Are the Numbers too High?   

There may be even fewer people of color working in America's newsrooms than ASNE's statistics suggest.

It's become a newsroom rite of spring: the release of the American Society of Newspaper Editors' annual survey of minorities in the newsroom. Each April industry leaders wring their hands over the slow progress in the efforts to diversify U.S. newspapers as they pore over data that is measured in hundredths of percentage points. Over the past 10 years, the annual increase has averaged 0.25 percent.

But the heart of the story – the numbers – is rarely questioned. It is, however, likely that the total number of minority journalists is actually lower than the much-analyzed ASNE census figures reported each year.

The problem lies in what pollsters call non-response bias – the likelihood that those who decline to participate in a survey are in some way significantly different from those who do respond. In the case of the ASNE survey, the question is simple: Are newspaper editors who have particularly low numbers of minorities in their newsrooms less likely to respond to the survey, hoping to avoid looking bad in the eyes of the industry? And conversely, are editors who have unusually high numbers of minorities more likely to respond in the hopes of showing off their papers' strong diversity records?

ASNE, which has conducted the newsroom census for 27 years, looked at the non-response issue early on and found that the overall proportion of minorities was probably about .3 percent less than the projected figures. But that follow-up testing was done at a time when individual newspapers' statistics were kept confidential. For more than 10 years, ASNE has disclosed the percentage of newsroom minorities for every responding paper. And ASNE officials say there has been no follow-up testing to measure differences between responding and non-responding newspapers since the association changed the reporting method to make individual paper data public.

ASNE Executive Director Scott Bosley says he is confident that non-response bias is not a factor in the survey because of the high response rate, which is traditionally nearly two-thirds of all U.S. dailies and higher for the bigger newspapers. "Of course, we're not sure with certainty, but we're very comfortable with our numbers," Bosley says. "We're pretty sure it's a good figure because of the high response."

In 1990, a year when ASNE gave newspapers the choice of keeping their data confidential, the minority newsroom population at papers that permitted their information to be made public was 8.5 percent while the papers that kept the information private was 7.2 percent.

ASNE accounts for the newspapers that do not respond to the survey by assigning to those papers the average number of minorities in newspapers within their circulation category. However, an AJR analysis of the data shows that often, the papers that do not respond to the survey historically have minority employment levels far below their category averages.

For instance, in the largest circulation category, dailies with circulations 500,000 and above, 11 of the 12 newspapers responded last year, averaging 18.7 percent minorities in the newsroom. That average was then projected on the one newspaper that did not respond – the New York Post. But in the most recent years when the Post did respond (1997 through 2001), it had the lowest number of minorities among the largest newspapers.

A similar pattern is found in the 250,000-500,000 category. The 25 newspapers responding to the ASNE survey reported an average of 20.6 percent of minorities in their newsrooms. But two of the three newspapers not responding had minority newsroom levels below their category averages in the last years they responded. One of them, the Boston Herald, had by far the lowest minority employment of any of the newspapers in that category the previous year – just 5.5 percent.

In the 100,000-250,000 category, the minority average was 15.3 percent, and eight of the 65 newspapers in that category did not respond. Of those, seven had minority levels well below their category averages in the last years they responded. The Providence Journal, which reported 5.4 percent minorities in the survey published in 2000, was lower than almost every newspaper in the category that year.

Such differences between those who respond to such a survey and those who don't were on display during a similar diversity census conducted jointly by Unity: Journalists of Color Inc. and the University of Maryland's Philip Merrill College of Journalism. That study, which looked at diversity in Washington bureaus of daily U.S. newspapers, found that, in general, bureaus with relatively high numbers of minority editors and reporters quickly provided the data while those who had few or no minority journalists often refused to participate. (The data on non-responding bureaus was obtained through other sources.)

G. Evans Witt, a former Associated Press editor and now chief executive officer of Princeton Survey Research Associates, says the error in the ASNE census probably is "relatively small" because of the survey's high response rate among larger newspapers, which have many more journalists than their smaller counterparts. But he said non-response bias could be a bigger factor when looking at non-population statistics such as the overall numbers of newspapers without minorities.

About 40 percent of the 927 newspapers that responded to the most recent survey reported no minorities in their newsrooms. But of the 490 newspapers that did not respond to the ASNE questionnaire, 266 of them – 54.3 percent – reported all-white newsrooms the last time they did respond to the survey.