Reveal’s Jim Briggs with the choir at the First Unitarian Church of Oakland. Credit: Byard Duncan/Reveal

We wanted to use sound to tell the story of racial and gender disparities in the San Francisco Bay Area’s tech sector. The result, featured in this episode of Reveal, is the Silicon Valley diversity choir. The historic First Unitarian Church of Oakland in California agreed to host our experiment, and we invited Reveal fans and friends to lend their voices to investigative reporting for an evening.

Reveal host Al Letson plays along with the Silicon Valley diversity choir at First Unitarian Church of Oakland. Credit: Byard Duncan/Reveal

Reveal from The Center for Investigative Reporting obtained never-before-released diversity data for 177 large tech companies headquartered in the Bay Area. Perhaps it won’t surprise many listeners that people of color and women are underrepresented in Silicon Valley, but the scale of the diversity is stark.

For example, in 2016, 73.2 percent of all executives and senior managers in these 177 large tech companies were white, 21 percent were Asian, 3 percent were Latino and 1.4 percent were black, according to data obtained by Reveal.

To get this data, Reveal collaborated with professor Donald Tomaskovic-Devey and his team at the Center for Employment Equity at the University of Massachusetts Amherst. Companies are required to provide information about their racial and gender representation through annual EEO-1 reports sent to the U.S. Equal Employment Opportunity Commission. The government won’t release those records to the public. So Tomaskovic-Devey and his team took criteria we provided and compiled anonymized data. That information allowed us to calculate percentages for the largest race and gender groups across job categories at the largest Silicon Valley tech companies. The data and percentages are can be found here.

As we have several times before at Reveal, we decided that music would help us convey statistical facts in an emotionally engaging way. For each data point we wanted to illustrate, we divided our 43 singers – 20 male voices and 23 female voices – into proportional groups so the volume and number of voices would match the data.

Screenshot of Sinduja Rangarajan’s spreadsheet used to calculate numbers to divide the singers into different groups. Credit: Sinduja Rangarajan/Reveal

For example, since 73 percent of all executives in these large tech companies were white, 73 percent of the 43 singers sang “white executives.” Then 21 percent of the 43 singers sang “Asian executives,” and so on. For black executives, 1.4 percent of 43 is less than one person, but we had a single person sing “black executives” to represent people from that racial group in those positions.

Katharine Mieszkowski, the lead producer for the episode, helped assemble the choir and figure out the logistics for managing the group. Fernando Arruda, our associate sound designer and engineer, recorded the sound for the evening.

Reveal’s Jim Briggs (at left) with the choir at the First Unitarian Church of Oakland.Credit: Byard Duncan/Reveal

To divide up the choir without putting volunteers in an awkward position of choosing which race their voices would represent, we handed out numbers on paper to each singer as they arrived, then called for a certain number of singers for each cue. (For example: “One through 31, you’re ‘white executives.’ ”) We didn’t divide up the choir based on the singers’ actual racial identity.

For each example, we assigned each group a note, which we then stacked into a chord. Knowing that we usually had four groups to represent, we decided to use the largest group as the “tonic” or root of a classic seven chord. Each tone that we stacked on top represented a new (increasingly smaller) group. We decided to feature the “lone voices” first, then add each group, smallest to largest. Our hope is that listeners can hear just how outnumbered minority groups often are.

Our data also showed gender disparities. We found that 79 percent of all executives were men, and 21 percent were women. We wanted to take advantage of all the male voices who were in the choir. We recalibrated our scale so that all 20 male voices represented the 79 percent male executives and, accordingly, recalibrated our scale so that five women from the 23 female singers sang “female executives.” This gave us a powerful way to use the choir to present the gender disparity data.

Then we zoomed in on the racial makeup within female executives: 70 percent were white. We divided the 23 women into groups to represent different racial groups again.

We also found that minority groups made up a smaller and smaller share of employees as they moved up the ladder from professionals to managers to executives. This is especially true for Asian women, who made up 12 percent of the professional workforce, 8 percent of the managerial workforce and 4 percent of executives. To represent this dropoff in representation among Asian women, all the women choir singers sang first to represent professionals. Then a portion dropped off, then another portion dropped off with only handful of singers left as executives.

More details on the music

Recording always means making some choices and dealing with variables that can affect the listener’s perception: how close the singers are to the microphones; varying loudness among individual singers; and even performance energy at different times during the session. We also happened to be in a creaky church building set next to a freeway. Regardless of those tough-to-fix sound concerns, the disparities in the data were so pronounced we were able to easily hear the differences.

Conducting a choir of mostly strangers proved to be far easier than we anticipated. We can’t say enough about how game the singers were for this project. Not only did they show up for free (we did provide pizza and Girl Scout cookies), they powered through three hours of work. And by the end, they were actively collaborating with us to help come up with the best approach for each sonification.

We also wanted to have the choir singing some songs they’d be familiar with, and added other fun elements to the show, including the “Mission Impossible” theme and the Otis Redding/Aretha Franklin soul anthem “Respect.” We decided to use “Respect” because it reflects the focus of this story.

More details on the data

All companies in the United States with more 100 employees have to report the demographic breakdown of their employees by race, gender and job category to the EEOC. This data is very useful to analyze disparities in companies, but it is notoriously difficult to get. The EEOC does not disclose this data to the public or journalists.

In 2017, Reveal asked 211 large technology companies to release their diversity data. Only a couple dozen companies said yes, but it’s still the largest collection of published EEO-1 reports to date.

The researchers at the Center for Employment Equity were able to take Reveal’s list of 211 companies and run it through their databases. They found 178 matches of companies that submitted their EEO-1 reports to the EEOC in 2016 and gave us the anonymized numbers of who works in Silicon Valley by race, gender and job category. (Reveal excluded Clover Health from our analysis because they told us they had submitted erroneous data to the EEOC, thus reducing our final number of companies to 177.)

Other than the 23 companies who have publicly released their data, Reveal does not know the identities of the companies, except that they are among the 211 companies we sought data from.

We have posted the data used in the sonifications on Github, along with a data dictionary.

Sinduja Rangarajan can be reached at srangarajan@revealnews.org, and Jim Briggs can be reached at jbriggs@revealnews.org. Follow them on Twitter: @cynduja and @jimbriggs3.

Creative Commons License

Republish our articles for free, online or in print, under a Creative Commons license.

Sinduja Rangarajan is a data reporter at Reveal, focusing on academic collaborations around workplace issues. She is the organizer of Mind to Mind, a symposium that brings academics and journalists together to foster conversation and partnerships. She is a former Google News Lab fellow. She is based in Reveal's Emeryville, California, office.

Jim Briggs III is the senior sound designer, engineer and composer for Reveal. He supervises post-production and composes original music for the public radio show and podcast. He also leads Reveal's efforts in composition for data sonification and live performances.

Prior to joining Reveal in 2014, Briggs mixed and recorded for clients such as WNYC Studios, NPR, the CBC and American Public Media. Credits include “Marketplace,” “Selected Shorts,” “Death, Sex & Money,” “The Longest Shortest Time,” NPR’s “Ask Me Another,” “Radiolab,” “Freakonomics Radio” and “Soundcheck.” He also was the sound re-recording mixer and sound editor for several PBS television documentaries, including “American Experience: Walt Whitman,” the 2012 Tea Party documentary "Town Hall" and “The Supreme Court” miniseries. His music credits include albums by R.E.M., Paul Simon and Kelly Clarkson.

Briggs' work with Reveal has been recognized with an Emmy Award (2016) and two Alfred I. duPont-Columbia University Awards (2018, 2019). Previously, he was part of the team that won the Dart Award for Excellence in Coverage of Trauma for its work on WNYC’s hourlong documentary special “Living 9/11.” He has taught sound, radio and music production at The New School and Eugene Lang College and has a master's degree in media studies from The New School. Briggs is based in Reveal's Emeryville, California, office.