“Round up the usual suspects,” said the character Captain Renault in the film Casablanca. As an editor, when I see round numbers, especially round percentages, I am suspicious. People casually say 90% all the time when they mean “almost all.” But when it comes to science or surveys, a percentage indicates a level of specificity. When I see 90% or all round numbers in something that’s supposed to be factual, I immediately wonder whether there are any actual measurements behind the percentage or whether it’s somebody’s wild guess.
After I saw the sign that is partly shown above on the outside of a Washington, DC, Metrobus (the 80% part is enlarged so you can read it), I wrote to the Nearest Green Foundation asking where the number 80% came from. I also mentioned that it doesn’t say 80% of deaths from the Covid-19 pandemic, just 80% of deaths. I politely asked for the numbers behind the percentage: total deaths and total black and Latino deaths. The foundation, or at least the computer running its website, assured me of a prompt response. That was almost a month ago. (Why didn’t I capitalize Black just now? I don’t always follow Associated Press style, but the guidance, which I discussed in another post, is to capitalize it when referring to people who self-identify as Black, and who knows whether all the people labeled “black” in the death count identified themselves that way?)
Meanwhile, I started searching on my own for the source of the 80% number, because I saw it mentioned in other places but without any further information. Georgetown University kept coming up, and I found a June 9, 2020, article in the Georgetown Hoya, “Georgetown Report Highlights Racial Disparities in Health in DC,” and a June 2, 2020, press release from the Georgetown University School of Nursing & Health Studies, “New Georgetown Report Highlights Health Disparities and Calls for Racial Equity in the District of Columbia.” Both the Hoya article and the press release cited a report from the School of Nursing & Health Studies: Health Disparities in the Black Community: An Imperative for Racial Equity in the District of Columbia, dated 2020 but, as the Hoya article notes, “prepared from findings gathered before the COVID-19 pandemic.”
The report “notes that Black residents account for 80% of deaths caused by the coronavirus in the District [of Columbia],” stated the Hoya article.
The School of Nursing & Health Studies had a different take on what Health Disparities in the Black Community says: “Approximately three quarters of the deaths associated with COVID-19 in the nation’s capital have been among the African American community.”
What the report actually says is “At the time of this report, Black residents represented close to 80% of deaths caused by the virus in the District.” It also states, “The report is a synthesis of findings and does not include new quantitative content.” The only source given for the figure of 80% is an end note: “Coronavirus Data | coronavirus. Accessed April 17, 2020. https://coronavirus.dc.gov/page/coronavirus-data.” “Coronavirus Data” is a vague source, and the hyperlink is dead. In this case, the Health Disparities in the Black Community authors got lucky—sort of. The dead link has a redirect to another web page of the Washington, DC, government, but the information for April 17, 2020, consists of the numbers of tests, positives, lives lost, and recovered, but nothing about ethnicity. The web page has a link to a spreadsheet with more numbers but no mention of black, white, or other racial data. If the Washington, DC, website cited by Health Disparities in the Black Community ever had such information, it seems to be gone.
But there are other sources (or suspects).
The DCist website has a May 6, 2020, article by Becky Harlan of WAMU, the radio station of American University in Washington, DC: “Black Washingtonians Make Up Less than Half of D.C.’s Population, but 80% of Coronavirus Deaths.” However, the article includes a pie chart indicating that 86%, not 80%, of the people in Washington, DC, who died from Covid-19 were black. (My book The Editor’s Companion has a sample editing checklist; one item on the list is “Check repeated information.” This is a good example of repeated information: the title says 80%, but the pie chart says 86%.)
Another news story, on July 16, 2020, by Biba Adams in The Grio, “Washington D.C. Has the Worst Racial Disparity in COVID-19 Deaths in US: Report,” stated, “In the District of Columbia, in the shadow of The White House, more than 550 people have died from COVID-19, more than 74% of them are Black.” If you’re going to die in the shadows, it might as well be the shadow of the White House. Leaving aside the poor grammar and the political complaint tossed in by mentioning the shadow of the White House, did the proportion of black people dying of Covid-19 really go from 80% in April to (maybe) 86% in May and down to 74% in July? The Grio in turn cited a July 15, 2020, news story from APM (American Public Media) Reports, “Failing to Protect Black Lives” by Christopher Peak, which said that “the coronavirus had left … 570 dead in Washington, D.C.” So The Grio rendered 570 as “more than 550,” which technically it is, but the APM Reports story had numbers of black and white deaths (though you have to place the cursor over the bar chart to see the numbers). It said there were “421 Black deaths,” and 421 is 73.9% of 570, so the percentage (if those numbers were correct) was slightly less than 74%, not “more than 74%,” as The Grio put it. (The Grio certainly got its information from the APM Reports story, which The Grio cited, because the APM Reports story said, “The fatality rate among the city’s Black residents is 5.9 times higher than for white residents,” which The Grio changed to “The fatality rate for Black residents in D.C. was 5.9 higher,” leaving out the word times.)
At least we got away from the round numbers. But who got left behind on the bus? The sign sponsored by the Nearest Green Foundation said that the 80% included Latinos.
And I found statistics that included them, on the COVID Tracking Project’s Racial Data Dashboard, updated twice weekly with data reported by U.S. states and territories (and the District of Columbia). As of October 21, 2020, 75% of those who had died from Covid-19 in Washington, DC, were “Black or African American” people, and 13% were “Hispanic or Latino” people. So the big, scary round number on the side of the bus actually understated the problem: of the people in Washington, DC, who have died from Covid-19, approximately 88%, not 80%, were black or Latino.
The Georgetown School of Nursing & Health Studies press release was correct after all in stating that “approximately three quarters of the deaths associated with COVID-19 in the nation’s capital” were among blacks, but I don’t give the school credit for accuracy in this case, because the sources the school’s press release cited did not contain that information.
In the case of black and Latino people in Washington, DC, there were actual numbers behind the percentages that some people were reciting, but in other cases, sometimes I doubt that there are any real numbers involved. For example, I saw these statements in print:
“More than 1 in 5 mobile searches are pornographic in nature.”
“50% of Christian men and 20% of Christian women admit being addicted to pornography.”
Maybe I have underestimated the prevalence of pornography, but the numbers seemed high. I tried to find the sources that were cited to find out what they said.
The first statement had a footnote citing “A Large Scale Study of Wireless Search Behavior: Google Mobile Search,” by Maryam Kamvar and Shumeet Baluja of Google. The copy I found online was undated, but the most recent source it cited was from 2006, so it may have been published soon after that. The data for the report were sampled in 2005. It examined only “Google’s mobile search interface.” Maybe the results would have applied to all search engines in the first years of this century, but the report doesn’t say that. The report did indeed say that more than 20% of Google mobile searches were for “adult” content (I’m not using adult as a euphemism for pornographic; I put the word in quotation marks because it’s the word the report used for the “most popular type of query that users performed.”) However, the report also cited a 2002 article in IEEE (Institute of Electrical and Electronics Engineers) Computer, according to which, in the words of the Google report, “pornographic queries only accounted for less than 10%” and “found that the proportion of pornographic queries declined 50% from 1997 to 2000.… The high percentage of pornographic queries may be on a declining curve.”
If such queries may have been on a declining curve whenever the Google report was published (maybe more than ten years ago), I don’t think that the statement “More than 1 in 5 mobile searches are pornographic” is supported today .
The other source was more problematic. The percentages are suspiciously round: exactly half of men (50%) and exactly one-fifth of women (20%) supposedly admit being addicted. The footnote read, “ChristiaNet, Inc., ‘ChristiaNet Poll Finds that Evangelicals Are Addicted to Porn.’ Marketwire, Aug. 7, 2006. http://www.marketwire.com/press-release/christianet-poll-finds-that-evangelicals-are-addicted-to-porn-703951.htm (accessed Dec. 27, 2012).” The link for the 2006 Marketwire press release is dead, but I found a ChristiaNet press release online that says, “Copyright© 2017”; however, it might be the press release from 2006; maybe 2017 is when the web page was last updated. This press release says that ChristiaNet “conducted a survey asking site visitors eleven questions about their personal sexual conduct … there were one thousand responses” (another suspiciously round number). The other results mentioned in the press release are similar: 60% and 40%. The press release doesn’t say what the eleven questions were; whether the answers were yes or no, open ended, or multiple choice; or how the one thousand people were led to the poll: was it on the website’s home page or maybe associated with an article aimed at people who use pornography? The people who were polled may not have been representative of average Christians: the press release says it was a poll of Evangelicals, which it equates with Christians, but there are many other kinds of Christians, and even if the poll was valid, it might not be representative of Christians in general.
I could not find the poll itself or the results. With the percentages of responses being all round numbers, it sounds like something clumsily made up. Christia.net seems to be defunct too. As an editor, I would say that the claims are unsubstantiated.
So after chasing some suspicious round numbers down rabbit holes, my advice to other editors is this: ask for the sources that support the numbers. The round numbers might be correct, they might be someone’s wild guess, or they might be entirely made up. In fact, I think that 90% of them are made up. (That’s a joke.) And make sure the reference notes have full details about the source, because if they rely on a hyperlink, the link might be dead soon.