appcpenn.org
202.879.6700
December 20, 2000
Are Voluntary Standards Working?
Candidate Discourse on
Network Evening News Programs
by Erika Falk, Researcher, Annenberg Public Policy Center
and Sean Aday, Ph.D., George Washington University
Key Findings
• None of the networks provided the recommended five minutes a night of
candidate-centered discourse during the 30 days before the primary and general
elections.
• There was more candidate-centered discourse broadcast during the general
election than the primary. During the general election there was on average 64
seconds of CCD per night/per network.
• Strategy frames dominated the stories. During the general election 62 percent of
stories employed a strategy frame. Twenty-seven percent employed an issue
frame.
• During the general election, the length of the average sound bite remained short at
nine seconds.
• NBC and CBS aired special interview segments with the candidates during the
last week before the general election and these segments both contained
significant CCD and employed issue frames.
Despite a trend away from network news watching, about 52 percent of Americans watch
news on one of the major networks every night with an additional 19 percent saying they
watch several times a week (Roper, 1999). Though television remains the dominant
medium for finding out about election information, network news has been dropping in
popularity. A recent poll by the Pew Research Center found that only 23 percent of
people turned to the network news as their primary source of information about the
election (down from 36 percent in November of 1996) (Digital Divide, 2000). The
dropping popularity of network newscasts for election information may be a result of the
dropping quality of election coverage.
A number of studies have shown that broadcast news is remarkably lacking in the kind of
substantive discussion of politics that people need to make informed decisions. Rather
than reporting on the candidates stands and providing analyses of the relative merits of
their positions, network news is far more likely to reduce an election to a game --
covering who is ahead and who is behind and what strategies the candidates are using in
order to win (Jamieson, 1996). Broadcast news is more likely to tell voters why
candidates said something than what they actually said. If a candidate makes a health care
reform proposal, for example, network reporters tend to reduce their analysis to how the
plan will appeal to certain interest groups, instead of discussing the specifics of the plan.
Studies have shown that this type of strategy reporting not only leads to less learning on
the part of audiences, but also encourages people to adopt cynical attitudes about politics
(Cappella and Jamieson, 1997).
If the press tends to focus more on spin and strategy than issues and substance, it should
come as no surprise that candidates themselves are relatively scarce in campaign
coverage. After all, candidates on the stump spend most of their time making a case to
voters for why they should be elected, and those arguments tend to be issue-oriented.
Indeed, studies have shown that over the last 20 years candidates have been increasingly
silent on the network evening news, with the length of candidate sound bites dwindling
from an average of 43 seconds in 1968 to fewer than 9 seconds in 1988 (Hallin, 1990).
In order to improve the quality of broadcast election coverage, many have begun to lobby
the networks to reverse the trend toward less candidate discourse. One prominent
proposal came from the Advisory Committee on the Public Interest Obligations of Digital
Television Broadcasters. The panel was established by President Clinton to study what
obligations the stations had for operating in the public interest in light of the additional
spectrum space that the FCC was allocating for digital television. The panel was
comprised of representatives from the broadcasting industry as well as academic and
civic organizations. The Committee recommended that television broadcasters
voluntarily air five minutes a night of "candidate-centered discourse" (CCD) in the 30
nights before all elections in order to offer more voter education. This proposal came to
be known as the "5/30 Recommendation." Though the recommendation called for
voluntary standards most of the Committee members had supported mandatory free time
for candidates. However, since the broadcasters on the Committee had opposed making
the provisions mandatory, the Committee adopted a proposal on which all members could
reach a consensus (i.e., voluntary compliance).
This study, conducted by the Annenberg Public Policy Center of the University of
Pennsylvania in conjunction with the Alliance for Better Campaigns and funded by The
Pew Charitable Trusts, analyzed network news and news magazines (between 5:00 PM
and 11:30 PM) for the 30 days prior to the Super Tuesday primaries on March 7, 2000
and the 30 days before Election Day on November 7, 2000 to see whether ABC, CBS,
and NBC lived up to the voluntary standards set forth by the Committee (i.e.,
broadcasting five minutes a night of candidate-centered discourse for the 30 days before
an election).
The first task in a study like this is to fill in the details left unaddressed by the
Committee; specifically, what constitutes candidate-centered discourse? What if, for
example, the candidates are simply asked strategy-oriented questions and never given the
opportunity to discuss issues? Even if candidates are heard discussing their ideas and
platforms, should newscasts be lauded for wrapping those quotes into stories that are
otherwise dominated by horse-race frames? Clearly, these examples would not be in the
spirit of the Committees goal of encouraging more voter information. To this end,
APPC and the Alliance, in conjunction with researchers at the University of Southern
Californias Annenberg School for Communication, adopted a formula that factored in
both the total amount of time candidates were heard expressing themselves and the news
frame of the stories. Thus, the following metrics were used in calculating candidate-
centered discourse (CCD):
If the story mainly used a strategy frame, only the total amount of
time a candidate was heard speaking would be counted as CCD.
If the story mainly used an issue frame, AND the total amount of
candidate sound bites represented at least 30 percent of the total
length of the story, then the entire length of the story was counted
as CCD.
If the story mainly used an issue frame, BUT the total amount of
CCD represented less than 30 percent of the total length of the
story, then only the length of the candidate sound bites was
counted as CCD.
1
The advantage of this coding scheme is that it captures not only CCD, but also whether
the networks ran substantive coverage. Without such a two-pronged analysis, it would be
impossible to know if a substantial amount of CCD was mired in strategy coverage, or
even if a lack of CCD obscured a laudable amount of issue-oriented news. If the networks
provided any measure of useful coverage, it would be best captured by this approach.
During the primaries all three networks failed to meet the voluntary standard set out by
the Committee. Indeed, only one of the networks ABC even averaged five minutes
of total election coverage a night during the 30 days before Super Tuesday. During the
general election, the total campaign coverage was better with each network averaging
about 7 minuets and 44 seconds a night of election coverage. ABC showed the most
improvement by almost doubling its coverage from 5 minutes 5 seconds to 9 minutes 48
seconds a night. NBC also significantly increased its coverage from 4 minutes 34 seconds
to 7 minutes 35 seconds. There was an increase in election coverage on CBS, too, but to a
lesser extent. They increased from 4 minutes 19 seconds to 5 minutes 50 seconds a night.
2
(Table 1). In theory all three networks devoted enough time to election coverage during
the general election to broadcast five minutes of candidate-centered discourse a night,
but, in fact candidates themselves received little of this time.
Table 1: Average time devoted to campaign stories per night
Election coverage/Night ABC CBS NBC
Primary 5:05 4:19 4:34
General 9:48 5:50 7:35
When we looked more closely at this coverage the results were less encouraging. During
the primary, the average nightly CCD ranged from a scant 42 seconds on CBS to an even
more paltry 28 seconds on NBC. The picture was better during the general election but
still well below five minutes. NBC did the best. It more than tripled its CCD from 28 to
97 seconds per night. CBS also improved from 42 to 58 seconds while ABC held
constant at 39 seconds of CCD a night (Table 2). All told there were only four nights in
which any network broadcast the recommended 300 seconds (five minutes) of CCD or
more (three on NBC and one on CBS).
3
Despite the improvements from the primary,
these numbers are far from the standards recommended by the Committee and these
findings raise doubt about the effectiveness of voluntary compliance.
Table 2: Average amount of candidate-centered discourse per night
CCD/night ABC CBS NBC
Primary 39 seconds 42 seconds 28 seconds
General 39 seconds 58 seconds 97 seconds
Though ABC had the most overall election coverage, it was the least likely to provide
candidate-centered discourse. CCD made up only 7 percent of its news coverage. NBC
and CBS actually provided more overall CCD even though their total election coverage
was less than ABCs (CCD made up 21 percent and 16 percent of their total election
news respectively).
Because CCD is a function of both candidate sound bites and the news frame of each
story, it should not be surprising based on the small amount of CCD that the vast majority
of stories discussed strategy instead of substance. During the primary CBS was the worst
offender framing 77 percent of its stories in terms of a horse race. In fact during the
primaries all three networks devoted at least two-thirds of their coverage to strategy-
oriented discussions. At best, about one in four stories provided viewers with a discussion
of the important issues in the primary campaign, and that dropped to about one in five or
worse at ABC and CBS. For the general election over half of the stories on all the
networks were strategy-oriented while about one in four were issue-oriented (Table 3).
Table 3: Issue frames by network
Election Frame ABC CBS NBC
Primary Strategy 68% 77% 66%
Issue 21% 17% 26%
General Strategy 63% 54% 67%
Issue 24% 27% 29%
In addition to being a function of the frame of the story, CCD is also a function of the
seconds of actual sound bites in the story. The average length of a sound bite was
consistently quite low and as a consequence the percent of the story in which a candidate
was speaking also tended to be quite small. Of the 198 stories that mentioned a candidate
in the 30 days before the general election, there were only eleven stories in which the
sound bites made up 1/3 or more of the total story time. The average length of an
individual sound bite was just nine seconds (NBC 12, CBS 9, ABC 7). It is not surprising
to find that ABC, which had the lowest proportion of CCD in its newscast, also had the
shortest average sound bite. This suggests an even lower quality of information since it is
hard to imagine what a voter can learn in seven-second bites (no matter how many there
are). Nor is it surprising that NBC, which had the highest proportion of CCD, would have
the longest average sound bite (12 seconds). Thus, while ABC had the most election
coverage it provided less opportunity for candidates to speak and what it did provide was
cut into shorter segments.
It is interesting to note that during the course of the general election both NBC and CBS
made a special effort to provide time to candidates. In the week before Election Day,
NBC invited both candidates to be interviewed live each night on the news. Gore
accepted the offer and Bush rejected it (instead appearing taped on a single occasion).
This was the most substantive example of CCD found on the evening networks. In each
of these segments, news anchor Tom Brokaw asked substantive questions (on Social
Security, education, foreign policy, etc.), and the candidates were given time to answer.
In all there was 1,640 seconds (27 minutes 20 seconds) of CCD just in these NBC
interview segments. That is just a little more than ABCs total CCD for the whole month
(1,170 seconds), what NBC would have had for the month if not for these special
segments (1,261 seconds), and just slightly less than the CBS total for the month (1,730
seconds). Moreover all of these segments were issue-oriented and the average length of
the sound bites for these five segments was about 35 seconds.
CBS also made an attempt to offer the candidates special air time in the form of a single
interview that was then broadcast over three nights during the evening news. This
approach seemed less effective in part because the questions were softer (including a bit
about whether the candidates had prayed to win); there were fewer of these segments
(three instead of five); they tended to be shorter (276 seconds on average compared to
328), and the network edited the interview so it seemed more crafted than NBCs special
interview segments. Still, issues made up the dominant frame for all of these segments,
and the average sound bite length was 17 seconds.
Together in the week before the election, these two networks tallied up 2,467 seconds (41
minutes 7 seconds) of CCD during their special interview segments or about 42 percent
of the total CCD for the three networks for the month. These segments also accounted for
three of the four nights in which any network met the recommended standard of five
minutes a night of candidate-centered discourse.
4
It is important to note that the special
interview time on both NBC and CBS was issue-oriented and since the sound bites made
up over 1/3 of the coverage, the whole story counted as CCD. In fact, of 11 stories where
the sound bites made up 1/3 or more of the total story time, eight of these occurred during
special interview segments aired by NBC and (3 on NBC and 1 on CBS).
While these segments show that it is possible to put higher quality information on the
news, all of the networks (including NBC) still fell well short of the standard of five
minutes a night of candidate-centered discourse for the 30 days before both the primary
and the general elections. These data show that when the networks make an effort to
provide candidate-centered discourse they can vastly improve their coverage, but they
may be unwilling to do so on a voluntary basis.
1
The length of the story included anchor lead-ins. CCD included only direct quotes from candidates (not
paraphrases by reporters or non-candidate sources). Neither election night coverage nor debates were coded
due to the fact that each is atypical and, in the case of election night coverage, inherently about winning and
losing. Newsmagazines such as "Dateline" and "60 Minutes" were included in the analysis along with the
evening news. Intercoder reliability greater than .70 was achieved on all subjective measures.
2
One reason that CBS had less election coverage than the other two networks is that they pre-empted their
weekend news for sports while ABC and NBS did not. CBS total network coverage divided by the 21
days in which they aired the news resulted in an average of 501 seconds of total coverage a night more
than NBC but less than ABC. Thus it is not so much that CBS was airing less election news when they
aired news. It was that they were less likely to air news at all.
3
NBC had an additional four nights where they had over 200 seconds of CCD. ABC had one.
4
The fourth occasion occurred when NBC ran an interview with Joe Lieberman and a story that used
significant quotations from the debates.
References
Cappella, J., & Jamieson, K. (1997). Spiral of cynicism: The press and the public good.
New York: Oxford University.
Digital divide. (2000, December 3). Pew Research Center for the People and the Press.
http://www.people-press.org/online00mor.htm.
Dorsey, T. (2000, November 6). Campaign coverage. The Courier-Journal, C-J.
Hallin, D. (1991). Sound Bite News: Television Coverage of Elections 1968-1988.
Woodrow Wilson International Center For Scholars.
Jamieson, K. (1996). Assessing the quality of campaign discourse 1960, 1980, 1988,
1992. Annenberg Public Policy Center.
Roper Poll. (1999, July 25). Accession number: 0334508. Question number: 11.
Available through Lexis Nexis on line.
Copyright 2000 Annenberg Public Policy Center
All rights reserved
