In data we trust – but should we?

By Sabrina Patsch, Universität Kassel and Freie Universität Berlin

For us scientists, data are our daily bread. We collect them, we compare them, we try extract general knowledge from them. Instead of speculating, we ask data to give us all the answers we are looking for. We collect data to improve the daily life for everybody. Or rather, for the average human being. Too bad the average person is between 25 and 30 years, weighs 70kg, and is a white man.

In her new bestselling book “Invisible Women: Exposing Data Bias in a World Designed for Men”, Caroline Criado-Perez addresses what she calls the gender data gap. Since the beginning of historiography, women have shone with absence. Instead, the life stories of men were assumed to be representative for all people. It is based on – or even the origin of – the unconscious thought that the man is the default human. If people say human, they usually mean men.
Criado-Perez addresses this issue on 425 pages (not counting the 75 pages of references) in seven chapters using examples from our daily life, the workplace, design, medicine, the public life and crisis management. She begins her book by quoting Simone de Beauvoir

Representation of the world, like the world itself, is the work of men; they describe it fromtheir own point of view, which they confuse with absolute truth.
– Simone de Beauvoir, The second sex, 1949

showing that, despite being 70 years old, this statement is as true as ever.

Criado-Perez claims that gender neutrality is in most cases mere illusion. Most of the data – or in general information – we gather, concerns men. Based on this, people make decisions that affect everybody – including the half of the population that is not captured by the data. Now most of the decision making is in the hands of healthy, white men which come in nine of ten cases from the USA. Without (necessarily) malicious intent, the decision makers assume themselves to be the “standard human” and they lack perspective. This is why diversity is crucial to design a world that works for everybody.

Let me give an example for the problem with gender neutral design. While real languages are historically shaped and might be influenced by sexist thinking from earlier ages, Emoji is a new language consciously designed by people. It is the Unicode consortium who discusses and selects the emojis which are part of the worldwide Unicode standard [1]. Originally, in Emoji 1.0, most emojis were present in a gender-neutral form, such as the “spy” 🕵️ . While the consortium defines the main specifics of the emoji (“An undercover investigator, wearing a hat, and sometimes using a magnifying glass to closely inspect evidence.” [1]), the specific design is up to every platform. And indeed, most platforms, such as Facebook and Twitter, interpreted the spy as a man. Even if they had been able to create a gender-neutral picture of a spy, most people (including women) interpret gender-neutral figures as men. We tend to assume things as male – until the opposite is proven. So, the seemingly gender-neutral language is not that neutral after all. The only way to make women visible is to name them explicitly. As a result, the Unicode consortium decided to add the “male spy” 🕵️‍♂️ and the “female spy” 🕵️‍♀️ to Emoji 4.0. An important step, even though Unicode still describes the “male spy” as “The male version of the Spy emoji. Currently identical in appearance to the non-gendered base emoji.” [1].

While the design of emojis could be dismissed as a trifle, Criado-Perez comes up with a shocking number of examples where the gender data gap poses a real threat for women. Take medicine. Studies on the efficacy of drugs are often conducted on men. One of the reasons put forward is that female bodies are more complicated since they undergo a hormonal cycle affecting the results. But if the hormonal cycle is affecting the effectiveness of the drugs, this must not be neglected in drug tests. But it is. As a result, many drugs don’t work for women in the same way as for men – or sometimes not at all. Even in medical school, women are often only treated as a variation of the standard. Students are taught anatomy and female anatomy, physiology and female physiology. How can half of the population be a variation?

In academia, we experience first-hand how the blindness to gender issues results in discrimination against women. In Germany, post-doctoral researchers can spend a maximum of six years in temporal positions. If they do not receive a permanent position afterwards, it is often the end of their scientific career. This system disadvantages women in particular, since the critical time for PhDs to achieve a tenure track position coincides with the time women might want to start a family. For many women, combining an academic career with raising a child seems like a Herculean task and they decide to drop out before even applying for their first tenure track position. Men become fathers too, one might think, so they should be affected in the same way, but the numbers tell a different story. A look at the figures is downright depressing – it is a story of lone she-wolves [2, study conducted in the USA]:
The rate of divorces is higher, marriages less frequent and the number of children less for female than for male professors. Among the tenured faculty members, 70% of men are married with kids – but only 44% of women. Women who are married with kids have a 35% lower chance of getting a tenured faculty position than married men with kids. Even without children, chances for women are lower than for men. At the end of the road, women receive a 29% lower pension than men – two of the reasons being a later promotion and parental leave. Men’s pension, on the other hand, is not affected by having children. This is a prime example of a system that was designed for only one half of the population. Currently, two years of half-time employment is simply not equal to one year of full-time work. One hard measure for success in science is the number of publications. If someone published half as much per year, their chances of a tenured position decrease significantly – full stop. Does the system have to be like this? Definitely not.

I presented only three of Criado-Perez’s examples of how women are affected by the gender data gap. In the afterword of her book, she breaks down the plethora of problems to three points that describe the position of women in a male dominated world. Firstly, the invisibility of the female body. It is often ignored, that the female body is simply different from the male one. In addition to medical aspects mentioned above, there are also technical or architectonical aspects. Gender neutral security clothes don’t fit, the keyboard of a piano is too wide, or voice recognition just doesn’t work. Secondly, and ironically with respect to the first point, the visibility of the female gender. It is not the female sex, but the gender – the socially constructed aspect of being a woman – that leads to women being ignored, interrupted in discussions, harassed or even abused. Equal behaviour of men and women does not cause the same reaction. And most dramatically, sexual violence of men against women is a threat to women’s freedom and well-being, and is not sufficiently studied and included in the design of our world. Thirdly, women do most of the care work, without which our society would not function. This work is not sufficiently acknowledged or considered in shaping the world which restricts the possibilities of women and complicated their lives.

In her book, Criado-Perez presents a staggering amount of statistics revealing the underrepresentation of women to make a simple point: This is a men’s world. Women are disadvantaged and discriminated against, treated as a variation of the norm. But women’s issues are no minority’s issues – they are issues of 50% of the population. We have to start questioning the implicit assumption of masculinity, just as Denna did in Patrick Rothfuss’ novel “The Name of the Wind”:

“How could we possibly hurt it?” (the protagonist said, talking about a dragon)
“We lure her over the side of a cliff,” Denna said matter-of-factly.
“She?” I asked. “Why do you think it’s a she?” ​
“Why do you think it’s a he?” she replied.

 

[1] https://emojipedia.org/
[2] https://slate.com/human-interest/2013/06/female-academics-pay-a-heavy-baby-penalty.html

Did anybody ask for some unique and ground-breaking research?

By Sabrina Patsch, Universität Kassel and Freie Universität Berlin

Gendered language is on everyone’s lips. Gender-biased terminology (“all men human beings are created equal”), gender-neutral pronouns (“The author of the article said… and he they also stated that…”) or gendered nouns (“the chairman chairperson”) are highly discussed and ridiculed by many.

But there is more to say about language than that. Instead of talking about how to talk about people, let’s ask the question: How do people talk? And more importantly: Is there a difference in the usage of language between men and women*?

Many comedians or sit-coms built up on this and have established, or at least strengthened, gender stereotypes with respect to how men and women talk.  There is even a whole paper [1] elaborating on gender differences in using language which almost sounds like the record of a 90s TV series. Some examples:

  • Women tend to answer questions with a rising intonation instead of a falling one. “In this way, they can show their gentleness, and sometimes this intonation shows a lack of confidence.”
  • There is a specific set of feminine vocabulary that men do not use: they say blue instead of aquamarine or azure, or they call it a good meal instead of gorgeous or heavenly.
  • Women use more diminutives (e.g. hanky instead of handkerchief) or words that show affection (dearie, sweety). “If a man often uses these words, people will think that he may have psychological problem or he is not manly“.

I repeat: “psychological problem”. For using the word sweety. It should be mentioned that the few references in this paper range, with one exception from 2004, from 1968 to 1980. Besides this text ranging from entertaining to incredulous, one might ask the question what to take from that. Does it matter if I say blue or aquamarine? Does my intonation matter?

At least for the last point: Yes, it does. Imagine the situation where you sit at a table with your professor and the rest of the group. You present your research with the words “Well, I guess we made a bit of progress. The results look promising, but we need to double check. There are a few things I don’t understand yet but maybe it leads to something.”. On the other hand, your more “manly” co-worker explains his work like this: “I made great progress, I have new results which look amazing. There’s only a bit of polishing missing but I’m essentially done.”. You could be talking about the very same results and still, your professor will most likely be more excited about your co-worker’s progress. Confidence is the keyword. Selling your research well let it seem to be of higher quality – independently of how good the research actually is.

In an ideal situation, your professor knows you and might understand what you mean by what you say. In a less ideal situation, you and your research might be underrated by your professor and he or she might think that you do not know what you are doing. And what is more: In the most important situations, when asking for grants or publishing your results, you will write about your research and the readers do not know you at all.

Does the difference in language between male and female scientists pose a real, measurable problem? A very recent study asked exactly this question [2]. To answer it, the authors have estimated how positively researchers present their work by counting the number of “positive terms”- such as novel, excellent or unprecedented – in the title or abstract of a publication. They found that these words are used 12% less often in articles with female first and last author than in publication with a male first or last author. In high impact journals women were even 21% less likely to present their research positively. While this observation is not an issue per se, the authors of the study found an increase of up to 13% in subsequent citations – the unit for the quality of your work – with the usage of positive terms.

The caveat of this study is that it may include some conscious or unconscious bias against women. In a working paper from last April [3], researchers of the National Bureau of Economic Research tried to avoid this issue. They studied and compared the success of grant proposals from which all personal information on the author was erased. They found that female scientists were 16% less likely to achieve a high score on their proposal than men – even though the review process was blinded. How is that?

The researchers suggest that the word choice is the major cause of this finding. They discovered that men have a greater tendency towards “broad words”, i.e. words which are used in a broad subject area. Women, on the other hand, prefer “narrower words” which are very special to a narrow field. Beyond that, broad words often detected in high scoring proposals while lower scores were achieved when using narrow words. In other words: the quality of proposals written by female scientists is statistically perceived of lower quality – even if the quality is equally high.

The essence of this study is that there is one form of expression that is more likely to be successful than another. The successful way seems to be praising your research and promising the moon. Statistically, more men cultivate a successful mode of expression than women. But on a case-to-case basis, there are many down-to-earth men and as many head-in-the-clouds women. The current “best practice” of writing proposals and publications is disadvantaging many people – mostly, but not exclusively, women.

What is the take on that? Are women (or “female writers”) hampered by the way they are talking? Should we talk “more manly” in order to be more successful? Is all this simply the result of the self-selection of the man-dominated world science is right now? Or is this just the way science works?

There is, unfortunately, no simple answer to that. A small step is to be aware of the problem. And again, the question we must ask ourselves is:

Do we want to play the game or change the rules?

 

*Since none of the considered references took non-binary gender options into account I only talk about male and female language in this article.

[1] Xia. “Gender Differences in Using Language”. Theory and Practice in Language Studies 3(8), 2013, doi:10.4304/tpls.3.8.1485-1489 https://pdfs.semanticscholar.org/478d/8227a766d8d836664f52fec37b8b34c03491.pdf

[2] Lerchenmueller, Sorenson, Jena. „Gender differences in how scientists present the importance of their research: observational study”. BMJ 2019;367:l6573, doi: 10.1136/bmj.l6573 https://www.bmj.com/content/bmj/367/bmj.l6573.full.pdf

[3] Kolev, Fuentes-Medel, Murray. “Is Blinded Review Enough? How Gendered Outcomes Arise Even Under Anonymous Evaluation”. NBER Working Paper No. 25759, 2019 https://mitsloan.mit.edu/sites/default/files/2019-06/w25759.pdf