What We Found Out About the Makeup of The Augur Bit’s Sources

In order to do its job well, any newspaper, The Augur Bit included, must draw information and ideas from as wide a range of people as possible. Otherwise, it risks missing out on certain stories, excluding certain perspectives and generally producing less credible work.

At the end of last school year, the two of us had a question: How good of a job do Bit articles do when it comes to quoting a diversity of voices?

So we set about amassing data from the 81 articles published in the spring semester that contained quotes. Those stories in total included 625 quotes—defined as paragraphs in which a source was quoted from an interview with a Bit writer, not a speech or written statement. 236 people were quoted in at least one story over the course of the semester.

We recorded information in two spreadsheets, one with data on each article we published that semester and the other with data on each source quoted in the Bit in that period. Lacking any database with specific self-identification, we relied on our judgment—no doubt imperfect—to classify each source’s race as white, Black or neither of those. (We went with broad racial categories to reduce the odds we’d err and to avoid statistically insignificant subsets.)

We conducted this self-study of sorts, and are making its findings public here, because the Bit owes its readers transparency. The paper should keep itself honest as rigorously as it tries to examine the institution at the center of its coverage. GDS itself undertook a third-party audit regarding diversity, equity and inclusion, which led to a 16-page report released to the community in May that lacked any quantitative results but called for GDS to better collect identity-based data in a range of areas.

We are presenting our results with little in the way of overarching interpretation, so readers may judge for themselves the paper’s performance on sourcing diversity. With a relatively small dataset, it’s important not to take small discrepancies in the results too seriously.

The heart of our project addressed the racial and gender demographics of the group of people whose words in Bit interviews wound up in the paper at least once in the semester. 62 percent of that group was white, 20 percent was Black and 18 percent were non-Black sources of color. As for gender, 53 percent of the sources were female, with male sources making up another 44 percent and nonbinary ones the remaining 3. The proportions were virtually the same when we examined only the sources quoted in multiple different stories.

Charts by Sadie Foer.

We were reassured to see a racial and gender makeup that appeared consistent with what we would expect from a newspaper covering GDS, where 40 percent of students consider themselves of color, according to the school’s website. But from there, we broke the numbers down to see any nuances obscured by the encouraging overall results.

First of all, we wondered, when a source was quoted in an article, did the average number of times they were quoted in the final, published version vary at all by race or gender? With race, slightly more quotes appeared on average per Black source (2.2) than per white source (1.7), for example; with gender, female sources were quoted 1.7 times per article, compared to 2.0 times for male sources, a number inflated by a few profiles of or Q-and-A interviews with men.

We also checked for discrepancies in sourcing diversity between different types of stories, homing in on news articles, which require a faster reporting process, and sports stories, whose sources are usually drawn from a specific team.

Here, we found variance from the aggregate numbers: 35 percent of sources quoted in sports stories were Black, compared to 18 percent in news stories and 20 percent when counting all articles. The proportions of white sources were practically constant across the three groupings, meaning that non-Black people of color accounted for a lower percentage of the sources in sports stories than in all articles.

A similar analysis looking at gender showed that a source picked at random from a news article would be female just over half of the time. Males, however, made up almost 60 percent of sources quoted in sports articles—suggestive of the teams that the Bit covered or the sources connected to those teams whom reporters interviewed, if not both factors. (It bears mentioning that GDS offers one more sport to male athletes than to female athletes in the spring season.)

The third lens through which we considered the data involved the race and gender of each article’s author. We were curious whether their identities correlated to those of the sources they were likeliest to turn to. To a certain extent they did: People of color made up 46 percent of quoted sources in articles by nonwhite writers, whereas they comprised 37 percent of quoted sources in white writers’ work.

It’s worth saying here that we did not seek in this project to absolve the Bit of the imperative to work towards diversifying its staff—far from it. Our findings suggest that diversity among writers is tied somewhat to diversity among sources, although reporters of all identities last spring largely avoided the mistake of homogenous sourcing. Besides, the advantages of a more diverse staff extend beyond the people writers interview or the quotes they use—indeed, beyond the words of the content they produce.

The fourth and last way we dissected the data on racial and gender representation was by types of sources—GDS high school students; GDS faculty and staff; and others—to find a number we could compare to GDS’ public statistic that 40 percent of the student population, albeit across all three divisions, is of color.

By that benchmark, people of color are overrepresented by 2 percent among GDS high school students who were quoted in the Bit in the spring semester; 58 percent of distinct student sources were white. A higher proportion, 66 percent, of faculty or staff quoted were white, and the percentage climbs higher still, to 71, for sources who fall into neither group.

But our study didn’t focus only on race and gender—among the other characteristics we examined were the grade levels of both authors and student sources. We wanted to see whether writers were more likely to quote members of their class, or older or younger students—and by how much.

Upperclassmen were unsurprisingly quoted more than their younger peers—they’re generally more knowledgeable about GDS and more likely to hold leadership roles. The largest gap had to do with seeking out freshman and sophomore voices—writers who were juniors (the oldest grade on the Bit’s second-semester staff) were notably less likely than their underclassmen colleagues to pay the freshman pit or the sophomores’ steps a visit for an interview.

We presented our research to returning staffers at the newspaper’s end-of-summer journalism boot camp. Whatever adjustments reporters may make in light of these findings, the biggest takeaway was that they should continue to keep sourcing diversity in mind. And though it’s unlikely anyone will repeat the laborious process we used here anytime soon, we hope that months and years from now it would yield similar, if not better, results.