When she was counsel on the Senate Small Business Committee, Caroline Bruckner’s boss made what seemed like an entirely reasonable request: Get some data on how often women claim certain tax provisions.

It turned out to be anything but.

She scoured all the places she would usually go for numbers, like the Congressional Research Service and the Joint Committee on Taxation.

“I had [Sen.] Mary Landrieu demanding information from me, and I would go to IRS, I would go to CRS, I would go to Treasury, I would go to JCT,” said Bruckner. “No answers, because they don’t track data this way.”

“This way” means by gender. “It’s something that’s not measured,” Bruckner was told, again and again.

What gets measured matters, or so the old policy saw goes. So when Bruckner decamped to academia, that’s where she began her research, specifically on whether women business owners claim tax breaks as much as men. (Spoiler alert: They don’t.)

Along the way, Bruckner decided to look at the legislative inputs that produced such unbalanced gender outputs: Who testified before the tax writing committees?

Bruckner herself had testified on Capitol Hill in her role as managing director of the Kogod Tax Policy Center at American University. Now she wanted to see who got a voice at the witness table as Republicans pushed through a sweeping overhaul of the tax code in 2017. The Senate Finance and House Ways and Means panels held 12 hearings on the topic. Just 19 percent of the witnesses were women, she found, and five of the hearings featured only men.

Every few years, a photo goes viral for what it doesn’t show: women missing from a hearing on birth control or a bill signing related to abortion. Public outrage flares. When Republican Rep. Darrell Issa convened a hearing on birth control with mostly male witnesses back in 2012, some Democratic women staged a walkout. (“Manel” is the pop culture nickname for this wider phenomenon, a mashup of “man” and “panel.”)

Less noticed is the lack of women in more mundane settings, where the effects of gender aren’t as obvious. Bruckner’s own research shows that small business tax breaks are aimed at capital-intensive sectors, which tend to be male-dominated, doing less for the service businesses that women are more likely to create.

Bruckner decided to expand her research, looking back at the two committees’ legislative hearings between 2007 and 2018. She found 47 percent of the Senate Finance Committee’s 355 hearings did not include any women witnesses, and 35 percent of House Ways and Means’ 479 panels were female free. Women made up just a quarter of the witnesses called between the two committees.

The tax writing committee’s man-heavy panels did not reflect the gender makeup of the relevant fields. More than 40 percent of businesses were women-owned as of 2019, and according to Bureau of Labor Statistics and industry data, women are nearly 60 percent of all accountants and auditors and half of all full-time staff at CPA firms.

Bruckner’s research popped up on Rep. Joaquin Castro’s radar when he headed the Congressional Hispanic Caucus last year. That led to the Tri-Caucus Witness Diversity Initiative, which in turn led the House to direct the Office of Diversity and Inclusion to address witness diversity on the Hill going forward.

“This effort aims to address that disparity by tracking the diversity of witnesses, and through greater public accountability, include more women and people of color at the witness table,” Castro said through a spokesperson.

The Office of Diversity and Inclusion is set to come out with recommendations in July, which Bruckner expects will include asking witnesses to volunteer their gender and ethnicity, making it easier to ensure future panels are representative.

In the meantime, Bruckner is looking at past panels, working with a team of other researchers at American University to track gender representation at legislative hearings for 15 committees going back 10 years.

The work mostly involves sorting through hearing transcripts and using names and context clues — members calling witnesses sir or ma’am, he or she, for example — to track gender. A graduate student is also looking at LGBTQ representation. Bruckner and her colleagues are focusing on legislative hearings because oversight usually just reflects the makeup of a given administration.

Looking back at representation still matters, said Bruckner, because it could provide clues for researchers and lawmakers alike searching for laws on the books with potential gender blind spots. A statute written without any female testimony is a good place for academics like her to see if there are gender disparities in its outcomes, she said. “Rarely did Congress reinvent the wheel and create brand new programs. More often than not, they build on what’s already there, and they do it by upping the funding or tweaking the rules.”

The idea is to make the information public, Bruckner said. “Congress is accountable to voters and taxpayers, and voters need to have access to that information too,” she said.

“Although,” she added with a laugh, “I’m such an idealist.”

Bruckner hopes to expand this line of research beyond gender, to include race and ethnicity. That work will be harder — there are no obvious linguistic nods to race like there are with gender — but she thinks machine learning can help create a comprehensive digital dataset going back decades.

“In order to have effective policy, you have to have good data,” Bruckner said. “And in my view, good data, absolutely, is inclusive data.”