NCWIT 2012 Summit – Workshop, Dr. Aaron Kay

DR. AARON KAY: I’m gonna talk to you about some recent research that I was invited to discuss, that I think you’ll find relevant to the kinds of issues that you’re concerned with. So, I’m a psychologist. My PhD is in social psychology. I got that at Stanford University. But since then I’ve been working in both psychology departments and business schools, and now I’m joint appointed at Duke University both in the psychology department and the business school, because my research really is about, my research focuses on the psychological processes that lead social inequality to be perpetuated and maintained, and people to be uninterested in redressing social inequality, and so business schools find that issue interesting, even though it’s really actually ultimately a psychological question, or at least the way I address it is ultimately psychological. So, I’m gonna share some research with you, and I’m gonna start off with an observation that I think is not that surprising to most people here, and so in the business world at least upper management levels tend to be very male dominated, so these are actually just the first nine pictures that come up if you put into Google Image board of directors, and so [group laughs] what you can see is that there’s not very equitable distribution. There is one company on the right who has actually a perfectly equitable board. Anyone wanna guess what industry that is?



DR. AARON KAY: Health care, it’s a hospital. Yeah, so in general, you see a serious lack of gender equality. There is one board there of the first nine that come up in Google Image though that is equitable, and this issue obviously is not just relevant to upper management, so in the areas that you’re more interested in, information technology, we also see this large inequity in terms of, or inequality, in terms of the gender breakdown in different areas, and importantly these gender differences across different domains don’t seem arbitrary. In fact, they seem to fall in line generally with sort of the power structure in society, such that the most gender-dominated occupations that are male tend to be the highest paying occupations, and women tend to occupy the lower-paying occupations. So, we have an issue of fairness here, an issue of job opportunities, and an issue not just of people opting into different areas, but people actually being given different opportunities. Now, this problem of diversity obviously one, is a problem of fairness and a problem of social justice. But it’s also an issue with the ability of the technology industry to make the best possible products, so when diversity is lacking, you often get products that are missing certain perspectives. This was the case, you know, in December of last year, there was a lot of publicity about the fact that Siri was really good at finding things that males would ask for, like, you know, escorts and so forth, but couldn’t find any sort of female health option at all, whether with birth control or Planned Parenthood and so forth, and what people noticed about this, or what they remarked was that this wasn’t an issue of Apple disliking women at all. It was an issue of the fact that the people who were programming Siri were missing a certain perspective. They didn’t hate women. They just, these were a bunch of men, and when they visualize who they are programming for, they missed a certain perspective. And so, basically, across lots of different research, it’s very clear, there’s one very clear finding, which is that more diverse groups, and this includes gender in diversity, leads to more creativity, leads to higher collective intelligence, that is to better group decisions, and leads to more effective decisions. So more diverse groups are better, but there tends to be a problem in that even organizations who realize this, and this applies to the IT world, who want diverse groups, and who are interested in fostering diverse groups, still sometimes fail at it. And so, where my interest is in my research, I’m not interested in the CEOs out there, the organizations out there that have these antiquated versions of who is good at certain jobs, and so they have a lot of inequality because they’re just flat-out prejudiced. That’s not that interesting psychologically. To me, what’s interesting are the organizations where they say they value diversity, and they really do mean it, and you have young girls also who at some point wanna go in those areas, but despite both those factors, once you look at the higher-level jobs in that industry, you still see large gender inequality. And so I’m interested in sort of the forces that lead inequality to be maintained, despite the fact people have really good intentions for trying to, whether or not they’re based on fairness or based on effectiveness, trying to fix these problems. So why is there this disjoint between people’s intentions and recruitment efforts? Now, the way my laboratory addresses a lot of these questions is through something that we call system justification theory, but you don’t have to really understand the theory at all to understand what I’m going to talk about today, but the basic idea is that the status quo tends to perpetuate itself, and this is for various structural reasons in society and for also psychological reasons, once people understand what the status quo is like, they engage in processes that tend to reinforce it. And so as an example, if people learn about a specific demographic breakdown in an industry, they’re gonna start engaging in psychological processes that make that demographic seem like that’s the way it should in fact be. And so we have shown in a lot of different studies that this even applies to gender inequality, so when we learn about a gender inequality, people start to think, or come up with ways of viewing that inequality in such a way that it reinforces itself. So let me give you an example. We could take women undergraduate participants and show them this figure, and this figure is showing the percentage of female politicians, the percentage of politicians who are female. This is in Canada, actually, not in the U.S., and if we showed them this figure, the overall percentage is pretty small, it’s increasing, but not at a high rate, and compared to baseline conditions, and we show them this, and then we ask these women, would you like to be a politician. Do you think women should be politicians? Would they be good politicians? They tend to think that they don’t wanna be politicians, and that women actually wouldn’t make good politicians, and both men and women engage in that process, but we can also show, we also randomly show another set of women this figure here. So this is also representing the percentage of politicians who are female, but the change here looks a lot bigger over time. Now actually the numbers are identical in both these figures. All we’ve done is we’ve changed the vertical axis so it looks like there’s a bigger change, or we’re representing the real truth in both situations. And when they see this, they think, wow, there’s a lot of women entering politics now, and compared to baseline conditions, they actually say they’re more interested in politics, and they also say that women will probably be better politicians than if they’d see for example the other figure or just compared to control conditions. So this is in the context of politics, of course, but it’s one example of the way in which once we learn about the status quo, we start making judgments and seeing the world in a way that reinforces that status quo. Now, in the real world, there can be a lot of things that tell us what the status quo is like that are actually very subtle, that we don’t realize that we’re communicating. And what I wanna talk about today is the extent to which in the job ads that we write and that are out there, people sometimes unintentionally communicate an inequality that actually exists in an industry, even if they’re not trying to do it, and that then can lead to that inequality being reinforced. So I’m first gonna start actually giving you some other examples that aren’t from my research on how, within the information technology world, reminders of minority status, whether they’re other people, or they’re stereotypic environments, can actually make people feel like they don’t belong, and this has effects on who enters these areas, and I’m gonna do that mostly just to convince you that in the really blatant ways, these things actually matter, and then I’m gonna shift gears, and for most of the talk, I’m gonna focus on recent research that I was asked to actually come here and talk about, which is how wording in job ads can systematically convey to people even if the authors aren’t aware of it, what kind of gender tends to fit what kind of job. Okay, so when people are looking to apply to work somewhere, they’ll often observe, whether consciously or not, they observe for cues that suggest what kind of person typically works in this environment. In one study, what psychologists did is, this is actually with African Americans. They were looking at African Americans being the minority and not women, but they, at a job fair, had African Americans look at brochures for a company they thought was real, but was actually fictitious. On that brochure they described their philosophy toward diversity, and they describe one of color blindness, which is the idea that we don’t see color in our employees. We just see people for what they do, and that’s all, and then that was paired with pictures of the company, and for half of our participants they saw these pictures, and for the other half they saw these pictures. So the color, everything’s kind of faded in here, but the difference in these two conditions is that in one you have an ethnically diverse group of people, and the other people are seeing basically all white employees, and what happened is that if you learned that this company adopts this colorblind philosophy, but you see these pictures of just straight white people, then African Americans are less interested in working there, and feel like they belong less at that organization then if that colorblind philosophy wasn’t actually espoused at all. If instead you have a philosophy of what they call embracing diversity, where we see everyone as different, but we want these differences, it doesn’t matter so much what kind of cues you give to people visually. This is one example of how these cues can be really explicit, they can be in your brochures. Here’s another example where they have math students. These are in university, university math students learn about a conference on STEM leadership, basically, so the conference is going to be about leading in science, engineering, and math, and for half of them, they all watched a video about last year’s conference, and these were male and female math university students. Half of them, in the video that they watched, there was a three to one ratio of men to women, so they think they’re just watching a video about last year’s conference, but embedded in this video is the fact that for every one woman there’s three men at that conference last year. You’re just noticing it as people were on the screen, and the other condition there’s a one to one ratio of men and women. These pictures aren’t from the actual video. And what they measured after was the extent to which people wanted to attend this conference, felt like they would fit in at this conference, and they also mentioned their physiological morale, the extent to which their heart rate and skin conduction suggests that they felt threatened about who they are. And what they found is that when women saw the three-to-one ratio in this video, they didn’t wanna attend the conference. They felt like it wouldn’t be interesting to them, and they also exhibited these signs of physiological threat, whereas they didn’t show any of this in the one-to-one condition, and men were unmoved by whether or not there was a big gender imbalance or there wasn’t a gender imbalance in the video they watched about this conference. You also get these kinds of effects is not just with other people, but with what sits in your environment, so for example, in studies they’ve had people go into computer science classrooms at Stanford in the Gates Building, which is the computer science building, and they’ve asked, people thought they were doing things on their career aspirations, and in that room, the walls were either adorned with things that pretesting showed were related to the idea that sort of the male geek nerd computer stereotype, or not related to that, so for half of them the walls had things like posters having to do with science fiction, comic books, and video games. For the other half, they were just art posters and neutral images, and then they asked these people if they were interested in computer science or not. And what they find, if you look at that one bar there really is that women become less interested in computer science as a field if the room has images to suggest that it’s a male type environment, so the big deficit there is women in geek poster condition are showing a decreased interest in pursuing computer science. They also get the same effect, so a lot of classrooms now are moving virtual, literally to the extent which you take your classes in a 3D environment which looks like an actual class, and you get the same effect if in these 3D environments posters are of science fiction type related ideas, so the point of telling you about those studies is just trying to convince you that cues to suggest who does and doesn’t belong actually affect people’s interest in a career, but I think that these kinds of cues are very obvious to people. If I said to you, what would make someone think they belong, you might say something like well if there’s more men than women there, then I don’t belong. I can count, and I can figure out where I do and I don’t belong. We can easily remove things like posters from the walls. What I’m interested in, and what I understand wording in job advertisements matters a lot is because compared to these other things, it’s basically invisible. So, people, unless they’re trained to understand what gender wording means in an ad aren’t aware of the fact that their ads are written that way, and people who are being affected by them aren’t really aware that’s why they like a job or do not like it, or that’s one factor that leads to them liking a job or not liking a job. So I’m gonna transition now and talk about the research on this idea of gender stereotypes being reflected in job ads. Now we’ll start with a bit of history. And at first this gender bias in ads was not very visible at all, so this is the New York Times in 1964, and what you’ll see here in a few different ways ads were very explicitly sex segregated, so first you could see that the New York Times itself categorized ads as a function of whether they were in the male category or the female category. That’s independent of what advertisers we’re picking. You’ll also see the jobs the advertisers had a male or female name to it, so salesman, for example, rather than sales person, and then within the ad itself, they would say, we want men or we want women, and they would use pronouns he or she. Now amazingly, once people started becoming concerned this was actually causing some sort of inequality, both the newspaper and the companies insisted that although they put these words in there, they’re just for convenience, and they’re confident that women are not being deterred from any sort of job based on any of this language. So they insisted that this wasn’t having any actual effect. Everyone understands, it’s for convenience, and they can “look past” these convenience-type labels that we put on there, so psychologists at the time disagreed. And they ran studies to show this wasn’t the case, and so for example, this is they found advertisements like this one that were targeted for males, where you know this was for a telephone frame man. “The telephone frame man plays a vital role “in telecommunications,” “skilled craftsman,” and “he also works,” etc., all these male terms, and they also found feminine ones, female ones. These are real ads from the 60s. [mumbles] Behind every man’s telephone call there’s a woman, so this ad’s targeted directly towards women. And they show that if they took these labels off and replaced them with gender-neutral terms like person, frame person, for example, that women became much more interested in the male jobs, and males became more interested in the female jobs, so they were able to prove that AT and T, for example, who there was a law suit against at this time, despite their claim that this wasn’t causing any bias, was in fact causing a bias. It also became the case then that the newspaper said that well, we might need to change the wording in the ads, but we can still keep our categories, because it’s still more convenient for people to be able to search by gender, and so they said what we’ll do is we’ll put this note at the top of the categories that will tell people that jobs are arranged under male and female classifications for the convenience of our readers, but this is done because most jobs generally appeal to more persons of one sex than the other, but you should know that you can apply for any job at all, so the psychologists went back to the lab and said, okay, we’ll put this thing right on the top, and then we’ll just sort our jobs as a function of male and female. Are people able to disregard this? And no they weren’t, so eventually, by the 70s, it became legislated, unconstitutional to have any of these kinds of advertisements. I think it’s interesting that, by the way, in these advertisements, it always said at the bottom, an equal-opportunity employer. So they used the term male or female, but at the bottom they put this equal-opportunity employer classification, and so what happened is eventually these all disappeared, so there was no help wanted male category. There were no gender-specific terms in ads, and for 40 years, from the 60s or 70s through maybe five or six years ago, the assumption was okay, this isn’t a problem anymore. We don’t have to worry about job ads communicating preference for one gender or another. But what my lab started to ask was maybe, although these explicit references to men and women, or he and she have disappeared, maybe these implications are still in there. They’ve just sort of gone underground, so to speak. Specifically what we were interested in was whether or not there is a systematic difference in the type of wording used for occupations that would tend to be male dominated or female dominated, that without saying men or women, use language that was either more male or female, that people were actually able to pick up on and realize that this was a job for a male or a job for a female. Okay, so this is a bit of history. This is where we are today, so what we’ve developed, is this idea of what we call gendered wording, and gendered wording is basically phrasing or vocabulary that subtly conveys gender stereotypes, both masculine and feminine. And so there’s a big literature on things, on words or traits that are associated with men, and words or traits that are associated with women, and you can look across this literature, and you can make lists of words that basically people think apply more to men or apply more to women. And so we essentially did that, and so if you look at words that fit the category of masculine wording, you get these types of words. You can look at them quickly. I can always send these lists to you later if you wanna see them. There’s also lots of words that fit the feminine stereotype. And not all these words you’re gonna find in job ads, but they’re words, and they’re both negative and positive, but the literature suggests these are the kinds of things that people definitely associate with one gender or the other, are you looking at whiny?

AUDIENCE MEMBER: [laughs] Yeah, I see that.

DR. AARON KAY: A lot of people look at whiny in there, and find that… What’s the other one?


DR. AARON KAY: Nag. There’s positive and negative in both, and like on the male you have stubborn.


DR. AARON KAY: And hostile, and these aren’t words that you ever would find in a job ad, but these are words that are associated with the genders differentially. So what we asked, and this is what I’m gonna tell you about now, is three questions, really, or it’s really two questions that I turned into three bullet points. Does gendered wording exist in real ads? So, is it the case that male-dominated areas for example have more male-type words in them than female-dominated professions, and if so, does this gendered wording actually cause men and women to be differentially interested in these jobs? So just like you putting the word man there or woman there, or he or she, has it now moved to just be implied through words that are associated with men and associated with women? And so men and women actually pick up on these differences? So this is what our research question asked. By the way, I should let you know that feel free to ask questions at any time. I’ve left off a time at the end of the talk for question and answer, but I’m happy to have that happen throughout the talk, or if you have any thoughts or questions, just interrupt me at any time. So, in the first test of this idea, basically what we wanted to see is is it the case that there are systematic differences in the types of words used in advertisements as a function if they’re male or female occupations? And so we looked at 11 different occupations that were clearly differentiated, in the extent to which they are dominated by men or women, and so this just shows you, we have from very strong male-dominated areas like plumbing, so you feel like in IT you have a problem. [audience laughs] Plumbing, you’re really missing the female perspective in plumbing much more intensely, and so we have six male-dominated areas, and five female-dominated areas, and what we did for this first study is we took the first, this is very labor intensive, because for this we found the ads on things like, and so forth, you know, online recruitment sites, and we had to code every single one individually and through this software that we have, and so we took the first 50 or so that came up on a search for each of these professions, and what we did with them is we ran them though a software that was built to sort of code for the percentage of different words used, and it had these dictionaries, and they can tell us the extent to which these ads had more male words or female words, as a function of the list that we’d created based on past research. So here’s an example of what a masculinely-worded ad would look like in computer science, and then I’ll show you a less masculinely worded computer science ad. Does anyone see the words in here that connote the masculine stereotype?


DR. AARON KAY: Individual, yeah.

AUDIENCE MEMBER: Self starter.

DR. AARON KAY: Anything else?


DR. AARON KAY: The very last sentence has a very common one. We already said individual.


DR. AARON KAY: Analyze, right. So analyze is also very common, so there’s individual in there twice, and analyze. Self starter actually does capture the male stereotype, but I don’t think it’s in our dictionary. We have independent and so forth, but it should be in the dictionary. Then you can see a computer science ad that was very high in the feminine wording, so if you look at the feminine wording, you have things like this, provide, support, service, understanding, and so often one of the big differences in male and female ads is that they’ll ask for the same skill, the ability to analyze something or the ability to understand something. Which really is getting at the exact same idea, the ability to lead versus support, or we want someone who’s aggressive versus someone who’s persistent and vigorous, so they’re often asking for very similar skill, but just doing it through language change that really wouldn’t have any consequence in terms of what kind of person or what they’d expect to do in the job, but actually are differences in terms of whether they convey masculinity or femininity. So, this is an example of a femininely worded ad. These are just two examples I pulled out to show you. If you then take all the advertisements and collapse them together, what you have is the blue is stereotypically male occupations. The red is stereotypically female occupations, and we can look at the percentage of masculine wording, feminine wording in each, so what we saw in this first study was masculine wording is much more likely in stereotypically male occupations than female occupations, whereas feminine wording isn’t actually more prevalent in female occupations than male occupations. So you get this difference in the extent to which these types of words are used in advertisements, and this effect holds within each, what’s really interesting is so if I go back to this, sorry, this slide here, you can see that some of the male-dominated areas are more male dominated, so plumber, then electrician, then mechanic, then engineer, then security guard, then computer programmer. And the percentage of the imbalance between male stereotypic words and female actually follows that trajectory, so the more males are in that area, the bigger the imbalance is. You can even, you don’t have to collapse them into two big groups. You can actually look at the correlations, and there’s a strong, and it actually tends to nicely follow the demographic pattern in the actual world. So let me just skip that, and okay. Before I got to Duke, I was an assistant professor at the University of Waterloo, and U of Waterloo is notable in a few ways, so this is in Canada. First of all, it’s like supposed to be sort of the MIT of Canada. It’s just focused on computer science and engineering. That’s one aspect, and the other aspect is that it has North America’s largest coop program. The students spend half a semester in college and half a semester at a job, and half a semester back in classes, and half a semester in a job, so companies actually engineering and computer science companies from everywhere, Reem being one of the large ones ’cause they’re in Waterloo, they recruit right out of the university and they recruit students who are currently in university, and what they do is they send their advertisements and target them to different majors, so as a company I might say, I wanna send this ad, I wanna get a coop student from Waterloo. I’m gonna send this ad to engineering and computer science, or I’m gonna send it to environment and applied math, and so the coop group at Waterloo gave us access to all of their ads, which allowed us to do much larger tests of this idea, with college-age people looking for jobs, than we were able to do just with our online ads, and so what we did here is we categorized people’s target ads, those targeted to the disciplines at Waterloo that were male dominated, engineering, math, computer science, science, physics, economics, and accounting and financial management, and then the ads that were given to people who were targeted to people in the more female-dominated areas, which were applied health, arts and environment. Waterloo as a computer science school doesn’t have things like history and English lit and so forth. And then we just looked to see, and now we had 4,000 ads, rather than 500 ads, are we seeing the same bias in language usage with these ads, and will we get it to replicate the same effect? So, masculine wording is more common in stereotypically male areas than female areas. Feminine wording doesn’t actually vary as a function of what type of domain it is. So this first piece is just that yes it does seem to be the case that systematically there tends to be more advertisements for occupations that are traditionally male dominated like IT, tend to have more words to suggest that fit the male stereotype than have words that fit the female stereotype. Now the question is does this actually make any difference? Do people notice it, and does it affect how much they like these jobs? In the 60s, when they wanted to argue that putting the word frame man versus frame person was having an effect, they had to bring it to the laboratory and show that if they take it out or put it back, men versus women are more or less interested in that job, and so we basically did the same thing. So what we did is we created ads based on the ads that we’d already analyzed, but the beauty of creating ads is that we could keep the job identical in every single way, expect for we could swap out masculine and feminine terms at a few spots in the advertisement, and you could see if people’s interest in a job that has the same pay, the same basic jobs, is in the same place and so forth, changes as a function just of whether or not there’s male or female words in there, and we could see if it differentially affects men and women. So, we did this. We built this ads in six different areas, so we had two in male-dominated areas. We made a male and female version of an engineering job, a male and a female stereotyped word version of a plumbing job. Same with two female-dominated areas, registered nurse and administrative assistant, and then also in gender-neutral areas, real estate agent and retail sales manager, so retail sales is actually very feminine dominated, but retail sales manager is actually pretty gender neutral. So, each participant in our study would see six job ads, and they would either, they would see one from every category, and they would be randomly assigned either to see the job ad that has the male-type words or the female-type words. And we look at seeing if that made any difference in men and women’s interest in these various occupations, so here’s an example of our engineering ad. It had lots of information, but here are the only things that were changed, okay? So, in the more feminine version of the ad, they are told that we are a community of engineers, so the word community, with effective relationships, the word relationship, with many satisfied clients, and in the masculine version, we’re a dominant engineering firm that boasts many leading clients. The feminine version also adds, we’re committed to understanding the engineering sector intimately so you have understand and intimate, versus we’re determined to analyze the competition, so determine, analyze, and competition are very male words, compared to understand and intimate. And then we also have sensitive to clients’ needs, and develop warm client relationships, versus superior ability to satisfy customers and manage company associations with them. So, basically these are all just a bunch of words that aren’t really gonna change, you would think, what kind of job, what you’re actually gonna do on the job, but in one condition, these were embedded, more feminine wording is embedded into the larger ad, in the other more masculine wording, and then after there’s the question, do men and women actually have different interests in these different jobs? So, after each job advertisement, either the male or female version, they rate the extent to which they find it appealing, six different questions on a one to seven scale. And then because we think if there is an effect on their appeal, it’s gonna be driven by the extent to which they feel that they belong there, to which they fit in there. We also asked them the extent to which they anticipate they would belong at that particular place. So, I would fit in well with this company. I am similar to people who work there. My values and this company’s values are similar, etc. Okay, so what we have in this figure, and I like to walk up to the figure usually but apparently I’m not allowed to leave the proximity of this microphone, so I’m just gonna point. We have female and male participants who look at either the masculinely worded advertisement or the femininely worded advertisement for the same jobs, and this is looking at how much appeal they have for the job, so what you see for female participants is that, when the exact same job has the feminine words in there rather than the masculine, they’re more interested. Something interesting to note here is that if that doesn’t look big, it’s significant. But four is the midpoint of the scale. Four basically means neutral, so what’s happening is people are going from neutral on these jobs to moving to the positive side. They’re now, all else equal, positively inclined to work at that job, and if you look at male participants, you see the opposite. They are more interested in these jobs when they have these masculinely worded pieces into it rather than the femininely worded pieces. Now, in the real world, like we’ve shown you, it doesn’t tend to be the case that there ever is fewer male adjectives in an a description than female adjectives, even in femininely dominated occupations, so men would be sensitive to these cues, but they don’t ever have to actually contend with it, but women we’re seeing, not only are these cues out there in the world. When we actually embed them in actual job ads and they say how interested they are, we’re seeing an effect. I should also note that this applies for every single job on our list, so for plumber and for nurse, if we take a plumber ad, and we use more female adjectives in describing that job or what work gets done, women become more interested in becoming a plumber, just like they become more interested in becoming a nurse, or an engineer. It doesn’t seem to depend on what area it is. We can actually make women more interested in becoming a plumber, just by making these little changes to the wording. If we look at anticipated belongingness, you see the same thing. Females feel like they’re gonna belong more there, they’re gonna fit in more, if there are the feminine words compared to the masculine words, and males feel like they’re gonna belong more if we use the masculine wording compared to the feminine wording. And then what you can do in a nuanced statistical analysis, that I don’t have to give you any details about, what we can show is that, so you find that being a female participant makes you less interested in jobs if they’re masculinely worded, and that effect is entirely mediated or driven by their feelings of belongingness, so the reason they have less appeal for the job is because they feel like they’re not gonna fit in, they’re not gonna belong there. So basically what this study has shown, and we have many that replicate this. I’m only gonna show you a couple, is that these subtle wording differences make people feel like they will belong there more or less, and that leads them to claim that they are more or less interested in the job, and what’s important here is that when we ask people why they’re interested in the job afterwards, no one ever says anything like, well, the words used to describe it sort of reflected a male stereotype, and so I feel like I’m not gonna belong there, and so therefore I don’t have any interest in it. No one ever says something like that. They say things like, that feels not really for me. It doesn’t look stimulating. It doesn’t look interesting. They come up with rational reasons, and no one ever notes that they’re being driven to some extent just by these arbitrary changes in the wording. One thing that I, I wanna show you one replication of this, because I think it eliminates an interesting possibility, which is that some have said, so I’m arguing that if we change analyze to understand, for example, in two ads, a male ad and a female ad, they haven’t really changed the nature of the job at all or what they have to do. You’ve just used male versus female wording, or if we change aggressive to persistent and vigorous, same idea, but someone could say, no, you’ve actually changed the skills that are required, and now people, women maybe feel like they’re actually going to be worse at the job, so we looked at this same idea. We basically replicated it where while also asking, do you think you’ll be any good at this job? And what we find is you know we replicate our, these are just female participants, they think they’re gonna like the job more and belong more, in the femininely worded ad, but when we ask them how good they’ll be, and if they have the skill, there’s no effect. So it’s not that the women are saying, I can do that better, or you know, I can analyze but I can’t understand, it’s that those words are just having, are exerting a feeling on if they feel like they’re gonna fit in there, be surrounded by other similar people. Okay, so if you look at the psychological model in a broader sense, what you’re seeing is that masculine wording affects job appeal through feelings of belonging there, not through feelings of I can do it or not. Women feel like they can do it in both conditions, it’s just one they feel like they’re not gonna fit in very well. So, just to summarize so far what we’ve seen, the obvious environmental cues like the graphics, a bunch of men in the room, or posters on the wall, can affect the extent to which men versus women feel like they are gonna fit in at a job, or would be interested in that job. More subtle things like changing the wording of a job ad also have this effect, and what I wanna suggest now is that there’s one other part of job ads that also exerts a similar effect. This one is not invisible, but I think it’s very counterintuitive. So, how many of you actually do hiring? Some people, so do you ever put these things in the bottom of your advertisements? Of your job ads?

AUDIENCE MEMBER: It’s required.

DR. AARON KAY: It’s required for some jobs, right? Now, there could be, basically I’m ignoring the fact that the current status quo triggers people where they will and will not belong and it has a big effect on what they feel like they’re interested in. If you put this on the bottom of your job ad, you could imagine two ways in which this might affect people. So, one possibility is that it suggests to people that wow, this place cares about women. They don’t actually just think it’s legal. They think this people care about diversity, and I’m wanted there, so I’m gonna be more likely to apply and feel like I belong. But what’s the alternative hypothesis? What else could happen?

AUDIENCE MEMBER: The implication could be they’re putting that there because they’re lacking in that.

DR. AARON KAY: Right, so if you think about it from the perspective of people sort of rationalizing whatever currently is the case, this might suggest to people that this is there because they don’t have any women or minorities, and as it stands right now, I’m not gonna fit in very well at that company. So the question we ask was, does adding these employment notices actually make women feel like they belong more? Or does it ironically deter them from wanting to work there? So, we tested this in the lab, and we basically gave men and women, lots of them, an engineering ad, and we either just added that equity notice that you just saw on the bottom, and they thought it was a real ad, and we asked them about their interest in the job. What you see is that for female participants, they have more interest in that job when that notice is absent compared to when it’s present. Would anyone have a guess what happens with males if that notice is absent or present?

AUDIENCE MEMBER: It goes up when it’s absent.

DR. AARON KAY: It goes up when it’s absent? That would’ve been neat. They don’t really care. [audience laughs] They don’t really care. They’re not moved by it one way or the other. I’ll show you in a subsequent study when they actually do care, but basically females are negatively impacted by the presence of that ad, and again this effect is working through feelings of belongingness, so when the ad is there, women feel like they’re not gonna belong at that position, which in some ways is sort of a rational inference to make. If this company needs to start hiring women and minorities, there must not be very many women and minorities there right now, so I’m not really gonna belong, and I’m gonna show up right now. I’m not gonna show up there in 10 years, when they’re affirmative action has actually yielded some sort of actual change in the demographics. And males aren’t really, again, aren’t affected by the presence or the absence of the notice. So, I don’t wanna imply that I’m against affirmative action, ’cause I’m very pro affirmative action. I think what people have to be careful of is the signal that affirmative action riders basically on these ads send, and how you might wanna write them, so maybe if instead of saying something like, you know, we’re committed to hiring women, something like we’re committed to maintaining our diversity, or we’re committed to maintaining the trajectory we’re on, would be more effective. We start to look at more nuanced ways of framing these employment, the affirmative action equity notices, so within the affirmative action world, there’s sort of two debates on how people are gonna react to affirmative action. Some people think that affirmative action means what’s on the top there. So our company is an equal opportunity employer. We have an affirmative action policy, and we’re committed to broadening the overall talent pool by actively seeking female and minority employees, so in all cases we give primary consideration to women, basically, versus imagine an affirmative action policy which there are some that say this instead. In support of achieving our goal of diversity in the workforce, our selection is gonna be limited to people of these underrepresented groups. So you can have one which is that we’re giving priority to underrepresented groups, or one that just says that we’re hiring only underrepresented groups, and do these actually have any differential effect on men and women, so we ran the study. And so we have three different categories. We have the notice is absent entirely, primary consideration to women, or selection limited to women, and we have belongingness and appeal for women and male participants, so anyone wanna guess what happened, you know, what the comparative data is gonna look like for females? In which case are they gonna be most interested in the job?

AUDIENCE MEMBER: Notice absent.

DR. AARON KAY: Notice absent. Is there gonna be any difference between primary consideration or selection limited to women?


AUDIENCE MEMBER: Ah, sorry. [audience laughs]

DR. AARON KAY: The answer is no. So women like the job less, or they like the job the most when the notice is absent. They like it less when there’s any sort of equity, affirmative action rider at the bottom of it. What about men? Any guess for men? And this time we do get an effect, so anybody wanna guess which of these men like the most, or the least?

AUDIENCE MEMBER: Well, I’d hope they would dislike selection limited.


AUDIENCE MEMBER: I hope they can read that. [audience laughs]

DR. AARON KAY: Are they rational at all? Yeah, so that’s what you see. Men don’t care if the notice is absent or if there’s gonna be some policy that gives primary consideration to women. They feel like they’re gonna fit in just fine. They don’t like to work at that place, though, if they’re selection is actually just limited to women. So the only time men are affected is when there’s a very aggressive affirmative action policy. Women, though, any sort of affirmative action policy, even though that’s actually supposed to benefit them, is making them less interested, and it’s because, we think, we’re still working on it. This set of data that I’m showing you now is just brand new. We haven’t published it yet. It suggests to them that right now they don’t fit in very much. Okay, so what you’ve seen then throughout the talk is that there are factors that reinforce the status quo because they suggest a current inequality in the demographics. This can be stereotypic environments, masculine wording in ads, or cues of current diversity problems, which include things like riders about equity, about affirmative action attempts at different organizations. These make women feel like they’re gonna belong less at that job. They anticipate less belongingness, and because they anticipate less belongingness, they come up with reasons why they actually don’t want that job. It’s not interesting to them and so forth. So implications. Gender wording is a problem obviously because of fairness. It prevents a level playing field. It’s obviously a problem in terms of creating the best possible products out there, but I think that gendered wording in particular is very problematic because it’s so subtle and difficult to notice, and this applies both to the people who are reading the ads, to the women or members of minorities reading the ads, and also to people writing the ads. People who are writing these ads, they’re not putting these words in there, I don’t think they’re putting these words in there, because they’re trying to cue what kind of gender fits in here. These are messages of the language used around these kinds of environments over many, many years, and so no one is really aware of these problems. What’s so nice about it, though, is that, there’s nothing nice about it per say. The good news is that, compared to trying to change the actual demographic makeup of a company, changing the wording in job ads is actually a very simple intervention, so whereas you need to get a lot of buy in from your CEO if you say, okay, I want an aggressive affirmative action policy, where we’re gonna change the numbers at our company, you don’t need a lot of buy in for them to change something that they probably think isn’t gonna make any difference anyways. You say, I wanna change every time we use the word analyze to use the word understand, in our job ads for example, or I wanna change lead to be about support, so it’s a change that can be enacted relatively easily, even though it’s one that I think is particularly insidious, because people don’t notice it. They think that as long as we’re not saying he, or men in our ad, it seems really open to both genders. And so I think, I prefer the term careless language use, rather than sort of malevolent or sexist language use, because I don’t think people are trying to discourage a specific gender. It’s that they don’t actually pay any attention to these kinds of male versus female words, and this can cause them to miss out basically on a lot of the best applicants for their jobs. And so the question is, can we write better ads. This morning in the Entrepreneurial Alliance, we discussed with everyone who brought their job ads, ways in which we could change their ads. There’s lots of things you can do, and I said to them that I’d be more than happy if anyone wanted to discuss their specific ad their company uses to run them through our program, so we have this program that’ll tell you the percentage of words that are male versus female, and we’d be happy to give feedback, and so forth, and we’re looking into putting something on the NCWIT website that shows just easy things that you can change in your ads and not really change the content and so forth, but that’s where we stand. So, would be happy to open it up for discussion, or any thoughts you might have, or anything like that, ’cause we do have time, right?

AUDIENCE MEMBER: Mm-hmm. 15 minutes. [audience applauds] Yes.

AUDIENCE MEMBER: You may have said this, but when you’ve done these specific studies, how many ads are you pulling in? What’s your…

DR. AARON KAY: So, in the second study I showed you, it was 5,000, every single ad that came to our coop pool of applicants. In the first one, it was 50 per occupation. And randomly, so the first 50 that would come up when you put a search term into the job search website. Yes.

AUDIENCE MEMBER: I remember you saying you used online job ads, and I’m curious if you took into account the nature of the online site. So for instance Dice, being so technical is more male driven than something like Monster, and if that was factored in.

DR. AARON KAY: Well, no. Do you have any thoughts on how it could make a difference? No.

AUDIENCE MEMBER: I don’t know. I mean I know just from personal experience that I feel like when I go to a site as a technical woman, like Dice, it just feels more heady and tech and male based versus Monster is a little more…

DR. AARON KAY: Even if you find that an ad for the same industry on Monster.


DR. AARON KAY: Yeah, that’s interesting.

AUDIENCE MEMBER: And it’s more of the high-level environment of that website. It’s not even the wording of that specific ad. It could be the same ad on both sites, word for word, but it’s something about…

DR. AARON KAY: And a lot of these job ads now people are just finding on the employer’s website, right? So you can go to any of the websites of the companies attached to NCWIT and they have their job ads listed right there. You know, you can look at those employment opportunities, and when I look at them I cringe, because they fail this test in almost all ways, and so, I think that even regardless of what website you find, the ones they post internally on websites are the same, and this effect’s really robust, so we’ve started looking at it with university mission statements, for example, going through our software, and you can sort universities as a function of ranking, for example, or private versus public, and we find that, now we’re not looking at gender differences, but SES, so how people are sort of wealthier, or wealth perpetuated, and we find that the higher the rank of the university, the more male adjectives there are in their mission statements, and also private universities have more male language than public universities in their recruiting statements. So you see this across all kinds of areas. Yeah.

AUDIENCE MEMBER: This is kind of a follow up to that, but first a comment. I’m a historian. I write about the history of computer programming, and particularly the kind of period in the late 50s and 60s where it becomes gendered male, so it’s actually fairly gender neutral or feminized prior to that, and job advertisements are a big part of what I look at. I can’t do that empirical stuff, kind of research you’re doing, but you do see different kinds of languages and visual tropes emerge.

DR. AARON KAY: Over time, yeah.

AUDIENCE MEMBER: And clearly shifting the perception of who belongs in this field. That was just a comment. The question is about other associations that those words have, so for example, one of the arguments historically about professionalization is that it’s also masculinization, so as a field achieves greater status, it also becomes male, so in your university case, have you done something where you’ve taken words like analyze and not tracked it via gender, or relative to gender associations, but status associations, right? So is Harvard using this language, and achieving as high rank, and becoming masculinized, but are they really thinking about status rather than gender?

DR. AARON KAY: Right, yeah, I mean so a lot of the research shows that stereotypes associated with high SES and low SES status, economic status, match on quite well to stereotypes associated with masculine versus feminine. And so it’s not necessarily that one is masculine and one is feminine, one is status. It’s just the same, so people in high status are thought of as individualistic and driven and achievement oriented, aggressive, and low status people are thought of as communal and group oriented, and worried more about interpersonal, interdependent things, and so they just tend to covary, but your profession, it’s an interesting idea, so this idea, as an industry becomes more feminine, it becomes less professional, which is a typical argument that’s happening with medicine now, so as medicine’s becoming more equal in terms of women, it’s actually becoming slightly imbalanced with more women, it’s becoming seen as less prestigious is the argument, and it’d be interesting to see if over time it also happens that the description of that job becomes from one of deep level individual analysis to one of care and independence and so forth.

AUDIENCE MEMBER: Yeah, and so is that a source of resistance if you go to your CEO for example and say we wanna change the language of this ad from analyze to understand, and here’s the reason that has to do with gender diversity, does that provoke resistance, to say well, but understand is less…

DR. AARON KAY: Yeah, I wonder, and you know…

AUDIENCE MEMBER: Effective than understand.

DR. AARON KAY: Right, and I don’t know the answer. Maybe you could tell me, if you came to your CEOs and you said, I wanna change my ad just like this, would they resist it or not? I really don’t know, or if they want to change the lead to support, for example. You know, I teach leadership at Duke School of Business, and we teach that leadership really is just supporting from above. I mean that’s all it basically is, but will there be resistance, because they’re afraid that they were gonna get a wimpier type of candidate? That’s a good question. I don’t know if anyone has any idea about that.

AUDIENCE MEMBER: That’s definitely a concern of mine, and I was sharing with you this morning that even though we have a woman CEO, I work with primarily male PhD geoscientists and geologists. So I am concerned that I’m going to get some pushback and resistance to that.

DR. AARON KAY: Right. I mean the one nice part is that if you have a company that has the intention, they understand that a diverse workforce is a more effective workforce, if they want it but they can’t achieve it, they’re more likely to want to make these small little steps, but if you have one where they’ve been doing it only because it suggests that they’re a fair, equitable place, it might be harder, even though I’d be very convinced that if you changed analyze to understand, throughout all your ads, your applicant pool is not gonna get any worse in terms of talent. I mean, it doesn’t seem feasible to think people are thinking about it at that level. But it is, it’s an interesting question. Will there be resistance just because at some level they feel the association with feminine, and there’s a fear of that or something. Yeah.

AUDIENCE MEMBER: I have to say that even though I’m a woman, if I was reading an ad which said support, I would be less inclined to apply, than I would be if I saw direct, lead, and maybe it’s because, as a woman in IT, I have associated those kinds of words, maybe I’ve associated the field with I have to be more like males, more leadership and strong and those kinds of things.

DR. AARON KAY: Yeah, it’s interesting. I got some of that feedback this morning, so clearly when we run these studies, with women who are college age, who are the ones who we ultimately care about, ’cause they’re the ones who are making these decisions, right now. It’s not, your intuition isn’t the case with them, so they prefer it when we replace lead with support, and don’t forget we’re not saying we want someone to support others. The idea is that we’re saying something like, you’re just embedding the word support in there, like, we need to support other customers’ infrastructure, right? So it’s just choosing to embed that word, not saying we want a supportive person, using it as a trait word, but just embedding it through the ad, but maybe it’s the case that if you look at women now who’ve been in IT for awhile, they are acting somewhat similarly to the male participants, but if you look at the women who aren’t there yet, they are resonating more with the female language compared to the male language, but your comment is interesting, ’cause I heard that this morning too, from some people. Yes.

AUDIENCE MEMBER: It’s unbelievably striking to me that the word that’s coming out in this is analyze, ’cause that’s my work, statistics and math, and in fact I’ve just written a piece pointing out that women’s participation in statistics and mathematics is equal to men’s, and it’s the highest level of participation in any of the STEM fields, so it’s remarkable to me that this particular word comes out. I’m very interested in the kind of work that he mentioned over here, and have been reading about that, but then I back up and I think, all right, you got that list of feminine words, masculine words. Did that come out purely with regard to your research, or was that somehow more fundamental?

DR. AARON KAY: The list?

AUDIENCE MEMBER: I mean, ’cause you said they weren’t necessarily words that showed up in job ads. Is that?

DR. AARON KAY: That list was generated not by my research at all, but by going through past research that has showed that there are traits that are clearly associated with one gender or the other, and we just put in that list only the ones there that there’s really good data already to suggest it’s either a male word or a female word, and we didn’t pay any attention to whether or not it was relevant to the work force, you know, to job ads at all. We just generate these two lists, and then run them through the software, yeah.

AUDIENCE MEMBER: And did any of that research ever go beyond English? Is that an English language, or an American thing? Or is it somehow broader?

DR. AARON KAY: Well, certainly, I can’t speak to words, but the general stereotypes of women being more collective, interdependent, communal, and men being more agentic, individualistic, achievement oriented, they tend to exist pretty much everywhere where there’s a hierarchy in which men are more powerful than women. There’s a few small little cultures, pockets you can find of small cultures where there isn’t that hierarchy, that traditional power difference, but everywhere that there is, you tend to see that, yeah.

AUDIENCE MEMBER: Interesting, thank you.

DR. AARON KAY: So it’s very cross cultural. In fact, it’s smaller in the U.S. than most places. So you know where there’s less mobility for women in certain countries, the stereotypes or them being you know more apt to the child care thing, which is what communal stereotype is about, are even much stronger. Okay, well thank you everyone.

AUDIENCE MEMBER: Thank you. [audience claps]

DR. AARON KAY: And feel free to get in touch if you want to talk to me. Feel free to get in touch if you have any questions or anything.

Scroll to Top