A larger Swedish consulting company recently claimed that “A simple change in recruitment ads significantly increased women applicants”. “Wow”, I thought, “that’s so 2017”. I remember the first wave of claims like this when I was working in the automotive industry 5 years ago. We had a job advertisement “scanner” as part of our diversity tool box – rumors had it that the scanner would highlight male words so that you could change them to something more female. Which in turn was supposed to make more women apply. Even today, some companies offer paid services on improving your ads, other tools are readily available online or you can just check out the original word list here.
I got exited about the claim by the Swedish consulting company – especially since they wrote in there press release that there were “a number of studies […] [showing that] the presence of masculine gendered words discourages women from applying to male-dominated roles, as they can make women feel they don’t belong in that work environment.” Maybe that actually was true. Maybe much has changed since 2017 and new research had been published to find a causal link between carefully gendered job ads and higher application numbers of female candidates. Spoiler alert – not much had changed.
A primer
Looking closer at the said press release I found one main research paper by Gaucher, Friesen & Kay, 2011 [1]. But before we deep-dive into this, let’s pause for a moment.
- Based on your experience, would you say that there are typical female and male words?
Let’s take this one step further even, by putting this into the context of attracting and recruiting candidates.
- Do you believe that changing your job advertisements accordingly (= replacing male for female words when you want to attract more female candidates) will make a significant difference in numbers of female applications?
The original research
One of the first studies investigating wording in recruitment messages was done by Gaucher, Friesen and Kay in 2011 [1]). Actually, the publication consists of five studies all looking at two different questions:
- Are there systematic differences in the wording of job advertisements “in real life” and
- Do these differences affect potential applicants‘ attitudes.
Take-away #1: The original research is not about increasing the number of female applicants.
Looking at the title of the research itself reveals that Gaucher, Friesen and Kay present Evidence that gendered wording in job advertisements exists and sustains gender inequality. Or in easier words: they found systematic differences in the way job advertisements were phrased and that this could affect gender inequality. Nothing about discouragement of women during the application process. Nothing about actually applying for jobs.
So far, so good.
Let’s look at what the researchers looked at more specifically.
Was there gendered wording in real-life job ads?
The short answer: Yes. But it`s a vanishingly small amount.
Take-away #2: Real-life job ads contained a vanishingly small amount (= overall around 1%) of gendered-wording.
The long answer: The researchers took thousands of job advertisements from male- and female-dominated occupations. Meaning that we started off by comparing job advertisements from two extreme settings. In Study 1 they find that for all job ads for male-dominated occupations 0.97% masculine words are used. Just pause for a moment and think about this. That´s 1 word in 100. Note here though that for some occupations this number was higher, for example for engineers (11%) and computer programmers (9%). An unpublished more recent analysis [2] says to have found similar effects for ads from 2017. But anyways, in practice, this difference is really nothing much to talk about. And given the fact that the study looked at extreme cases, gendered wording in reality might be even lower.
What effects did gendered wording have?
Remember, that we started off with a very (!) small amount of gendered wording in real-life job advertisements? Now, in Study 3-5, the researchers create their own job advertisements to conduct an experiment in order to investigate how the experiment participants are affected by gendered wording. In these job ads, the gendered wording is increased to 7-8%, thus inflating the findings from Study 1-2.
Take-away #3: The job advertisements used in the study are not comparable to real-life examples.
Something else to keep in mind is the participants of the study. It’s actually Canadian university students who take part in the experiments. If you don’t want to attract Canadian university students, the generalisability of the results is limited. Also, students are enrolled in introductory courses, meaning it’s unlikely that they are job seekers. Not even speaking of the fact the jobs advertised don’t correspond to what they will be working with after their degree (e.g. “plumber”, “nurse”, real estate agent”). Overall, it’s unlikely, that the study participants represent your reality, whatever that reality might look like. If you were to do the same study with a representative sample of your potential applicants, results might differ. Another important aspect is language. The ads in the experiments are written by English native speakers for English native speakers – words might differ in other languages, and even if the same words in English were used, they might be perceived differently by non-native speakers.
Take-away #4: The participants in the study are - I'm quite sure - not representative for your reality.
Now that we know all that – what about the results? I mentioned earlier, that the reserachers did not look at application numbers as an outcome. They randomly presented the manipulated job advertisements to Canadian university students and asked them a bunch of questions. For example: “How many women (…) work in this company?” ([1], p. 115), “This job is appealing” ([1], p. 116) or “I could fit in well at this company” ([1], p. 116). You can categorize the questions into three buckets of personal skills, job appeal and belongingness. In Study 5, the effect that advertisement wording for female participants had on all of these three buckets is small (estimates ranging from -0.4 to -0.19), and in practice I would say almost insignificant since the job ads used are inflated from the beginning. Also, you are measuring individual perceptions in a self-rated questionnaire and in practice, you would want to know more about how the gendered wording affects behaviour.
Take-away #5: Even though some of the effects found in the study are statistically significant, you might want to question their significance for your daily work.
What do I make out of these results?
A word of caution: we are looking at a research article in a reknown academic journal here. As in all high-quality articles, many of the aspects I mention above are transparently addressed by the researchers. What has happened – if you ask me – is that the article has been taken out of context and cited for something that it simple can’t provide: evidence for how gendered wording affects the number of female applicants.
If you’d ask me – based on this research – if it was worth the effort to put effort into gender-decoding your job ads in order to increase female applicants – my answer would be no. But surely, almost every company making money from offering “gender-decoding” services, would tell you otherwise.
Also, as with all research, this paper presents preliminary finding that would need to be reproduced (“redone” in different contexts or in slightly different manners for example) in order to be able to say more about the quality of the results. I could only find one unpublished study that directly tried to replicate Study 1 and 2 from Gaucher et al. (2011) and found similar differences in gendered wording in job ads from 2017 [3]. However, since it is unpublished and only pre-registered I could not verify the results. That’s it. And it has been more than a decade since the paper was published. Maybe it’s time to move on.
Take-away #6: What has happened - if you ask me - is that the article has been taken out of context and cited for something that it simple can't provide: evidence for how gendered wording affects the number of female applicants.
Now, what about the Swedish consulting firm that “significantly increased women applicants“? When doing some research by yourself, you will stumble upon a number of reports saying that when job ads used non-gendered language, they received 40% more applicants. Be aware that these studies usually are correlational in nature, that is you can not determine if it was the gendered language that made a difference, if there were more candidates attracted to particular jobs or if there was another factor at work (maybe there was a major employer branding campaign launched just before posting the new job ad?).
If you want to find out if gender-decoding works in your context, you could try testing different job ads in an A/B testing manner. Meaning that you randomly present the versions to potential applicants and either measure what kind of candidates apply and/or measure company/job appeal with self-rated questions.
Or you just gender-decode because it feels good. There is nothing wrong with that. It’s just hard to argue for an idea if you do it just because it feels good – especially in the corporate context.
Take-away #7: If gender-decoding your job ads is something you still want to do simply because it feels good, go ahead. I truly mean it.
Taking a step back
When was the last time a word kept you from doing something your really wanted? Because that’s what we were talking about really here, 1% of gendered wording in real-life job advertisements. If your job posting contains only one gendered word, rephrasing it does not make a difference to candidate attitudes.
Words are always embedded in the wider context of a sentence, a paragraph and a whole text. Consider the following examples.
We challenge the bold to be humbled, the high achievers to take a break and the introverts to find their individual way to speak up.
We are bold challengers without the need to be humbled, high achievers who do not need breaks and introverts will have a challenging time to express themselves.
It’s time to move beyond words and consider the entity of the design of job advertisements. Maybe it’s time to move even further and look at how your company is perceived. Maybe it’s more about how easy it is to apply for you roles or how user-friendly your career page is. Talent acquisition is so much more and I have a hard time believing that changing single words, arbitrarily defined as male/female, will solve your problems. Changing words in your job ads won’t change your company culture.
Take-away #8: Changing words in your job ads won't change your company culture.
Take-away #9: Academic research is a continuous process of building on and refining existing knowledge. After publishing this post, I actually found another study which addressed some of the limitations in the study I analyse in this post [4].
References
[1] Gaucher, D., Friesen, J., & Kay, A. C. (2011). Evidence that gendered wording in job advertisements exists and sustains gender inequality. Journal of Personality and Social Psychology, 101(1), 109-128. doi:http://dx.doi.org/10.1037/a0022530.
[2] Pietraszkiewicz, Agnieszka; Formanowicz, Magdalena M.; MĂĽller, Petra; Sczesny, Sabine (1 June 2019). Agency and communion in job advertisements: A replication study (Unpublished). In: 19th EAWOP Congress (European Association of Work and Organizational Psychology) – “Working for the greater good: Inspiring people, designing jobs and leading organizations for a more inclusive society”. Turin, Italy. 29.05.-01.06.2019.
[3] Pietraszkiewicz, A., & Formanowicz, M. (2019, December 31). Monster Study. Retrieved from osf.io/g6bcn
[4] Oldford, E., & Fiset, J. (2021). Decoding bias: Gendered language in finance internship job postings. Journal of Behavioral and Experimental Finance, 31. https://doi-org.ezp.sub.su.se/10.1016/j.jbef.2021.100544
Further readings / resources that I found but have not used in my article
Recruiting (dis)advantage: Men’s versus women’s evaluations of gender-based targeted recruitment Webster, Brian D; Smith, Alexis N; Kim, Joongseo; Watkins, Marla Baskerville; Edwards, Bryan D. Sex Roles: A Journal of Research Vol. 83, Iss. 11-12, (Dec 2020): 706-721.
Reducing women’s lack of fit with leadership positions? Effects of the wording of job advertisements, Lisa Kristina Horvath &Sabine SczesnyPages 316-328 | Received 17 Sep 2013, Accepted 24 Jun 2015, Published online: 30 Jul 2015, https://doi.org/10.1080/1359432X.2015.1067611
Kroeper, K. M., Williams, H. E., & Murphy, M. C. (2020). Counterfeit diversity: How strategically misrepresenting gender diversity dampens organizations’ perceived sincerity and elevates women’s identity threat concerns. Journal of Personality and Social Psychology, doi:http://dx.doi.org/10.1037/pspi0000348
Marise Ph Born and Toon W Taris. 2010. The impact of the wording of employment advertisements on students’ inclination to apply for a job. The Journal of social psychology, Vol. 150, 5 (2010), 485–502
Tanja Hentschel, Susanne Braun, Claudia Verena Peus, and Dieter Frey. 2014. Wording of
advertisements influences women’s intention to apply for career opportunities. In Academy of
Management Proceedings , Vol. 2014. Academy of Management Briarcliff Manor, NY 10510,
15994.
Lea Hodel, Magdalena Formanowicz, Sabine Sczesny, Jana Valdrová, and Lisa von
Stockhausen. 2017. Gender-fair language in job advertisements: A cross-linguistic and
cross-cultural analysis. Journal of Cross-Cultural Psychology , Vol. 48, 3 (2017), 384–401.
franzidoesblog 