How Variety Studied the Diversity of Hollywood Unions

Some unions reacted positively to the request for stats, while others were critical

America Ferrera On Set
Evans Vestal Ward/NBC

California is starting to ratchet up the pressure on Hollywood to diversify its workforce. Last week, the Legislature passed a new $150 million tax credit that requires productions that benefit to be “broadly reflective of California’s population.”

But nobody knows exactly what the industry’s demographics look like. Unions mostly do not collect such data, and the state’s data are imperfect and incomplete.

So Variety compiled more than 51,000 names from the rosters of 21 California-based unions, which are posted on the Contract Services website. We then hired Political Data, Inc., to help analyze that information. The firm is the leading provider of data to California campaigns, and uses surnames to sort voter rolls by ethnicity, which helps clients target their communications.

Political Data, Inc., used their dictionaries of surnames to sort the union rosters. Not every name could be matched to an ethnicity, but 78% of them were.

Of those, we found that 16.1% were Latino and 4.7% were Asian American. Both groups were dramatically underrepresented compared to their share of the state population (which is 39.4% and 15.5%, respectively).

Surnames are not perfect. For one thing, they cannot be used to estimate Black or multiracial populations. Some surnames -- like Lee or Lang -- are common among both Asian Americans and white people. And for various reasons, including marriage and adoption, a person’s surname might not line up with their self-identified ethnicity.

Nevertheless, surnames are used in fields such as public health and political science when, as here, good self-reported data are not available.

Still, we wanted to check our numbers against the self-reported data that does exist. The California Film Commission has reported diversity stats for productions that received the state tax credit from 2015-20.

That data shows that 17.2% of those whose ethnicity could be determined were Latino, and 4.1% were Asian American -- or within about a percentage point of the surname analysis.

Self-reported data suffers from its own problems -- primarily non-responses. Some people might not answer such questions because they fear discrimination, or fear being seen as a "diversity" hire. Some may simply think it's nobody's business. Depending on how a survey is conducted, the share of non-responses can become quite large.

Compounding that problem, the California Film Commission has lumped non-responses into a broad "ethnicity other" category, which also includes multiracial people and Pacific Islanders. For the 2015-20 data, 22% of total responses were classified as “ethnicity other.” This issue will be fixed in the future, according to a commission spokesperson, but for the data that exists now it poses a real problem.

When using the CFC data, we have generally excluded the "ethnicity other" responses from the denominator, on the grounds that the vast majority are likely non-responses. (If the film workforce mirrors the state population of multiracial people and Pacific Islanders -- a generous assumption -- then those groups would total 4.5%.)

The CFC has also counseled caution in relying on its data, noting that the responses are voluntary and that the data cover only those projects that have received a tax credit.

But lawmakers nevertheless relied on the CFC data in the recent debate over setting diversity goals in the credit program. On the Assembly floor, one member noted that only 5.6% of the film workforce is Black (compared to 6.5% of the state population). But if the “ethnicity other” responses are excluded from the denominator -- because most are actually non-responses -- then the Black percentage jumps to 7.2%. So Black workers are either underrepresented or overrepresented, depending on how one chooses to look at an imperfect dataset.

Clearly, better data are needed.

But even if the industry-wide data were perfect, their usefulness would be limited. It does not help very much to know that Latinos and Asian Americans are vastly underrepresented (which is true no matter how one slices the data) without also knowing precisely where those disparities lie.

And that is where surnames become useful. By employing the same method to analyze all the locals, we can be fairly confident that we are drawing valid comparisons.

The study shows, for instance, that Latinos are well represented in some jobs. The laborers, plumbers and electricians all have Latino representation that is close to, or above, the Latino share of the state population.

Among Asian Americans, that is true for only one union: the animators.

Latinos and Asian Americans are underrepresented in every other local, but the unions differ in the way they are unrepresentative.

One group is more Latino than the industry average, but less Asian American. That group tends to represent blue collar jobs: property crafts, grips, painters, plasterers, Teamsters, and studio electrical technicians.

A second group is less Latino than the industry as a whole, but more Asian American. Those unions tend to require specialized skills: cinematographers, editors, art directors and publicists.

There are also a handful of unions that are less Latino and less Asian American than the industry average. They are harder to categorize, including some office jobs and blue collar jobs as well as some skilled jobs. They include costume designers, hairstylists and makeup artists, sound technicians, office coordinators and script supervisors, story analysts, and location managers. These jobs are the least representative, compared to the state population, according to the study.

Variety reached out to each of the unions to give them a chance to dispute the analysis. Three locals provided their own numbers, which tracked fairly closely with the surname analysis.

IBEW Local 40 said that it is 38% Latino and 3% Asian American, which matches the surname analysis almost exactly (37.5% Latino, 3% Asian). The Motion Picture Editors Guild shared a survey that showed it is 7.3% Latino and 5.6% Asian American (compared to 10% and 6% using surnames). And Local 755 (the plasterers union) said it is 36% Latino and 1.2% Asian American (compared to 30.9% and 1.4% using surnames).

A few union representatives said they believed the surname results were off base, though they did not have their own figures. Thom Davis, business representative at IATSE Local 80, said he believed the union is more than 25.4% Latino, as shown by the surname analysis.

"That number sounds low to me, but I can’t dispute it,” he said.

Others said the Variety analysis sounded more or less on target.

"That’s pretty accurate," said Salvador Perez, president of IATSE Local 892, the Costume Designers Guild, when told that the study shows his union is 6.7% Latino and 3.7% Asian American. "I would confirm those numbers."

The surname method showed that Laborers Local 724 is the most heavily Latino entertainment union, with 52.4% Latino membership. Alex Aguilar, the head of Laborers Local 724, confirmed that: "We are majority Latino."

Steve Dayan, secretary-treasurer of Teamsters Local 399, estimated his Latino membership at 20%. The surname method showed it is higher than that -- 27.5%.

For the Motion Picture Costumers, IATSE Local 705, the study shows 13.5% Latino and 5.2% Asian American membership. Adam West, business agent of the union, suggested that surnames might be less accurate for their group, because costumers are mostly female. They conceded, however, that the union is "overwhelmingly white."

"We have a lot of work to do," West said.

Some were more pointed in their criticism of the method, and refused to engage on whether the numbers were accurate. Rick Markovitz, a publicist who represents the cinematographers and the art directors unions, expressed “concern” with the surname methodology, which he said is “flawed” and “troublesome.” He said that neither guild collects its own demographic data.

"They really have no comment on this issue at this time," he said.

Laurence Abrams, the spokesman for the sound union, IATSE Local 695, called the surname method "racist and inaccurate." He also did not have numbers for his union, saying that when the union has tried to gather such data in the past, many members refused to fill out the survey.

Abrams acknowledged that film crews are "not diverse," but he also cited success stories the union has witnessed through its training program. "The efforts you make to diversify your local aren’t based on the metrics. I don’t need to measure it to recognize that we’re achieving it."

Several unions are working, however, to gather their own data. Some have even considered hiring university researchers, at significant expense, to do an analysis.

"I think it’s good you’re pursuing this," said Jeannette Moreno King, president of the Animation Guild, IATSE Local 839. Regarding the lack of diversity, she said, "everybody kind of knows, but it’s so easy for somebody to blow it off."

"We need numbers."