to be the default, there haven’t been many education efforts around women and heart attacks, women and heart disease, or the fact that heart disease is a huge cause of death in women.
Medicine, particularly treatment of mental illness, is one of the places where these biases have significant consequences. There are TONS of examples of this, but to name a few: women’s concerns are more likely to be written off as “merely” psychological, white children are far more likely to get a proper diagnosis of autism and appropriate services, depression is diagnosed most often in white women with men and black women being seen as angry. Borderline personality disorder is diagnosed most often in women, and the symptoms are associated with traditionally understood “female” qualities like hysteria and excessive emotion. Perhaps these two things could be related. The things that we view as problematic and thus in need of treatment are influenced by our biases.
Bias also affects the course of treatment. I don’t really even want to touch on medical masturbation because it squicks me out massively, but fatphobia is a great place to go for examples of this. There are tons of examples of fat people being told their medical issues are all the result of their weight and if they simply lost weight then they would be healthy. It’s also extremely difficult for men and people of color to get treatment for eating disorders or for women to get appropriate treatment for their reproductive systems (see: the defunding of Planned Parenthood). All of these things are ways that science has learned to incorporate bias into how it views diseases.
A lack of women and minorities in science is a problem. But it’s just the tip of the iceberg in terms of the existence of bias in science. Bias affects everything from research results to how we diagnose and treat diseases. A lot of these are more subtle ways that bias influences science. We’ve learned to trust peer reviewed papers, but those papers are coming from human beings who have grown up with the same biases that we have, biases that affect what they think is important and what could potentially be true.
I was recently told that most early drug trials, the ones at the mouse stage, are typically done only using male mice. The rationale being that the estrus cycle of female mice complicates the results.
I don’t have independent confirmation of this, but it sounds about right: “non-estrus cycle” is considered the default, the “normal,” and treated as such, while “estrus-cycle included” is counted as a confounding factor. As a consequence, the way drugs interact with the female body is just not investigated at all in many cases, and when it is it is treated as though being female is an abnormal complication of the male norm. There’s no reason why the reverse could not be the case, with female as default, and no reason at all for there to be a default in the first place.
Please tell me you understand why a system that stays at a steady hormonal state would be easier to test on than a system that cycles from that state to another. If a drug messes up the steady state it is going to mess up the unsteady state, but the reverse is not necessarily true. It is good scientific practice to start at the simpler system first and then see if it can also be applied to the more complex system. Worst case starting male you get a drug that is only effective in males. Worst case starting female and you’re passing up on multiple possibly life saving treatments because of complicating factors that are not present in fifty percent of the population.
Citation needed for showing that men’s hormones don’t fluctuate. (Please tell me you’re not a research scientist.)
I really don’t understand how someone named Samantha can’t see how freakin’ ridiculous and *sexist as hell* the highly incorrect and overly simplistic (irony!) statement of “men are really simple, and women are complex, duh!” is. Because that’s about all Samantha said: “Well, duh. I mean women have periods! So of course they are more complex. Men are so simple. I don’t really show any proof of this and am just talking out of my ass because I make a lot of sexist assumptions, but hey! Men are simple and women are complex. Duh.”
There’s a few issues with your sexist bioreductive comment. First, you prove the point of the article quite brilliantly (though I’m assuming unintentionally). In your view, biomedical scientists are biased towards “stable systems” (gendered and read as “male” or “man”). Setting aside for a moment the sexist view of female bodies as “unsteady systems” and male bodies as stable, your logic is that if it works for a stable system, maybe then it could be applied to work on a complex system–and if it doesn’t, oh well! This seems to me indicative of another kind of bias in science: a bias towards “easier” science. It is no coincidence that what is considered “simple” is male and therefore more useful. Which brings me to my second point.
Ideas of the female body as “unsteady” and “complex” and “unstable” are regular sexist tropes in science (and elsewhere). It’s exactly that sort of thinking that leads to the problems Olivia is referencing in her post. I have not seen any credible scientific evidence (feel free to link some!) that demonstrates that female bodies are “more unsteady” and “more complex” than male bodies. What does it mean to be “more complex”? What does it mean to be “unsteady”? You’re throwing these words around uncritically as if their meanings are clear and uncontested.
Third, to build on Mary’s comment, male bodies also undergo hormonal fluctuations. Here are a few examples of research on male hormonal rhythms and cycles from Google Scholar. The ideas you have put forth in your comment as scientific fact are, in reality, sexist tropes you (and other scientists) have internalized and assumed as true. Again, this is an example of exactly what Olivia was talking about with regard to how cultural beliefs influence the practice of science.
Fourth, you are working under the assumption that the most important factor in whether or not a medication works on a human body is sex hormones. Citation needed.
Yes. What Will said.
Even I know more about this subject than you and I’m not a freakin’ research science. Wow.
Your comment proves this post, really. Men don’t have hormones, you guys! Do you REALLY believe that?!
*scientist
Aside from what Will said, you are using bad logic here:
“If a drug messes up the steady state it is going to mess up the unsteady state, but the reverse is not necessarily true.”
The only evidence for this in medicine exists because the male body is tested first (because it is presumed as the normal). There is a subset of drug testing that is generally performed first on female bodies, namely those that are being aimed primarily at women. Drugs for treatment of breast cancer, for instance, which affect women far more often than men, or for regulation of female infertility or unwanted fertility.
These drugs, which, when all goes correctly, act as desired in the female body, sometimes “mess up” the male body. So your assertion that it’s going to always be best to test the male body first whenever feasible is demonstrably untrue.
How many drugs have been passed over for conditions that affect both men and women, because they “messed up” the “steady-state,” which would have acted quite well in women if that had only been investigated? WE DON’T KNOW. We can’t know, because this is not the typical order of operation. And that is the problem. You’re potentially throwing away good therapies for half the population because they don’t work on the other half. The only reason this is acceptable is because the half for which it would work is female, and the half on which it doesn’t work is male. There’s no reason whatsoever to not test simultaneously on male and female, except that it isn’t considered to be worth the effort and cost to test females.
I would have thought it was utterly incompetent for a drug study on rats not to include female test subjects. There are even international guidelines for this sort of thing: http://www.oecd-ilibrary.org/environment/test-no-408-repeated-dose-90-day-oral-toxicity-study-in-rodents_9789264070707-en