A Critique on Diversity & Inclusion

A year-long data journalism investigation surrounding D&I in big tech

By Jiaying Brust | April 16 2021

Hello đź‘‹ My name is Ji!

I am on the cusp of being a Millennial and Generation Z.

While I got to catch the tail-end of a 90s childhood, I also grew up with Facebook, Amazon, Apple, Microsoft and Google (or FAAMG), watching them transform from garage startups to tech giants. 
‍
Growing up in the tech revolution meant I got first hand experience watching these few corporations completely change the world around me.

They have become a dominant force in our everyday lives - holding power and influence, not just in the U.S., but globally. 

Are you a Millennial or Generation Z?
Millennials include anyone born between 1981 and 1996. Generation Z includes anyone born between 1997 and 2012.

I aspire to a future in tech because I want to take part in building up the world around us.

We all want to make the world a better place. I see technology as a way of doing that.

But, as I have grown older, and have realized my own unique positioning in the world (as an Asian American and a woman), I am seeing some pretty scary numbers:

Source: https://www.latimes.com/business/technology/la-fi-tn-asian-american-tech-20150506-story.html
Source: https://www.kaporcenter.org/black-tech-workforce/
Source: https://hired.com/page/wage-inequality-report

The reality is, there are structural barriers for people of color.

Art inspired by https://www.kaporcenter.org/black-tech-workforce/

This is a growing concern considering that post-millenials are already the most ethnically and racially diverse, and the following generations will be even more so. In fact, a majority of the U.S. will be comprised of People of Color (POC) by 2060.

If POC are not being represented in companies that are building up the world around us, than the tech is not serving everyone.

And while there is growing recognition of how critical diversity and inclusion is to business performance, the numbers told a different story.

What are EE0-1 reports?
An EE0-1 is an annual compliance survey that all employers must file with the Equal Employment Opportunity Commission (EEOC).

It gives a demographic breakdown of the company employment by job category, race and gender. 

All US companies with 100+ employees are required to submit the report to the EEOC, but most do not publicly release these reports. 

I know this because that is what the EE0-1 reports are showing.
‍
Out of the FAMMG companies, Google was the first to release their report in 2014. The others quickly followed suit. 
‍
The great thing about EE0-1s is that they give us context to the numbers. We get to see the totals of each category [figure 1].

Figure 1: Source: https://static.googleusercontent.com/media/diversity.google/en//static/pdf/2018_Alphabet_Consolidated_EEO-1_Report.pdf, Google's 2018 EE0-1 Report

That being said, there are also some limitations that we should be aware of:
‍
1. The Multi-racial category does not specify which races. 
2. These reports don’t include age, gender identity, sexual orientation or disability data. 

These reports are far from perfect but they give us a baseline to compare diversity across big tech.

Across all diversity reports, there was little progress.

Especialy in the Latinx and Black representation.
‍
To the right is a comparative graph showing the percentage of workers by race in 2014 and 2018.

The EE0-1s validated what we already knew: that these companies are overwhelmingly male and white.

Source: https://www.wired.com/story/five-years-tech-diversity-reports-little-progress/

So in May of 2020, when there was an outcry on social media by corporations in response to the police killings of Black Americans...

‍

I couldn't help but think that these pledges, promises and statements felt HOLLOW.

According to a Pew Research survey, many Americans have been questioning the timing and sincerity of these messages as well. It was found that a Americans see pressure rather than genuine concern as a big factor in company statements about racism. 

In an effort to try and better understand why these messages felt HOLLOW, I began to map formal journalism on big tech (specifically at Google since they have always been the trendsetters), and critical race-related events in the past 10 years (since 2012). 

I wanted to see the relationship between what companies were saying V.S. what they were doing

I found key race-related events by tracking the top trending activism hashtags (#MeToo, #BLM) on social media by year.

For reporting on Google, I would search "Google race diversity" and pull the top news articles from the query. But I realized  that I couldn't look up the journalism on Google while using Google's Search Engine. Not only is it ironic but Google filters the information we see with paid ads, SEO, etc.

I bring up my methodology because the process to do the mapping itself showed me how embedded these tech companies are in our everyday lives.

To fix this bias, I was directed to use Nexus Uni, an academic archival database with a robust collection of news. I liked this tool because it would filter and find relevant articles easily. It would also strip them down to its barest bones (title, source, link, and preview of the body text).

Below is the mapping of major-race related events to formal journalism on Google from 2012 to 2020:

After mapping the journalism data I collected, I wanted to see if there were any patterns in the data. The first thing I noticed was a flurry of articles around the same dates....

When I took a closer look, I found a very disturbing trend.

Created by Ji Brust

To make matters worse, the trend was cyclical.

This was not just happening with George Floyd either, it has been happening since 2014.

Likewise, it goes beyond the tech space. Evidence of this can be seen in Pew Research's mapping of the global usage of the #BLM hashtag to major race-related events.

Above is a graph revealing that the usage of the #BLM hashtag would spike in response to any race-related news events.

Finally, I had an answer:
The marketing by corporate America is nice, but it's not working.

These social media promises, pledges, and statements feel insincere because while it appears they are taking action (donating $, making statements addressing racism, making D&I visible in their branding, etc.) the diversity numbers are flat-lining when we look at them. 

If we continue to do what we are doing, nothing is going to change.

We need to break out of this cycle. But to do that, we need to understand what is going on: why are the numbers flat-lining (what are we doing wrong)? And why does the conversation drop off?

Discussing diversity is complicated. We will need a guide.

This is @anonymous. They don't know much about diversity, but they're willing to learn.

I invited @anonymous to the conversation because we need to rethink everything we think we know about diversity - and having someone ask the 'seemingly obvious' questions will help us uncover challenges that we might have overlooked.
‍
With that, let's Get started!

Generally, when industry discusses diversity in the workplace, it's in terms of the EE0-1s. That means looking at the yearly snapshots of the company's demographics and seeing if they are representative of the U.S. population as a whole.

For example, if 13% of the U.S. population is Black, then this number should be reflected in corporate. However, currently, Black professionals only make up 5% of the entire tech workforce. Therefore, Black Americans are underrepresented.
‍
The other highly emphasized diversity metric is hiring/recruitment, or how many minorities are getting through the door.

Companies are "checking off the boxes" when they track these metrics alone. This is limiting.

Remember that EE0-1s are required by law, and are therefore the status quo to diversity reporting. It doesn't give a truly comprehensive understanding of diversity as a whole.

At the topmost level, the definition for inclusion is: feeling like you belong. 

Deloitte's definition, which they adopted in July of 2012 after surveying the perceptions around D&I in 1,557 employees across 3 different companies, was determined to be:

1. Being treated with fairness and respect
2. Feeling valued and a sense of belonging
3. Feeling confident and inspired (enthusiasm, empowerment and trust) 

They coined this as their "new maturity model" for inclusion and have used it ever since.

It is subjective - and this is the problem. These terms are very general and eft to interpretation. This means that companies are in the drivers seats. They are in control of determining what is considered to be "diverse and inclusive".

Something that is not subjective however, is equity. Equity does not leave room for. interpretation. It is measurable and the metrics are very clear. An example would be equitable pay OR tracking how fast POCs are being promoted within the company.

In fact, we can't even begin to think about inclusion (a sense of belonging) without meeting the basic needs first.

I think of it in terms of Maslow's hierarchy of Needs:

Source: https://www.simplypsychology.org/maslow.html#gsc.tab=0

In imagining ways to create a more equitable workplace, I quickly ran into a problem.

The retention/attrition rate is how long a person stays at a company before leaving. It is also trying to understand reasons for their departure, positions worked and the definitive types of employees who let.

I thought a good avenue to better understand D&I and equity would be to look at retention/attrition data. This number could give us better insight into why people of color were leaving tech.

BUT THIS KEY DATA WAS MISSING.

I spent half a year trying to answer that very question.

And in this sub-investigation to see why I could not get access to that data, I spoke to experts and professionals across several different disciplines.

I found that it was highly likely that these companies are tracking retention. However, it was not in their interests to share these numbers (nor were they equired to).

This is highly sensitive data, and they know the numbers are bad.

While looking for retention data, the first place I looked was on the FAAMG company websites (companies will release self-made diversity reports every year).

I realized that all of these companies have different methods of collecting and sharing that data (and they put it into their own contexts).

Even if I had access to the retention data, the data would look different across the board and I wouldn't be able to compare.

There is no standardization, therefore the numbers are unreliable.

The result fo this investigation was that big companies are throwing around D&I, so much so, that they have become buzzwords.

These numbers sound good but they don't really mean anything. The words themselves are not enough to solve or actually address the issue. Likewise,

1. The data itself is messed up
‍we are overemphasizing D&I metrics (which are the status quo and don't give us that much insight overall) while underemphasizing equity metrics.
‍
2. We can't even find/access the data
‍
3. Even if we did find the data, it is not standardized
Whereas in most industries there are processes in place

In the next section @anonymous will be addressing the current solutions for the "diversity and inclusion" problem in big tech.

I will then use real life examples to explain why these solutions are failing to actually fix the problem.

No. There are actually plenty of highly qualified, diverse candidates.

Let me give you an example of why the #pipeline problem is complete BS:

LATimes reported that in fall of 2020, preliminary data showed that Latinos were the leading group of prospective freshman accepted into the University of California for Fall of 2020. They made up 36% of 79,954 California students offered admission, while Asians made up 35%, Whites 21% and Blacks 5%. Likewise 44% were low-income students.

This was the most diverse first-year class ever to be admitted, increasing the number of accepted Black and Latino students more than 40% from the previous year.‍

In the phase out, SATs/ACTs would be optional through 2020-2021. From 2022-2023, UCs will do a blind test (not look at SAT/ACT scores) during the admissions process, up until they make a new exam.

The UC Board of Regents decided to "phase out" SAT and ACT tests as an admission requirement.

It was decided that SAT/ACT scores would not be used in the admission decision process because "standardized testing unfairly disadvantaged applicants because they lacked access to testing centers with accommodations during the Covid-19 pandemic".

A highlighted argument in the case was that standardized testing correlates with income (i.e. low income neighborhoods suffer from schools without resources or activities in the summer).

As I mentioned earlier, the decision resulted in the most diverse class of UC students, likewise-the average GPA , ACT and SAT scores remained the same.

Decisions like these can impact the candidate pool, removing barriers for diverse, high potential candidates.

First, Forbes points out that this is a denominator problem. Tech companies are growing fast, and the hiring of these underrepresented minorities remain too small to make any real impact. This means that the numbers of employees they would need to hire to achieve anything close to proportional representation keeps getting larger.

But also, we have hired more People of Color at these big tech companies: Asians.

Asians are in an interesting predicament because they are actually OVERREPRESENTED in terms of the overall U.S. population. According to a 2016 UMass Amherst study on diveristy in Silicon Valley, Asians made up 28.1% of tech workers in Silicon Valley, but only made up 17.56% of the U.S. population (U.S. Bureau of Labor and Statistics, 2016).

This makes them the largest racial cohort in the tech industry. However, research has found that they were the least likely to become managers and executives. In fact, white men and women are twice as likely to become executives compared to Asians. Also, Asian women fared the worst, being  the least likely to be promoted into executive roles overall.

The point is, even when diversity is present, it is not equitable.

Okay, the first reason why that will not fix the problem is because the current POCs in high position role are devalued.
‍
A primary example happened in December of 2020 when a prominent AI Ethics Researcher at Google, Timit Gebru, got fired.
‍
According to Wired, Gebru was a “superstar” in her field: that is, examining the ethical and societal impacts of technology. Her work consisted of making sure AI models didn't perpetuate bias on gender and race.

However, after refusing to retract or remove her name from a research paper that urged caution with Google’s AI (doing her job), she was openly fired. Not to mention she is also a Black woman with a history of speaking up about Google’s lack of diversity. 

Have you ever thought about who is restocking the snack bar at Google? How about mowing the Apple campus lawn? Delivering Amazon’s packages? 

These are what I consider to be the “invisible workers''. They are behind the scenes, doing the day to day operations.

While their jobs aren’t highly technical like software engineers, dev ops or data analytics, they are just as important as these skilled workers as have helped build FAMMG into the tech giants that they are today. 

Amazon wouldn’t be Amazon without its same-day-delivery drivers hauling packages.

Looking at how low skilled workers are treated right now is another reason why putting POC into technical and managerial roles won't necessarily work.
‍
In April of this year (2021), nearly 6000 Amazon workers in Bessemer, Alabama tried to unionize. This was a big deal because of the rarity of Amazon unions (the last one failed in 2014).

Amazon has 800 facilities staffed by 950,000 full time and part time workers. It has grown exponentially during the pandemic, increasing sales in 2020 by 38%. This is a huge pressure on warehouse staff.

The response from Amazon was that their wages and benefits were the top of the industry (they pay $15/hour). However, this isn't just about wage issues but poor working conditions: Warehouse work is demanding, dangerous and workers have little control over shifts, time off, sick leave and being fired. ‍

Amazon's response to Congressman Mark Pocan. Source: https://theintercept.com/2021/03/25/amazon-drivers-pee-bottles-union/
‍

In the end, Amazon successfully won the vote. This is not surprising considering that Bessemer workers were going up against a company that spent millions of dollars to aggressively disrupt the votes. But while they squashed the unionizing, this doesn't mean the poor working conditions have gone away.
‍
In this example workers may be paid high above industry average, but they are far from fair treatment. 

These two examples show that POC are devalued (publicly/overtly), whether they are in technical and managerial positions or not.

Hopefully, the examples I gave you illustrated why these current solutions for diversity and inclusion don't actually fix the problem.
‍

Final Thoughts.

By taking a year to explore D&I (taking a deep dive into journalism, social media, archives and databases, contacting experts in the field, etc.), I was able to examine my biases/assumptions, test my gut feelings and figure out where I was overestimating, and where I was underestimating.

Below were the key-takeaways:

  1. The current way we are looking at D&I is not working. Numbers are flatlining. 
  2. This is because we are using DIVERSITY & INCLUSION as BUZZWORDS, when we should be using more grounded data (like retention) over generalizations. 
  3. We need to make progress on equity and the fair treatment of POC before we can focus on diversity and inclusion.

‍

Ultimately, 2020-2021 was a great year to do my thesis. It was the year that flipped everything on top of its head.

I had some very painful, cringe-worthy and uncomfortable moments but I was able to learn, listen, discuss, investigate and test my assumptions.

I hope that we continue to challenge/grapple with and understand why things are the way they are. In that way, I can work towards building up a more equitable future.

‍