This report was written by Thomas E. Patterson, Bradlee Professor of Government and the Press at the Harvard Kennedy School’s Shorenstein Center on Media, Politics and Public Policy. It has been adapted from the originally published version.
Election Night offers an unusual opportunity to inform the American people. It is one of those increasingly rare moments when an uncommonly large number of Americans gather in front of their television sets to hear about politics.
Given that Election Night 2012 may feature a very close presidential race, it is well worth returning to some of the lessons learned from the controversial election of 2000, between George W. Bush and Al Gore.
To that end, this article contains ideas for strengthening Election Night telecasts. The recommendations were developed in consultation with television executives and journalists, by examining the internal reviews that the networks themselves conducted after their star-crossed Election Night broadcasts of 2000, and through a systematic comparison of Election Night coverage of the 1968 and 2000 presidential elections — both elections were cliffhangers but occurred at very different stages in the evolution of television news coverage.
Although the recommendations are based on journalistic values, they are designed to strengthen the audience appeal and impact of Election Night telecasts.
1. Anticipate problems with the new and the untested
Election Night always brings a few surprises that confound the coverage. One area that requires close attention is exit polling. Exit poll procedures have been revised with the aim of overcoming the problems that surfaced in the controversial 2000 election, but questions remain.
Allegations of balloting irregularities are also likely to arise on Election Night. Many locations have changed their voting machines and voter verification procedures in response to the Florida debacle in 2000. If the presidential race is close, it is almost certain that balloting procedures will be a large issue on Election Night.
Absentee ballots are another source of Election Night uncertainty. Numerous states have relaxed their absentee ballot requirements in the past decade, such that in many states a significant percentage of votes are now cast absentee. States and localities differ greatly in how quickly absentee ballots are counted and added to the raw vote totals. (Information on absentee voting and other balloting issues can be obtained at the state level from the secretary of state’s office and at the community level from the local election office.)
The presidential contest is the big story on Election Night and naturally gets more attention than do the congressional or statewide races. However, our comparison of the 1968 and 2000 Election Night network telecasts revealed that the presidential race now gets substantially more attention than it once did while congressional and statewide races get substantially less coverage. The presidential race accounted for slightly more than two-thirds of the airtime in 1968 and a full four-fifths of the airtime in 2000. Congress is a co-equal branch of government, and partisan control of Congress is a key element in a president’s policy success. Congressional elections should get more attention on Election Night.
Local stations must provide much of this coverage. The networks can be expected to concentrate on the presidential race and to a lesser extent on the Senate races. The responsibility for in-depth coverage of House races has effectively devolved to local broadcasters. If they don’t provide it, viewers are unlikely to hear much about the House of Representatives. One obstacle to such coverage is that the large majority of House races are now so one-sided that there is no large story to tell about them. However, significance can sometimes be found in the national consequences of local races. For example, reelection in a House race may position a local incumbent to take over the chairmanship or the ranking position on a House committee or subcommittee that is important to the nation or the district. Whether this happens will depend on what is happening in districts elsewhere — for example, is the current senior committee member retiring or facing a tight race? A ranking committee position can mean more federal grant-in-aid money for the local district, which is always a great local story. Such possibilities, if recognized in advance, can be built into the coverage.
3. Give seasoned correspondents more opportunities
Just as evening newscasts have become more anchor-centered, so have Election Night telecasts. Our comparison of the 1968 and 2000 network telecasts found that anchors now get substantially more airtime than they once did while correspondents get fewer on-the-air opportunities and, when they do speak, get far less airtime in which to present their arguments.
Yet, audiences crave the insights that experienced reporters can deliver — keen appraisals of what’s happening and what can be expected. Seasoned reporters should be brought more fully into the Election Night telecasts and, while on the air, given adequate time to make their points.
For news organizations with sufficient resources, it can make sense to assign a seasoned journalist the task of monitoring the Election Night coverage for mistakes or soft spots. Journalists directly involved in the telecast, whether on-the-air or at a decision desk, are swamped with demands and information. Errors of omission and commission occur frequently, and the monitor’s role would be to blow the whistle when they do.
At some local stations, seasoned political reporters are now in relatively short supply. The trend toward “softer” news has reduced the number of journalists who cover politics closely. Local newspaper reporters and political officials can in some instances be employed effectively on Election Night. Even top elected officials are sometimes available. Because most governors and U.S. senators do not run in a given presidential election year, they may welcome the opportunity to appear on television in a major market on Election Night. Smaller-market stations could include a former member of the U.S. House on their broadcast team.
In earlier elections, as vote returns in a state trickled in, journalists helped viewers to understand that early returns are not necessarily indicative of the final results. In the process, they also taught viewers about a state’s voting patterns — for example, that upstate Illinois is heavily Democratic while downstate Illinois is heavily Republican. Likewise, exit-poll projections must also be explained to viewers if they are to interpret them properly. Some news organizations provide this explanation but our analysis indicates that others do not. A detailed explanation of exit polling should occur early in the Election Night telecast and, because of audience turnover, should be repeated in abbreviated form every hour thereafter. You might also consider posting an exit-poll primer on your website and using a crawl to direct interested viewers to it.
5. Establish guidelines for making exit-poll calls, and stick to them
Some of the problems associated with exit polls have occurred because television organizations either had unclear rules for their use or ignored their pre-established guidelines. Although different news organizations might reasonably establish different guidelines (for example, on whether to withhold a call until all the polls in a state have closed or whether to withhold it until all but a specified percentage have closed), a news organization should stick to its rules even if another news outlet makes the call. Of course, no news outlet wants to withhold a call that others have made, and every outlet likes to boast “you heard it here first.” However, there is no evidence that viewers are attuned to this competition. On the other hand, an erroneous call rightly gets the attention of critics as well as viewers and, in the worst case, can adversely affect turnout or the legitimacy of a candidate’s claim to victory.
Predictably close races deserve exacting scrutiny. No system of estimation based on exit polls and initial returns is foolproof. A keypunch operator’s 40,000 vote error in Duval County contributed to the faulty early prediction of a Gore win in Florida in 2000. Any indicator that something is amiss with a projection should be taken seriously. The only sensible rule on Election Night is getting it right, not getting it first.
Television organizations should also resist the pressures that can develop from leaks of the early results of the exit polls. Although the early results enable news organizations to better plan their Election Night coverage, they are also a source of mischief. The early results are invariably leaked to outside parties and then find their way onto web logs and talk radio, where pundits freely announce the results. Nevertheless, television news organizations should stick to their pre-set guidelines. For one thing, early exit-poll results can be misleading. In 2000, there were roughly a dozen states where the early exit-poll numbers were projecting the wrong winner. Early numbers reflect the preferences of early voters as opposed to the preferences of all voters and can lead to inaccurate predictions.
6. Use language that fits the numbers.
Viewers need to know at all times the weight to be attached to exit-poll projections. Precise words (both spoken and on the screen) are the only way to meet this requirement. Calls based on exit polls, sample precincts, or partial returns are not definitive. They are properly described as “projected” or “estimated.” Such words should be employed until actual returns make the outcome a virtual certainty. Until then, there is no useful journalistic purpose in unconditional statements such as “has won” or “is the winner.” Even when such statements eventually prove to be accurate, they set news organizations up for the complaint that they prejudged a race or a state.
Journalists should distinguish between races that are not called because they are “too close to call” and ones that are not called because of incomplete information. These are very different situations, and viewers should be told clearly which situation exists. Clarity in these situations will also protect on-the-air journalists the embarrassment of describing a one-sided race in which the numbers are slow in coming as “surprisingly close.”
7. Breathe life into the numbers
By itself, the vote reveals only the decision that voters have made. Exit polls supplement this information with evidence on the relationship of the vote to people’s personal opinions and characteristics.
Our study of the 2000 Election Night telecasts found, however, that most exit-poll analysis is shallow. Typically, the journalist did little more than recite a set of numbers that viewers could already see on the screen. In one case, for example, a network correspondent indicated that exit polls showed women had voted 10 percentage points more Democratic than had men, concluding that the difference represented a “gender gap.” Viewers were not given a cogent reason for this gap, nor told of its larger history or its political significance. Numerical distributions of this type should be accompanied by adequate explanations. Although experienced journalists will know the deeper story behind the numbers, many viewers will not unless they are told.
There is a choice to be made on Election Night in the presentation of exit-poll numbers. Viewers can be provided more numbers and shallower explanations or fewer numbers and fuller explanations. Audience research suggests that both viewer interest and viewer learning increase when the explanations are fuller.
Election Night telecasts go on the air before the polls have closed in most states. Just as the legendary mayor James Michael Curley urged his constituents “to vote early and often,” Election Night telecasts through words and banners should regularly remind viewers that polls in their area may still be open. Some telecasts already do this, but our analysis indicates that the practice is uneven. Local telecasts must be part of this effort. Such announcements will spur some citizens to vote. They also mitigate the perennial charge that early election calls serve to depress turnout in those areas where the polls are still open.
9. Avoid the “storyline” trap
Often, journalists develop storylines as a way of giving viewers, and perhaps themselves, a better handle on Election Night developments. An example is the statement that “Democrats need to hold onto their Senate seat in North Carolina if they are to have a chance of taking control of the Senate.” If the Democrats then fail to win the seat, the inference is that they have lost any chance of controlling the Senate. Such storylines can get embedded in the night’s commentary and mislead or confuse viewers. No such storyline in memory had more influence on the public than one voiced early in 2000 Election Night telecasts — the notion that Gore had to win Florida if he hoped to win the presidency. While Florida ultimately decided the election, this storyline was put forward well before the decision had narrowed and was not, in fact, true. Gore would have been elected, even losing Florida, if he had won any additional state. The Florida storyline, combined with the networks’ late call that Bush had won in Florida, contributed to the perception among some Americans that Gore, through his post-election legal challenges, was somehow trying to “steal” the election.
Journalists should also avoid words that send potentially misleading signals that might influence viewers who have not yet voted. An example is the overuse of the word “expectations” to describe how a presidential candidate or political party is doing in states with early poll closings. To say that a candidate or party is doing less well than was “expected” suggests underperformance. Yet, the “expectation” itself might be mistaken — the result, for instance, of faulty analysis or poll data. Moreover, even if the “expectation” is soundly based, it is not necessarily indicative of what will happen elsewhere — factors that drive the vote in, say, the earlier-poll-closing states of the deep South may be quite different than those driving the vote in the later-poll-closing states of the upper Midwest.
10. Enlarge the explanations
Election Night coverage now employs the same fast-paced format as the evening newscasts. Our analysis indicates that sound bites on Election Night telecasts have shrunk to an average of 15 seconds or less. As a result, many reports skim the surface, as on-theair participants jump from one comment to the next. While much of the information communicated on Election Night can be presented in short spurts, brief statements are rarely adequate when explaining the results.
Moreover, the explanations themselves need rethinking. Our comparison of the 1968 and 2000 Election Night telecasts found that journalists are now far more inclined than in the past to explain election outcomes primarily in terms of the candidates’ strategies rather than in factors such as policy issues and partisan loyalties. Strategy is undoubtedly an important part of election outcomes, but no serious analyst would place it first on the list.
11. Look forward, as well as back
In their Election Night analysis, journalists tend to look backward, toward what happened during the campaign, rather than forward, toward what can be expected during the transition and governing periods. In 1968 and in 2000, fewer than 5 percent of the broadcast segments contained even a passing reference to the election’s policy and political implications, and only 1 percent did so in a substantial way. Yet, aside from winning and losing, the crucial question on Election Night is the election’s larger meaning: What does the outcome portend for the nation? This question should be explored more fully on Election Night telecasts.