As they covered the Watergate scandal in the early 1970s, Washington Post
reporters Bob Woodward and Carl Bernstein frequently based stories on
information from sources they did not name. Sources in their stories were often
identified as "aides," "officials," "staffers," or simply "sources."1 In some
cases, the reporters protected sources' identities by introducing information
with phrases such as, "It was learned" or "The Post has been told."2
Notwithstanding the criticism of government officials who called their work
"shabby journalism,"3 Woodward and Bernstein helped create a niche in the
newspaper industry for a new breed of maverick journalist: the "investigative
reporter." Because of their work, their paper received the 1973 Pulitzer Prize
for meritorious public service.
Less than a decade later, another Post reporter, Janet Cooke, drew national
attention to the drug problem in inner-city slums with her profile on an
eight-year-old addict. The parts of her story, "Jimmy's World," that readers
found most shocking were based on interviews with unnamed sources. Like
Woodward and Bernstein, Cooke was praised for her courage and vision by some and
criticized for her lack of sensitivity by others. As had Woodward and Bernstein,
she too won a Pulitzer Prize.
Three days later, Cooke admitted that the boy-addict didn't exist and that she
had fabricated the quotes attributed to him. The Post returned the Pulitzer and
apologized, and the journalism industry absorbed significant blows to its
The present study is based on the supposition that the Washington Post relied
heavily on unnamed sources in its reporting during the Watergate era, and that
the policy of granting anonymity to sources had become an accepted practice by
the time Janet Cooke's fabricated story ran on September 28, 1980. It suggests
that the practice of quoting unnamed sources was not as prevalent prior to
Watergate, and that it was curbed after the "Jimmy's World" scandal. It is
expected that the results of this content analysis may be used to support the
assertion that the climate in the Post newsroom was influenced heavily by the
fame of Woodward and Bernstein, and that this climate engendered the
circumstances which allowed Janet Cooke's story to elude the gatekeeper.
The study, then, poses three research questions:
1. Did attribution to unnamed sources in the Post increase during and
immediately after the Watergate scandal?
2. Did attribution to unnamed sources in the Post decrease after the
Janet Cooke scandal?
3. Could it be said, on that basis, that there is an association between
the two events?
In each case, an affirmative response is expected.
It is sometimes said in the newspaper community that the Post's use of the
unnamed source "Deep Throat" during the Watergate era was not unethical because
Woodward and Bernstein never quoted Deep Throat and used his information only
for "deep background." It is said, too, that Woodward always confirmed what Deep
Throat told him. Consequently, Woodward and Bernstien did not really quote
unnamed sources in their stories. That is simply untrue. Woodward and
Bernstein routinely attributed information to unnamed sources.5 In one
instance, they quoted 21 unnamed sources in a 29-paragraph Watergate story.6
Moreover, their use of unnamed sources was noted and recorded by other
journalists of their era.7
Subsequent book and movie deals made Woodward and Bernstein instantly famous and
extremely wealthy, which, according to biographer Adrian Havill, satisfied an
abiding need for recognition and success: "Bob was consumed with naked ambition
. . . There were money and fame at stake. . . . In Carl, who had been telling
tall tales since he was a teenager, Bob had the perfect partner. . . . The
unnamed source was the cover."8
Janet Cooke used that cover in pursuit of her own fame, to satisfy her own
pressure to succeed. She hadn't been able to handle the pressure of competition
at prestigious Vassar College, and had dropped out after a year to return home
to Toledo, Ohio.9 However, when she arrived at the Post after earning a
bachelor's degree at the University of Toledo and reporting two years at the
Toledo Blade, Cooke encountered a new set of pressures, particularly the
pressure of a highly competitive newsroom atmosphere, which executive editor Ben
Bradlee purposely fostered. "There's a lot of pressure on young reporters
here," Bradlee said. "They get all the lousy beats . . . If they don't get good
stories, they will never make the national staff."10 Unofficially, Cooke
received the same indoctrination other Post reporters received: Be bold.
Stretch for the story. Management will stand behind you.11
Management stood behind reporters because Bradlee and Woodward (who by that time
was the paper's assistant managing editor for metro news) were always on the
lookout for the second coming of Watergate. They wanted what they called
"holy-shit" stories.12 Post editors sought them actively. For example, despite
the alarm bells that went off when Cooke first presented the Jimmy story, city
editor Milton Coleman was eager to believe it. This is not so hard to
understand when it is unknown that just weeks before the Jimmy story appeared,
Coleman had engaged in a heated dispute with veteran reporter Courtland Milloy
over another questionable story. Coleman had ordered Milloy to find a family in
Washington whose financial structure would be ruined by President Reagan's new
economic policies. Milloy returned to the office and reported that he had been
unable to locate such a family. Coleman insisted, and the two engaged in a
shouting match in the newsroom that ended when Milloy told Coleman to find
someone else for the story; he, Milloy, would not invent evidence.13
Cooke entered that picture with a desire to make a name for herself. She has
since come to be identified with the overuse or misuse of unnamed sources,14 but
that may not be her fault entirely. She knew the impact would be greater if the
child in her story were perceived as a real person, rather than as a composite
representative of a social problem. She knew there was little chance she would
be found out. After all, who knew the true identity of Deep Throat? As Cooke
herself later admitted, she didn't think she would be caught. She had been
promised she would not have to reveal her sources.15 When government officials
began to ask Cooke to name her source so that "Jimmy" could be helped,
Woodward said the Post had invoked a tradition he had helped to cultivate. "We
went into our Watergate mode," he said. "Protect the source and back the
The idea that the use of unnamed sources increased following Watergate is
supported by Wulfmeyer's content analysis on the use of anonymous sources in
Time and Newsweek. Wulfmeyer found that 81 percent of the 388 articles he
analyzed quoted unnamed sources. The articles appeared in the national and
international sections of 12 issues (one per month for every month in 1982) of
When Wulfmeyer's data is compared to that from Culbertson's 1976 study on the
same theme, the increase of attribution to unnamed sources in the years after
Watergate is evident. Culbertson found that 70 percent of Newsweek stories and
72 percent of Time stories relied on what he called "veiled" sources. He also
found that 33 percent of all newspaper stories attributed information to unnamed
After the Cooke scandal, Anderson surveyed U.S. newspaper editors, nearly half
of whom reported that attribution to unnamed sources increased significantly in
their papers after Watergate. At the same time, 75 percent of the editors said
they believed newspapers overused unnamed sources and 92 percent agreed that,
because of the Pulitzer hoax, editors in the future would examine more carefully
stories relying on unnamed sources.19 Additional research by Wulfmeyer also
supported the notion that editors became more cautious of unnamed sources
following the Cooke scandal.20
Brickajlik's unpublished master's thesis lends support to the assertion that use
of unnamed sources at the Post increased after Watergate. She also found that,
while Post reporters continued to quote unnamed sources with great frequency
after the Cooke scandal, overall instances of unnamed attributed decreased.21
Stories appearing in the Washington Post front and city/metro sections during
four two-year periods were analyzed for use of unnamed sources. The sample was
selected from these four two-year periods:
y June 1970 to May 1972 (before the Watergate break-in).
y June 1972 to May 1974 (during Watergate reporting).
y October 1978 to September 1980 (after Watergate and leading up to "Jimmy's
World," which appeared in September 1980).
y April 1981 to March 1983 (after "Jimmy's World" was uncovered as a hoax in
For each two-year period, all stories appearing in four constructed weeks were
analyzed. The dates in each of the constructed weeks were selected at random,
using simple random sampling, so that all possible dates in each two-year period
had an equal chance of being selected.
Only investigative and enterprise stories appearing in the front and city/metro
sections of the paper were analyzed. Investigative and enterprise stories were
defined as articles developed by the enterprise of the reporter, through
cultivation of facts and sources or both. Generally, these stories shed light on
a problem or dilemma which might have remained hidden without the journalist's
enterprise. Excluded from consideration were articles reporting results of
public meetings, speeches, press releases or matters of similar origin. Stories
likely to have appeared in other newspapers were not considered, because they
were likely to have been based on information dispersed to all media on a given
beat. At the same time, a story may have been considered an
investigative/enterprise article if information routinely distributed to all
reporters was used as a springboard for further research, yielding a more
detailed and more compelling story.
When an article was identified as fulfilling these criteria, its date, headline,
author(s), location, number of paragraphs, nature (foreign or domestic) and
general topic were recorded. Then the article was submitted to statement
analysis, and the number of attributions to unnamed resources was counted and
The coding instrument was tested and found to be functional, after which coders
were tested for reliability. Intercoder reliability ranged from 91.5% for
nature and topic of story to 99.8% for number of instances of unnamed
attribution. Overall, intercoder reliability was 95%.22 Data were considered
statistically significant at the .05 level.
The sample, consisting of 162 investigative or enterprise stories, was
distributed unevenly among the four two-year periods. Fifty stories appeared in
1970-72, 32 in 1972-74, 45 in 1978-80, and 35 in 1981-83. The majority of the
stories (86%) appeared in the front section. Likewise, a majority (77%) were
domestic in nature. Articles varied in length, ranging from nine to 126
paragraphs. The mean was 35 paragraphs long. Thirty articles contained no
attribution to unnamed sources. The balance, 81.5%, featured at least one
instance of attribution to an unnamed source. The mean was 4.15 instances of
attribution to an unnamed source.
Crosstabulation of the presence or absence of unnamed sources by the four
two-year periods revealed no significant relationships between these two
variables. However, the data began to take shape when raw figures reflecting the
number of unnamed-source attributions for each article were collapsed into
categories and a similar crosstabulation was performed.
As illustrated in Table 1, 70% of the 50 stories from 1970-72 contained three or
fewer attributions to unnamed sources, and 22% contained between four and nine
instances. The proportion of sources containing between four and nine instances
of unnamed attribution increased dramatically during the Watergate period of
1972-74, to 40% of the 32 stories.
The trend toward use of unnamed sources seems to have accelerated by the
two-year period immediately preceding the Janet Cooke affair, 1978-80. The
proportion of stories containing four to nine instances of unnamed attribution
remained strong (33.3%), but the proportion of stories containing 10 or more
instances increased to 24.4% (up from 15.6% in 1972-74 and 8.6% in 1970- 72).
Thus, the first research question is answered in the affirmative. Instances of
attribution to unnamed sources in the Washington Post increased significantly
during the Watergate period, as well as during the two-year period leading up to
the Janet Cooke scandal.
After Cooke's hoax was revealed, attribution to unnamed sources in the Post
decreased. Of the 35 articles, 74% contained three or fewer instances of unnamed
attribution, while 23% contained between four and nine instances, and only 3%
contained five or more. Thus, the second research question also is affirmed.
Instances of attribution to unnamed sources in the Post decreased significantly
in the two-year period following the revelation of the "Jimmy's World" scandal.
A chi-square of 17.4 based on this data yielded a p-value of .008. Thus, it is
safe to conclude that an association between these two variables exists.
Further examination of the data reveal that 91.9% of the foreign stories relied
on unnamed sources, while only 78.4% of the domestic stories used unattributed
material. Also, while only 29% of the domestic stories contained five or more
instances of unnamed attribution, 46% of foreign stories featured five or more
instances of unattributed material. To eliminate bias that might be attributed
to the tendency of foreign correspondents to rely on unnamed sources,
contingency tables were prepared which disregarded foreign stories.
The first such table, Table 2, reinforces the answers to the first two research
questions. The number of attributions to unnamed sources per domestic story was
significantly greater during the Watergate period (1972-74) and immediately
before the Janet Cooke affair (1978-80) than it was before Watergate (1970-72)
or after "Jimmy's World" (1981-83). However, because four of the 12 cells
contain expected frequencies of fewer than five, the chi-square value cannot be
considered statistically significant.
To compensate for that weakness, two columns in the dimension "Instances of
Unnamed Sources" (4-9 instances and 10-plus instances) were collapsed to form a
single column (4-plus instances). Table 3 shows the statistically significant
results of collapsing these categories. It demonstrates that during the
Watergate period and before the Cooke scandal, instances of unnamed attribution
in the Post's domestic coverage approached or exceeded the mean for the entire
study (4.15 instances). Whereas domestic stories in the Post quoted unnamed
sources four or more times in 17.1% of the stories analyzed for the
pre-Watergate period (1970-72), more than half the stories contained four or
more such attributions during Watergate and immediately before the Cooke scandal
(57.7% in 1972-74, 53.8% in 1978-80). However, after the Cooke affair, the
proportion of domestic stories containing four or more attributions to unnamed
sources dropped below the pre-Watergate level, to 16% in 1981-83.
The findings presented here indicate affirmative responses to each of the three
research questions. The use of unnamed sources in the Washington Post increased
during and after the Watergate period and decreased after the Janet Cooke
scandal. Chi-square analysis indicates association between the number of
instances of attribution to unnamed sources in the Post and the four two-year
periods in question. The strength of that association increases as control is
placed on the effects of an intervening variable _ foreign stories.
Based on these findings, there are grounds to suggest that the use of unnamed
sources by Woodward and Bernstein during the Watergate era bore fruit at the
Washington Post, to the point that Janet Cooke believed she could get away with
such an elaborate hoax. This study demonstrates that the Post made increasing
use of unnamed sources during its coverage of the Watergate scandal and
continued to do so until the Cooke case taught its editors about the pitfalls of
relying on unnamed sources. The strength of the association between the two key
variables indicates more than a chance association.
Susequent research attempting to explain and strengthen this association might
seek commonalities in the paper's gatekeeping process during each of the four
periods. Also, research might attempt to pinpoint common denominators in the
personalties and character traits of the reporters in question.
As presented here, the results of this study cannot be generalized beyond the
Washington Post. It may prove fruitful to analyze the contents of other
metropolitan papers during the same periods and determine whether those papers
followed a similar pattern. Only then can the results of this study be
generalized beyond the Post.
1. See Carl Bernstein and Bob Woodward, "Nixon Ex-Aides, 5 Others Indicted In
Bugging Case," Washington Post, September 16, 1972, A1; Carl Bernstein and Bob
Woodward, "Grand Jury Quizzed Both Mitchell, Stans," Washington Post, September
22, 1972, A16; and Carl Bernstein and Bob Woodward, "Five-Man Panel, $500,000
Sough for Watergate Probe," Washington Post, February 6, 1973, A1.
2. Carl Bernstein and Bob Woodward, "Key Nixon Aide Names As 'Sabotage'
Contact," Washington Post, October 15, 1972, A1; and Carl Bernstein and Bob
Woodward, "Nixon Alerted to Coverup in December," Washington Post, April 23,
3. Carl Bernstein and Bob Woodward, All The President's Men (New York: Simon &
Schuster, 1974), 186.
4. "The End of Jimmy's World," Washington Post, April 16, 1981, A18.
5. Between September 1972 and April 1973 _ the period during which it could be
argued that Woodward and Bernstein wrote some of their most significant
Watergate stories _ they averaged nearly 10 attributions to unnamed sources per
story. For examples, see Bernstein and Woodward, "Nixon Ex-Aides, 5 Others
Indicted In Bugging Case," A1; Bernstein and Woodward, "Grand Jury Quizzed Both
Mitchell, Stans," A16; Bernstein and Woodward, "Key Nixon Aide Names As
'Sabotage' Contact," A1; Bernstein and Woodward, "Five-Man Panel, $500,000 Sough
for Watergate Probe," A1; and Bernstein and Woodward, "Nixon Alerted to Coverup
in December," A1.
6. Bernstein and Woodward, "Nixon Alerted to Coverup," A1.
7. See Samuel L. Dash, Chief Counsel (New York: Random House, 1976), 51; John
Dean, Blind Ambition (New York: Simon & Schuster, 1976), 289; Elizabeth Drew,
Washington Journal (New York: Random House, 1974), 28; Adrian Havill, Deep
Truth: The Lives of Bob Woodward and Carl Bernstein (New York: Birch Lane,
1993), 142; Jeb Stuart Magruder, An American Life: One Man's Road to Watergate
(New York: Atheneum, 1974), 246; and Clark Mollenhoff, Game Plan for Disaster
(New York: W.W. Norton, 1976), 244.
8. Havill, Deep Truth, 87.
9. See Naomi Munson, "The Case of Janet Cooke," Commentary, April 1981, 47; and
Havill, Deep Truth, 146.
10. Philip Nobile, "The Pulitzer Surprise," New York, April 27, 1981, 22.
11. "Ben's World," National Review, May 15, 1981, 532.
12. This identifying tag comes from the words Woodward said he uttered at the
arraignment of the Watergate burglars when he heard James McCord tell the judge
that he had worked for the CIA. See Bernstein and Woodward, All The President's
13. "Ben's World," 530.
14. David L. Eason, "On Journalistic Authority: The Janet Cooke Scandal,"
Critical Studies In Mass Communication, 3 (4), 429.
15. See Havill, Deep Truth, 147.
16. William Safire, "Bradlee's World," New York Times, April 20, 1981, B6.
17. K. Tim Wulfemeyer, "How and Why anonymous Attribution Is Used by Time and
Newsweek," Journalism Quarterly, 62 (1985), 83.
18. Hugh M. Culbertson, "Veiled Attribution _ An Element of Style?," Journalism
Quarterly, 55 (1978), 456-465.
19. Douglas A. Anderson, "How Newspaper Editors Reacted To The Post's Pulitzer
Prize Hoax," Journalism Quarterly, 59 (1982), 364-65.
20. K. Tim Wulfmeyer, "The Use of Anonymous Sources in Journalism," Newspaper
Research Journal, Winter 1983, 43-50.
21. Linda J. Brickajlik, "An Historical Profile of the Washington Post And Its
Use of Confidential Sources," Master's thesis, CBN University, 1989
22. Intercoder reliability was determined by the percentage of agreement method
suggested by Stempel. See Guido H. Stempel III, "Reliability in Content
Analysis," Journalism Quarterly, 32 (1955), 333-34.