Hired had a problem. Women on the online hiring platform were consistently underbidding themselves. Hired, which lets technology workers post their preferred salaries so companies can interview them and extend offers, saw female applicants ask for lower salaries 69% of the time compared to men with comparable experience and skills.
Fixing that, thought Jessica Kirkpatrick, Hired’s lead data scientist, was a matter of education: give women data about average salaries in their field and they would correct their mistake. She helped design a feature giving candidates a custom range of average salaries based on their skills and experience.
It didn’t go well.
After testing with thousands of candidates, the gender gap worsened. Women in the experimental group, newly aware of their salary distributions, placed themselves on the lower half of the salary curve. Men, with comparable skills and experiences, tended to assign themselves to the upper half. That led to a gender gap in preferred salaries that was even bigger than the gap among those who didn’t see the data at all (women in the experimental group asked for 9.3% less money than men compared to the control group which asked for 4% more money than men).
In retrospect, Kirkpatrick says the results don’t surprise her. In her first job as a data scientist, the former physicist—who holds an astrophysics PhD from the University of California-Berkeley—asked for a base salary near the lower end of her field. She eventually discovered men on her team with similar experience and job titles were paid $10,000 more. The income loss compounded over years and different jobs until arriving at Hired (which boosted her pay by 25%).
“I know that as a woman I am…likely to be more hesitant to put myself forward for opportunities where I don’t meet ‘all’ the qualifications—yet knowing this information doesn’t make it easier to advocate for myself or put myself forward for challenging opportunities,” Kirkpatrick wrote by email. “I still struggle with this on a regular basis. As a data-oriented person, I do find it easier to advocate for myself when I am armed with objective facts or statistics that can help back-up my requests. I also regularly put myself forward for opportunities that terrify me and where I don’t feel qualified, because I figure if I had the confidence and socialization of a man, I maybe wouldn’t have the same hesitations and self-doubt.”
Yet the experiment’s failure belied a promising trend Hired uncovered in a gender-wage gap report it released on April 4. The report was based on 120,000 interview requests and job offers from 8,000 companies in 2016.
Women in tech early in their career are now asking for higher salaries, and getting them. Analyzing the 2016 data, Hired found a sharp divergence between relative new and senior technology professionals. Female candidates with less than five years of work experience ask for higher salaries than their male counterparts, and those with less than one year of experience were the most confident, asking for an average of 4% more relative to comparable men. They also received 8% more in salary offers, equivalent to $1.08 in salary for every $1 offered to men. On average, that equated to more than $3,300 per year.
Yet that relationship reverses after six years. At that point, female candidates ask for about $5,000 less relative to comparable male candidates.
The reason is hard to isolate. Kirkpatrick suspects the more senior female job candidates may have seen lower salaries early in their career compounded over time, while new entrants are pushing for parity. A second possibility is that the demand for female tech companies to diversify their workplace has put female developers in high demand, and most of these new jobs are going to more junior candidates.
Hired is now exploring ways to give candidates salary data without widening the gender gap such as replacing a salary range with a single-number recommendation.