I collected and analyzed employment data by race for 57 of the biggest tech employers in the US (1). Here are the top level conclusions:
- Black and Hispanic/Latinx employees are significantly under-represented in tech relative to their overall percentage of the US population.
- 67% of tech companies are made up of less than 5% of Black employees.
- Controlling for overall racial disparities in tech hiring, white employees are significantly over-represented in leadership positions while all other racial groups are under-represented.
- Despite the fact that all US companies with at least 100 employees have to file an EEO-1 (which shows the racial composition of their employee base) only 16 out of 57 companies have recently made this data public.
- Here’s how you can help to encourage your company to be more open about their diversity practices.
Representation In Tech Leadership
It seems to be well known that racial minorities are under-represented in tech relative to their makeup of the US population. This is true (see the next section). What might be lesser known is that racial minorities are heavily under-represented in tech leadership roles even when you control for the fact that they are under-represented at these companies as a whole.
What does representative leadership in a company mean? It simply means that you’d expect the racial makeup of leadership in a company to resemble the overall racial makeup of that company. Simply put, if 10% of a company is made up of Black employees then equal representation means that 10% of leadership is also Black.
That is largely not the case in our data. The diagonal line in the scatter-plots below demarcates representative leadership within a company. If a company is above the line for a given race, that means a larger percentage of leadership is that race relative to the overall makeup of the company (over-representation in leadership). Conversely, if a company is below the line then leadership is under-representative of the racial makeup of the company (under-representation in leadership).
Comparison between the racial composition of employees at tech companies relative to the racial composition of leadership at those companies
Note: Data from 33 US based tech companies for US employees (2). The diagonal line represents a 1:1 relationship between the % of leaders in a company that are a given race relative to the overall % of employees of that race at a company.
You can see that nearly 100% of companies in this dataset have an over-representation of white employees in leadership positions relative to the percentage of those companies overall that are white. All other racial groups are largely under-representated in leadership. 94% of companies under-represent Hispanic/ Latinx employees in leadership positions, 82% under-represent Black employees in leadership, and 91% under-represent Asian employees in leadership.
How does this representation change when we aggregate this data across companies? For this comparison, I had to limit my analysis to the 16 companies that recently released an EEO-1 (3). We created a single metric to convey whether a given racial group has equal representation in leadership relative to their overall representation in these companies.
To do this I divided the % of leadership across all companies that are a given race by the % of employees at these companies that are a given race. So if, in aggregate, these companies have equal racial representation between their employees and their leadership this metric will have a value of 1. Over-representation in leadership will have a value greater than 1 and under-representation will have a value less than 1.
Comparison between the total racial composition of leadership at tech companies divided by the total racial composition of employees at those companies
Note: Data from 16 US based tech companies for US employees with recently released EEO-1 data. *Aggregate of Native Hawaiian or Pacific Islander, American Indian or Alaskan Native, and Two Or More Races.
This demonstrates that white tech employees are the only racial group that are over-represented in leadership in aggregate (as we’d expect from the scatter-plots). All other racial groups are under-represented. More than that, it shows that a white employee at a tech company is twice as likely as a Black tech employee to be represented in a leadership role. Another way of putting this is if 10% of a tech company is made up of white employees, you’d expect 12% of the leadership of that company to be white. If 10% of that company were Black, you’d expect only 6% of leadership to be Black.
Employees of color are under-represented in tech
Not only are employees of color under-represented in leadership positions in tech, they are also under-represented in tech in non-leadership positions. For this analysis we looked at the employee counts by race for 35 US based tech companies (4) and compared them to the overall racial makeup of the US population (5).
Racial distribution of tech employees relative to US population
Note: Data from 35 US based tech companies for US employees. *Aggregate of Native Hawaiian or Pacific Islander, American Indian or Alaskan Native, and Two Or More Races.
Hispanic/Latinx and Black people are the most under-represented in tech relative to their representation in the US. While Hispanic/Latinx make up 18% of the US population, they only account for 8% of employees in tech in our data. Black people make up 13% of the US population but only 5% of employees in tech. Conversely, while Asian people make up 6% of the US population they comprise 29% of tech employees.
The bar chart above looked at aggregated data across 35 of the largest US based tech companies. To avoid smaller companies in our data being drowned out by larger companies I also wanted to look at representation of Black employees in tech while controlling for company size. To do that I bucketed companies based on the percentage of their employees that were Black.
Number of tech companies by % of Black employment
Note: Data from 35 US based tech companies for US employees.
Not a single company in our data has more than 12% Black employees. 67% of companies have fewer than 5% Black employees while 89% of companies have fewer than 8% Black employees. Compound this with the under-representation in tech leadership and a startling picture emerges of how people of color truly don’t have a seat at the table in tech. In fact, they’re not even at the restaurant.
What you can do to help
In tech, we’re not even at stage 1 of ensuring diversity and inclusion. We’re still at step 0. Step 1 can only begin when there is data readily available so we can learn what works, and what doesn’t, in promoting diversity and inclusion.
All companies with 100 or more US employees have to file an EEO-1 with the government every year. The EEO-1 breaks out employee counts by race, job category (manager, professional, laborer, etc…), and pay. Of the 57 companies I looked at, 16 publicly shared their EEO-1 but withheld payment information by race and gender. Only 1 company, Intel, also shared pay data.
There is no good reason that a company should withhold their demographic EEO-1 data (the most common reason they give is to protect trade secrets, but that’s a bunch of nonsense). Every company in this dataset talks about diversity and readily uses pictures of people of color on their employment pages. Most even have a diversity and inclusion officer.
They sure know how to make it look like they care about diversity in their companies. Frankly, it’s time to put up or shut up. Unless there is external accountability and knowledge sharing between tech companies, we’ll be stuck in the starting gates when it comes to racial inclusion for years to come.
As a tech employee, demand accountability
If you work at a tech company and you have more than 100 US employees at your company, you should demand accountability from those in charge. Write an email to your executive leadership requesting they release their 2018 EEO-1 demographic data (2019 was postponed to 2021 due to Covid). Ask your colleagues to do the same. Here’s an email you can use:
I know as a company we’re committed to having a diverse and inclusive workplace. To further that commitment I strongly believe that we should be transparent about our progress towards that mission. To that end I think we should publicly share our EEO-1 demographic data.
This will enable us to keep ourselves publicly accountable to our goal. It will signal to prospective employees that we are truly committed to becoming more diverse.
If you succeed please send the EEO-1 my way so I can add it to this post.
As a business leader, share your EEO-1
If your company has 100 or more employees, send me your EEO-1. If you have less than 100 employees, just send me the breakout of your company and leadership by race. If you have any questions please reach out.
Send it to me. Send it as an image, a PDF. Send it via carrier pigeon or mail a hand-written note in hieroglyphics. I don’t care. The more data we have the quicker we can act on that data.
Open Diversity Data Project
I want to publicly share the source of my data for two reasons:
- So others can continue this analysis
- To hold tech companies accountable
I want this to be a living document. I want all tech companies to publicly share their diversity data so that it can be analyzed and scrutinized by people smarter than me.
With that said, here is where I got my data for my analysis (a blank cell means I couldn’t find anything for the given company)
- I looked at the largest US based internet and technology companies and tried to obtain their employment statistics by race. I also included two smaller companies (Buffer and 23andme that I knew reported on diversity data). First, I tried to find recent EEO-1s for each company. The EEO-1 is a standardized form that US companies with greater than 100 employees have to file with the US government each year. Only 17 of the 57 companies I analyzed publicly shared this information. 16 of the companies had EEO-1s from 2016, 2017, or 2018 so they were included in my dataset. Due to Covid, 2019 EEO-1s don’t have to be filed until 2021. One company, Amazon, hasn’t shared their EEO-1 since 2014 so I used their self-reported data instead. If a company didn’t share their EEO-1 I looked to see if they self-reported diversity numbers on their website. In total 19 companies that didn’t disclose their EEO-1 self-reported diversity numbers to some degree. Unfortunately, these weren’t standardized. I wanted the breakdown of racial composition for leadership positions, tech positions, non-tech positions, and overall company makeup. I will detail my methodology for how I standardized self-reported diversity data for different comparisons I do throughout this article.
- These scatterplots use data from 33 companies. Data for 16 of these companies come from their self-reported EEO-1s. 13 of these EEO-1s are from 2018, 1 is from 2017, and 2 are from 2016. The EEO-1s require that companies break out their employee counts by different job classifications. We categorized all “Executive/ Sr Officals & Mgrs” and “First/Mid Officials & Mgrs” as leadership. For the remaining 17 companies we relied on self-reported diversity data. All 17 of these companies gave the percentages (not totals) of employees by race at the overall company and the percentage of employees by race in leadership. Unfortunately, the definition of “leadership” changes company to company (and sometimes companies don’t define that term when they share this data) so the comparison isn’t directly apples to apples across companies. Still, it is helpful in looking at the racial composition of employees in power in tech relative to the overall composition of these companies. 5 of the companies that self-reported gave data from 2020, 11 from 2019, and 1 from 2017.
- Per the second footnote above, we can only get counts of employees in leadership positions by race through an EEO-1. Since companies that self-report diversity data only report percentages and don’t rigorously define “leadership”, there is no way to aggregate counts across those companies.
- In total, our data has 645,996 employees. For 16 of the 35 companies we used the EEO-1 mentioned in footnotes (1) and (2). For the other 19 companies we relied on their self-reported data on the makeup of their companies by race. These companies don’t give total employee counts by race so to approximate employee counts I looked at the total number of US based employees on LinkedIn at the time in which each company self-reported their data. While not a perfect measure, it’s a reasonable approximation for US employee count. For this analysis we excluded Amazon which totals 183,322 employees. The reason being that the majority of their work-force is concentrated in their warehouse (lower paying jobs) unlike all other companies in our data. Their most recent EEO-1 from 2014 shows that 53.8% of their workforce are “Laborers & Helpers” and people of color are largely over-represented in this work-force so including them severely and in my opinion, unfairly, overstates representation of people of color in tech.
- US population data comes from the US Census.
Originally published here.