Slate's Jack Shafer is on a mission to see which newspaper is most prone to using anonymous sources. He and intern Kara Hadge had Google News alert them whenever the words "anonymous" or "anonymity" showed up in an article published by the Los Angeles Times, the New York Times (NYSE:NYT), the Wall Street Journal, or the Washington Post (WPO) over a two-week period. What sparked my motivation here is this comment by Shafer:
I had a good chuckle when I discovered in the course of my newspaper analysis that the Wall Street Journal, which has no written, public standards for the use of anonymous sources, is the most reluctant of the four big dailies to cite anonymice. Maybe there's a lesson there.
But there's more than one way to be anonymous. Readers familiar with the Journal's style know that the phrase "people familiar with" the talks or negotiations, etc., is used quite often to report on a deal in the works. Dean Starkman at the Audit points out the same. Here is an example from today that just came via email (emphasis mine):
Merrill Lynch has a deal in place to sell its 20% stake in news and data provider Bloomberg for $4.5 to $5 billion, according to people familiar with the matter. Merrill also has recently held talks to sell part of its 49% stake in asset manager BlackRock, but the firm has not reached a deal with BlackRock on this.
Shafer didn't search for this phrase, so this is what happens when you do. With the help of Factiva, I looked at who uses the most anonymous sources among some major news outlets.
(My methodology, which I will say is far from academically robust: Based on the examples Shafer found, I looked for derivations of the word anonymous close to the words "condition," "insist," or "request." I also looked for the terms "people familiar with" or "person familiar with" in the first 200 words of an article. There are bound to be cases were there's a Factiva "hit" which doesn't correspond to an actual anonymous source. The same source may get counted twice, driving up the number, or an article might have more than one anonymous source, driving down the number. I went through and looked at about 50 hits across the different search results and didn't find any false positives, so I think this methodology is relatively good at picking up anonymous sources in short order.)
Here's a chart showing hits across the different media outlets. The red section shows the portion of time the "people familiar with" wording was used and the blue section shows when the "anonymous" method was used. The numbers next to the bars show the total number of non-cited sources.
Far and away, the Associated Press appears to use the most anonymous sources. But as you can see, the top two outlets are news wires, meaning that the results are likely being driven by the sheer extra volume AP and Reuters push out. Next, I divided the number of hits by the total number of articles published over a two-year period ending yesterday. This gives a sense for the propensity to use anonymous sources. A much different picture emerges:
The Journal looks like it's most willing to not cite people by name. We can also look at a possible Murdoch effect. The following chart shows the number of Factiva hits surrounding the 216 days before and after the News Corp. (NASDAQ:NWS)-Dow Jones deal closed on December 13, 2007:
On first, glance it looks like the use of anonymous sources may have increased by close to 40 percent since Murdoch took over. But when you factor in that there were more articles published in the second period than the first, the propensity to use anonymous sources has increased by a smaller but still sizable 21 percent.
Maybe this is how Murdoch plans to beat the Times?
We can also look at the effect of NYT's 2004 guidelines on the use of anonymous sourcing. Clark Hoyt, the NYT's public editor recently asked some graduate students at Columbia's journalism school to go through and "read every word of every article in six issues of the newspaper published before the policy and six from last fall. Here is what they found: The number of articles relying on anonymous sources fell by roughly half after the policy was introduced."
I searched all NYT articles in 2003, 2005, and 2007 to see if I found the same thing:
Doesn't look like it. The propensity to use anonymous sources rose 33 percent between 2003 and 2005 (and 11 percent between 2005 and 2007).
So, if true, what explains the increased use of anonymous sources? Perhaps people have become more savvy in dealing with reporters in recent years, or maybe companies are clamping down on reporter access to employees. Either of these could force reporters to rely more on anonymous sourcing.
Let me know in the comments any big holes in my methodology.
A friend lets me know that in its history, the NYT has used many different wordings for anonymous sourcing: "people close to" or "people who have been briefed on" or "people involved in" or "people who asked not to be identified."
I'll investigate this more later, but it looks like the effect of NYT's new guidelines was to reduce the use of phrases like these in favor of the "anonymous" terminology. For example, in 2003 the above phrases were used 622 times while in 2007 they were used 413 times. And that changes the increase in propensity from the 33 percent I mention above to about 11 percent.