One year: Examining the prominence of female sources in Gender Check study

Editor’s note: In January 2011, we set out to examine the ways in which women are represented in online news both as sources and as authors. To mark our first year here at The Gender Report, we’re revealing our findings from our year-long studies as well as other statistics and commentaries in a series of posts. View other coverage of our one-year anniversary here.


As we unveiled earlier this week, women made up 26 percent of human sources referenced in the articles we monitored as part of our Gender Check project. This percentage gives us a general idea of representation of women’s voices in the news, but it does not reveal how prominently these voices were incorporated. We’ve explored our data a little more indepthly so that we can address this issue and questions such as this: Was the first source in a story – a position of more weight in some respects — even more likely to male than female?

We looked at this after the six-month mark of our study, and found were at least some small signs of a lack of prominence of women sources in the online news articles we monitored. After a year, we find similar evidence.

In our year-long study, we had examined 354 articles from eight U.S. online news websites (for more details on the study, click here). Thirty-one of the articles contained no human sources. Less than 10 percent of articles had nine or more sources. The most sources a story had, as was true at the six-month point, was 25 — an article in June from ProPublica about the criminal justice process in murder cases involving children.

Roughly 15.5 percent, or 55 articles, were single-source stories. That source was male in 35 of the articles, female in 18 and unidentified in two. That made women 32.7 percent of sources in single-source stories.

The number of female sources only exceeded that of male sources twice: when we were down to two articles at source No. 21 and down to one article at source No. 25. Female sources only exceeded one-third at two other times – source No. 16 (five articles) and source No. 22 (two articles). This has remained consistent since since the six-month mark.

The first source of the articles in our sample was female 23.5 percent of the time, but the percentage of sources that were female jumped 5 or 6 percentage points for the second and third sources and then dropped back down to 21.2 percent for the fourth source.

Here’s what we found for the first five sources in a story as well as the last source in the articles.

  • First source: 23.5 percent female (in 323 articles)
  • Second source: 29.5 percent female (in 268 articles)
  • Third source: 28.3 percent female (in 205 articles)
  • Fourth source: 21.2 percent female (in 151 articles)
  • Fifth source: 25.9 percent female (in 108 articles)
  • Last source: 29 percent female (in 269 articles – not including single source stories)

In addition to the order of sources, we also examined the number of expert and non-expert sources of both genders. An expert source is an official or public figure, a person in position of authority or someone with significant knowledge on the subject.

In the articles we monitored, a larger portion of the female sources referenced in were non-experts compared to male sources. Non-experts made up 29.2 percent of female sources but just 14.1 percent of male sources. Overall, 17.9 percent of sources were non-experts.

These findings in particular raise more questions for us. Does this reflect a lack of female experts as a whole or is something different at work? Share your thoughts in the comment section below or using the #GRdiscuss hashtag on Twitter.

For more results from our year-long Gender Check project revealed this week, review the links below:

For more information on gender representations in online news, check out our “Findings and Statistics” and “Useful Resources” pages.


Findings show women as 26% of news sources in year-long Gender Check study

Editor’s note: In January 2011, we set out to examine the ways in which women are represented in online news both as sources and as authors. To mark our first year here at The Gender Report, we’re revealing our findings from our year-long studies as well as other statistics and commentaries in a series of posts. View other coverage of our one-year anniversary here.


Women were 26 percent of human sources referenced in lead articles monitored in a year-long study of eight online news sites.

The Gender Report’s Gender Check monitoring project began in January 2011 and concluded earlier this month. For this project, we monitored the articles weekly from two websites — one associated with a newspaper and one considered online-only — in each of the four different geographic regions as defined by the U.S. Census. These websites included the Seattle Times, Seattle P-I, New York Times, ProPublica,, St. Louis Beacon, Miami Herald and Patch (Seminole Heights).

The study time frame resulted in a total of 53 weeks, but due to a missing week, resulted in a total of 52 weeks of monitoring for at least two of our geographic regions. Between Jan. 18, 2011 and Jan. 19, 2012, we examined a total of 354 articles, averaging about 29.5 articles per month. In total, this broke down to 104 articles each from the West and Midwest regions, 78 articles from the Northeast and 68 articles from the South. These lower counts in the Northeast and South are the result of missed weeks. Though this may have have some slight affect on the overall findings, the standings of these regions among the others have remained consistent over time.

Of a total of 1,321 human sources in the articles we monitored, 978 were male and 343 were female. These totals do not include those whose gender could not be identified.

The year high came with women as 34.4 percent of sources in the 11th month with our low at 19.5 percent in the fifth month.

Approximately 123 of the 354 articles monitored contained only male sources (no female sources), or 34.7 percent. This compared to 25 articles with only female sources, or 7.1 percent. Thirty-one articles contained no human sources and two only had a single unnamed source whose gender was not identifiable.

Geographic regions

Among geographic regions, the sites we monitored from the South had the highest percentage of women as sources at 33.1 percent. The sites from the Northeast had the lowest at 19.4 percent. These regions have maintained the high and low percentages respectively during each quarter of our study.

Here’s how sources broke down geographically:

  • West: 258 males, 109 females (Women at 29.7 percent)
  • Northeast: 290 males, 70 females (Women at 19.4 percent)
  • Midwest: 315 males, 107 females (Women at 25.4 percent)
  • South: 115 male, 57 female (Women at 33.1 percent)

Newspaper vs. online-only websites

The online-only outlets we monitored have maintained a trend of using a higher percentage of female sources than those associated with a newspaper. After a year of study, women were 29.1 percent of sources at online-only sites versus 23.2 percent at newspaper-connected sites. In every geographic region, the online-only site had a higher percentage of female sources than its newspaper counterpart.

Newspaper websites had more sources overall, but the highest percentage of sources per article belonged to the online-only site, the St. Louis Beacon at 5.2 sources. The lowest also came from an online-only outlet — the Patch site, with an average of 1.6 sources per article.

  • Newspaper website: 546 males, 165 females (Women at 23.2 percent)
  • Online-only: 432 males, 178 females (Women at 29.1 percent)

Individual news websites

During our study, Patch (Seminole Heights) had the highest percentage of female sources at 43.4 percent in the lead articles monitored on its website. It’s noteworthy that, as previously mentioned, this site also had the lowest number of sources overall as well.

The lowest percentage of female sources came from lead news articles from the The New York Times at only 13.8 percent, followed by at 22.7 percent. These two sites have continued to hold the lowest percentages throughout the year, though more research would be needed to see if these percentages accurately reflect the content of these sites. We encourage readers to keep in mind that when broken down by each individual news site, our sample size is still small.

  • Seattle Times: 177 males, 70 females (Women at 28.3 percent)
  • Seattle P-I: 81 males, 39 females (Women at 32.5 percent)
  • New York Times: 168 males, 27 females (Women at 13.8 percent)
  • ProPublica: 122 males, 43 females (Women at 26.1 percent)
  • 116 males, 34 females (Women at 22.7 percent)
  • St. Louis Beacon: 199 males, 73 females (Women at 26.8 percent)
  • Miami Herald: 85 males, 34 females (Women at 28.6 percent)
  • Patch (Seminole Heights): 30 males, 23 females (Women at 43.4 percent)

Author gender and source selection

As part of this study, we examined how the author’s gender may have impacted  the use of female sources. As we have throughout the year, we saw a difference, with women having a slightly higher percentage of female sources than men.

Females were 28.3 percent of sources in articles written by women and 24 percent in those written by men. Articles with a shared byline between a man and women had women as 28.7 percent of sources.

  • Female authors: 269 males, 106 females (Women at 28.3 percent)
  • Male authors: 573 males, 181 females (Women at 24 percent)
  • Shared bylines: 124 males, 50 females (Women at 28.7 percent)

About this study

These findings were based on the results of our year-long Gender Check project to monitor U.S. news websites. For the monitoring, we selected the lead article from each site at the time we visited and performed a “Gender Check” by recording information on the gender of the author and the breakdown of the genders of the human sources referenced in the articles among other details. (For more on what Gender Checks are, read our introductory post here.)

As a note to our readers, this study does have some limitations. This data simply reflects our findings from eight news websites and results should not be assumed to represent all sites in the regions or the industry as a whole. Further research is needed to confirm or elaborate on these findings, which is something we’ll be looking to do in year two.

For other data on gender representations in online news and to compare these findings to others’, check out our “Findings and Statistics” category and our “Useful Resources” page.

Coming later today: Our findings on authorship of articles in this Gender Check monitoring project. Check back for more.

We’re interested in what you make of the findings. Share your thoughts in the comment section below or using the #GRdiscuss hashtag on Twitter.