Reprinted from Campaigns and Elections American Style,
Westview Press, 1995
In the spring of 1983, the ever precocious Democratic pollster Pat Caddell stated in congressional testimony, "Today, the opinion pollster has become the strategic center of most campaigns". In 1984, Republican pollster Chuck Rund sat before a monitor flashing up different electoral vote scenarios based on Dick Wirthlin's state-by-state-polling for President Reagan's re-election and said, "Gee, I feel like a voyeur into the American political psyche." And in 1992, Newsweek reports on an opinion research project designed and driven by Bill Clinton's campaign pollster, Stan Greenberg, along with general strategic consultant James Carville and media maven Mandy Grunwald, which was destined to change the public's perception of Bill Clinton -- and ultimately lead to his election as President.
Clearly the art and science of modern political polling has become the major influence in strategic decision-making in modern U.S. political campaigns. This uniquely U.S. profession was not always at the center of our election campaigns. The importance of analyzing the electorate via public opinion polls has grown as the need for direct feedback from individual voters has intensified. Political parties have become less influential; the power of traditional interest groups has atrophied; and television, telephone, and mail technologies have improved and begun to dominate direct communication with the voters.
The polling profession, over the past 25 years, has improved because of the steady build-up of experience gained by pollsters provided by the numerous, ever-present elections in the U.S. As experience was gained, improvement in communication technology, which was changing political campaigning in general, was also increasing the opportunity to use polls in this dynamic mileau. To understand how polls eventually moved to the epicenter of American campaigns, it is necessary to trace political polling's short but dynamic history.
In the beginning there was George Gallup, the ultimate psephologist. His disciples were Elmo Roper, Arch Crossley, and Hadley Cantril. But soon there came a generational split, led for a short time by Louis Harris and followers like Oliver Quayle, John Kraft and Charles Roll.
This description succinctly outlines the pre-Columbian history of American political polling. As most people acknowledge, the public pollster (now called media pollsters) preceded the private political pollsters by nearly 30 years. In the early 1930s, George Gallup believed that his desire for direct democracy called for public information and policy evaluation that was not filtered through the economic elite. Gallup first applied this new concept of going directly to the voter by polling for one of his mother's elections for local office in Iowa. Shortly thereafter he began to apply his new technique to predicting the results of elections for dissemination by the public media.
Throughout the 1930s and 1940s, Gallup and two or three breakthrough social scientists conducted numerous polls for media, while keeping their new businesses afloat with marketing research studies for private companies and advertising agencies. While a number of the new pollsters tried to interest candidates and political parties in using polling as a strategic tool for campaigns, most of the opinion polling remained public, paid for by the media for general publication. Then came 1948 and the year of "President Dewey."
Contrary to most accounts, the highly publicized "miss" by Gallup and others in 1948 did not automatically discredit polling after Truman/Dewey. The technique had proved so accurate in other instances and so enlightening to business/political decision-makers that there remained a market for the product, in the field of product polling, but also for public polling on public policy issues. However, parties and major political candidates continued to use polling only idiosyncratically, even then not for strategic purposes.
In the early 1950s, most politicians continued to show little interest in using political polling. A few political players would hire a pollster to conduct a quick poll late in the campaign just to "see how things were going," but the technique was peripheral to actual campaign operations.
During this post-war period of consumerism, however, some marketing researchers were beginning to apply "psychology" to their commercial studies and help marketing departments plan strategy. While the clients in the political field were slower to move in this direction, a few pioneers kept trying -- Roll with Crossley and Roper and his employees, Harris and later Quayle and Kraft.
Harris left Roper shortly after the 1956 Democratic Convention to begin his own research company, and one of his first clients was U.S. Senator John F. Kennedy. He continued working with the future president up through the successful 1960 primaries and general election, bringing with him Quayle and Kraft from the Roper organization.
On the Republican side, Charley Roll had a small New Jersey firm going but only handling a few clients. George Romney's Executive Assistant Walter DeVries, later of The Ticket Splitter fame, was working with Market Opinion Research to help GOP candidates in the state of Michigan. But it was Harris who truly spawned a new industry or profession in 1960 with the publicity surrounding his polling for the new President. Two years later, however, Harris announced that he was eschewing private political work and would, from 1962 on, conduct only public polls (for his newspaper column) and private marketing polls.
By 1964, Democrats Quayle and Kraft had left the Harris organization and begun what was to become a powerful and influential cottage industry in American politics. They were followed by a number of smaller operations on both sides of the partisan fence, including this author. In 1965, supporters of Ronald Reagan had asked an economics professor from Utah, Richard Wirthlin, if he could conduct polls to help the former actor in his campaign for governor of California.
In the mid-1960s private political polling began to take hold. In 1964, my first client (and only client at the time) wanted to run for statewide office in Florida. After two polls, I suggested it was an uphill race against the entrenched incumbent, and he finally quit the race. The firm struggled for a few months until we picked up our second statewide client -- a state senator running for governor in a six-way Democratic primary.
Our client's plan was to run second in the primary and win the nomination (tantamount to election) in the run-off. In Florida and much of the South, this was the traditional strategy of moderate-conservative candidates -- to get the liberals and ultra conservatives out of the race early, and narrow the contest to a head-to-head with a single moderate-liberal urban opponent. The number two finisher in the primary, often the "legitimate" moderately conservative Democrat, usually won these run-off elections before 1966.
The first primary poll showed our candidate running third -- behind the moderate-liberal favorite, but also behind the big city, ultra liberal mayor from Miami. When I presented the expected third place finish to the candidate/client, all he said was "Who'd you talk to -- all blacks" (he used another term). He promptly rose from his chair, left the room, and I never saw him again during the primary campaign.
We continued to poll for the public relations company handling his campaign, and finally showed him a likely fourth, which was the ultimate outcome. After the primary, the candidate called me to his office and the first words out of his mouth were "How'd you know that ...?" A new convert for using polling in campaigns had been made, but more importantly, the publicity of our accuracy had helped to launch a new polling firm.
Other pollsters from this prepubescent period have similar stories, usually told with different accents -- from New York to Pennsylvania to California. But the stories generated the same questions: How can you challenge the existing party or political elite's structure and thinking about a campaign? How can a poll be so correct? And how can it help a campaign win?
In 1966, now veteran consultant Joe Napolitan polled for Milton Shapp and helped an unknown millionaire defeat an entrenched Pennsylvania Democratic machine for the gubernatorial nomination. In the same year, Dick Wirthlin used the new technique to help fashion the Reagan gubernatorial victory in California. Regular polling helped that ultra liberal from Miami (Robert King High) knock off Haydon Burns, a sitting governor, in the Democratic primary, and DeVries continued to use MOR's surveys in Michigan to re-elect George Romney.
There was not a cataclysmic event which caused political polling to catapult forward to dominate the political dialogue in American election campaigning. It was, rather, a series of fairly compressed events during the middle political cycles of the 1960s -- when communications technology was changing the face of the politics of America, which caused polling to become necessary and critical to the operation of a modern campaign. It was at the end of this period that polling became a required line item in the campaign budgets of many major races, but many of us (pollsters) were still arguing with candidates, campaign managers, and fundraisers to allow our data to help guide the strategy of the campaign, rather than simply evaluate, too late, the impact of key campaign decisions.
In 1960, Bobby Kennedy was reported to have cut Lou Harris short during one of his analytical poll presentations and said, "Just give us the numbers, Lou." The movement from number cruncher to strategic counselor had changed little during this pre-history era of the polling profession. But polling was on the political map.
The change in U.S. pollsters' role from last minute evaluator to early strategic initiator was to begin soon enough -- during the first true era of modern campaigns (1967 - 1978). In the second era (1979 - 1988) polling was to become the strategic centerpiece of the modern American political campaign. In the 1990s the polling industry is continuing to refine its strategic research techniques to meet the changing needs of American political campaigns -- but not without some conflicts and controversies.
During most of this early era of the 1970s, there were only three or four national polling firms conducting most of the polls for candidates of each major political party. This was due partially to the fact that a fairly large infrastructure and network was necessary to conduct the type of polling used in those days, and partially to the fact that the market was not yet producing enough dollars to support many additional research firms.
During the early part of this period, only about half of the "top of the ticket" races (U.S. Senate and gubernatorial) were using pollsters, and less than a fourth of the congressional candidates commissioned polls for their campaigns. Very few legislative contests or local races even considered polling. Near the end of this era, however, most of the federal and all of the gubernatorial candidates were using pollsters extensively. Accordingly, major political research firms grew in size and influence.
The pollster and polling function during the early 1970s still served largely as the "provider of a unique set of numbers and data" to the media consultants and direct voter contact functions in the campaign. But the polling function had won a seat at the strategy table by the middle of this era, which allowed it to become increasingly more central to campaign decision-making. As many have often said, and rightly so, "information is power."
However, polling in the early-mid 1970s was largely conducted using in-home, face-to- face interviews. This procedure, while collecting extensive in-depth data for analysis and assistance in developing a campaign plan, was relatively expensive, very slow, and not as accurate as the telephone polling to come. This meant that numerous polls with the quick turnaround (time from beginning of interviewing to reporting of numbers) needed to manage strategy at midcampaign and endgame were costly, impractical or untimely. This was one of the main reasons why pollsters and their product did not move to the center of strategic decision-making sooner. At the same time, there was less need for an increased volume of data on a quick turnaround timetable. Media producers were still shooting 16mm film which took time to transfer and edit, TV stations couldn't (or wouldn't) change spot traffic daily, Federal Express was in its infancy, and there weren't satellite feeds or broadcast FAXs -- yet.
Near the end of this era, however, polling in campaigns began to change. Caddell polling for presidential candidate Jimmy Carter and Robert Teeter, polling for Jerry Ford, both used some telephone surveys at the end of their presidential races, and by 1978, most of the statewide contests were at least using the telephone for trend polling or panel-back surveys -- and were just beginning to conduct rolling-average tracking.
In the initial two election cycles of this period, there was an explosion of central telephone facilities with 20-100 telephone stations at which large sample surveys could be completed in a short time. Some of these were built by the major pollsters themselves and some were independent contractors hired by pollsters who had no phone bank. By 1980, most initial benchmark surveys were being conducted by phone as experience was gained with the different questioning techniques used on the phone, and as the in-house interview became even more impractical -- travel costs rose, the independent interviewer network atrophied, and some urban (and rural) areas became completely inaccessible to the survey researcher.
These basic improvements in technology either directly or indirectly changed the polling industry and how polls and pollsters were used in political campaigns.1 First, more frequent polls were conducted by major campaigns (all statewide and congressionals), but down ballot campaigns such as mayors, judges and statewide constitutional officers could also now afford polling.
During this period, the campaign committees of the U.S. House and Senate became volume purchasers of polls for their targeted candidates. And, to a lesser degree, major PACs began commissioning polls for their "favorite" candidates.2 Finally, during the end of the 1980s the leadership of major state legislative bodies became volume purchasers of polls for their marginal districts as part of a new mode of centralized strategic planning and operations for these campaigns.
Second, the increased availability of telephone interviewing facilities meant that establishment of new, independent polling firms was more feasible than in the past. Experience in questionnaire construction, minimal knowledge of analytical techniques and a personal computer was all someone needed if they could purchase samples, interviewing time, and data entry services. During the 1980s many of the younger people, who had built some years of experience at the six to eight established firms which had previously controlled much of American polling, established their own firms. This trend was augmented when a number of young academics, who had been teaching at a number of universities and "doing a little consulting on the side," saw an opportunity to work in this exciting field full time -- and with little financial investment. So the number of national firms doubled and then tripled in this decade - - not to mention the introduction of a number of excellent state or regional polling firms.
Third, the increased use of telephone interviewing, a technique that produces mainly quantitative data, brought about the need for a more qualitative, "touchy-feelie" type of arrow in the researcher's quiver. So, during the late 1970s, focus groups were borrowed by the political pollsters from the market researchers.3 It was in the early 1980s, however, that the use of focus groups expanded dramatically and were often a required companion to the telephone benchmark poll. In most cases, the principals in the polling firm became the focus group moderator and analyst. While some may argue these pollsters lacked training to conduct this research function, this argument is offset by the pollster's full understanding of their client's political situation and knowledge of what they needed to improve the more quantitative research phase to follow.
The combination of a larger, more diverse and competitive polling industry paired with the rather quantitative style of the telephone survey led to more R&D on research techniques during the 1980s. In short, more professionals were searching for a competitive edge -- not only in marketing themselves but in helping defeat their client's opponent, who was also using opinion research techniques.
During the latter part of this era different research products were being thought about and tested -- pre-post tests of on-air TV spots, the dial measured focus group, testing of radio spots over the phone, content analysis of verbatim responses for message development, more refined statistical techniques in data analysis, etc. As private political polling moved into its third decade, many of these techniques would become standard.
One other development during this period had a significant impact on both the polling industry and U.S. campaigning in general. That was the expansion of public polling during campaign periods -- not only the national polls but the conduct of multiple polls during state election campaigns by a single media outlet or a consortium of outlets. Not only did it make everyone (the public, press and politicians) more aware of polls, sampling, questioning techniques, etc., which (a) demystified the "black box" concept of a pollster's analysis, but (b) also made the pollster even more important to campaigns in order to interpret and help manage the impact of the media polls.
This decade-long combination of events and developments placed the pollster at the center of campaign decision-making. The speed and reasonable cost of polls brought their increased use as planning tools for strategy and message development as well as the monitoring of tactics. Focus groups and other polling techniques became the standard tests for expensive media products. Most of the pollsters had extensive experience drawn from numerous campaigns over a number of election cycles. And the increased use of polls by the media had forced campaigns to account for and monitor public opinion.
Caddell's remarks in his 1983 congressional testimony were rather prescient and it is now clear, based on reports of polling work in the Clinton presidential campaign, that the pollster/polling function is at the center of strategic planning and tactical decision- making in most major campaigns in America.
This trend toward the polling function (and the pollster as a personification of it) being at the strategic center of campaigns is not likely to be reversed. While no longer the only campaign professional with access to and understanding of data, the pollster can still best interpret the opinion research data and has the most varied experience in helping campaigns determine the strategic imperatives for a campaign to win.
Questions have been raised whether there is an ethical concern or conflict-of-interest in the pollster helping develop overall strategy while providing the initial data on which this strategy is based and then monitoring how well this strategy is working. My answer to these questions is "No," mainly because the job of building strategy is never the pollster's alone. It utilizes a number of different data bases (of which the polling data is probably the most important) and is a product of a campaign team -- among others, the candidate, campaign manager, a media consultant and often a fundraiser. Plus, it would seem unwise to discard the input of the most analytical and often most experienced member of one's campaign's inner circle.
It is possible, however, for a pollster to become so enamored or involved in the consulting aspect of the campaign that his/her attention is diverted from designing and executing the most creative and rigorous research design for the client's campaign. While an experienced counsellor to the campaign, the pollster is also the best instrument for producing specific and accurate opinion research data for that unique campaign; anything which unfocuses the pollster on this task is dangerous. This potential problem, of course, can be partially solved by expanding the size of the research firm (while maintaining quality) and/or the dollar size of the contract.
Regardless of how integrated the pollster is into the campaign team and regardless of what level of election (presidential or state representative), every modern campaign demands that the pollster provide major input to two critical questions:
There are other important questions on which some of the more experienced pollsters provide advice (What are the best channels to use in communicating with our targets? or What intensity, weight or emotionalism do we use in communicating with the targets? etc.), but targets and message are the critical elements which modern American campaigns need to address in order to develop a winning strategy. In other words: in order to win, what must be communicated directly to which group of voters (most of whom are not seeking the information) with a limited budget.
The question is : Is this message/target responsibility the same for a pollster working for a presidential candidate as well as any other level of election? The answer is: Yes. After that, the similarity of the pollster's role (like many other campaign professionals) is very different in a run for the Presidency. First, this contest is for the leadership of the strongest, most influential country on earth -- neither a Governor's race in California or Prime Minister contest in Canada is even close in terms of the raw power of the office. The knowledge of the potential power of the office being sought affects all political participants, of course, not just the pollster. Second, the campaign operates in a fish bowl with more constant and total press contact than any other American election contest. Third, more powerful people are interested in the race -- the top leaders of all special interests want to help or hurt you or your candidate. Finally, the campaign is geometrically larger than any other contest that any first-time presidential pollster has ever dealt with. And that's just the general election campaign.
The presidential primaries are different and in some ways more difficult for pollsters of the leading players. In these primaries, a pollster may do very little specific polling unless the candidate is well heeled, but still may be required to provide the campaign with advice on message and targets. If polling can be afforded in a primary effort, the pollster must be prepared to help keep the national message focused while providing the campaign with message and target direction for three or four different, ill-defined state primary electorates. Next week, the tasks will be the same but most of the states will have changed.
After the primaries decide the candidates, the polling firms then have the largest set of clients they'll ever have -- 50 different state elections and a national campaign (in reality, it is usually only 15-25 states which are in play in a given election). Not only must the pollster help the campaign devise and operationalize its national communication strategy, s/he must develop a research plan and execute it for at least the target states; this means helping each state campaign develop a campaign plan and then tracking the effects of each state's execution.
While it has been different for Democratic and Republican campaigns over the years, general election polling has usually, since 1980, taken on a structure which put the main polling firm (and winning) pollster at the center of all polling with special responsibility for national message (and possibly a few key states) and then dividing data collection and first-cut analysis duties between other party pollsters -- the tracking to be conducted according to a pre-determined plan. The secondary pollsters would feed their data and analysis to the major pollster, who would usually advise the state coordinators and put the actual data in some type of large model or simulation. In short, techniques are the same but the volume of work and pressure are unique to a presidential race.
All American electoral contests today attempt to use some type of public opinion research technique to gauge the mood or collective mind-set of the electorate. This information and proper analysis is simply essential to success in our present political system. Most campaigns with more than 10-15,000 voters and more than $50,000 in campaign dollars to spend will use some type of professional, modern research technique to guide them in developing or refining their campaign strategy.
Virtually all gubernatorial and U.S. Senate campaigns now rely heavily on the pollster as one of the two or three major consultants/operatives of their campaign brain trust. These campaigns will require significant amounts of personal consultation (time) from the pollster in addition to using many of the different research techniques in the pollster's repertoire: This cost to the campaign is usually 5-10 percent of the total campaign budget.
Most congressional contests also now use polls -- often to determine whether or not they need to mount a campaign in what is often considered to be a "safe" district. In only about 15-20 percent of the congressional districts is there some degree of competitiveness where a true campaign is run; although in 1992, the Democratic Congressional Campaign Committee (DCCC) reported nearly 150 districts "severely in play" following scandals, redistricting, and retirements.
These competitive congressional contests use the polling function much like their statewide counterparts -- except they usually have a smaller budget and therefore less need (or ability to pay) for the varying types and volume of data used in a statewide contest. Most of these congressional races conduct early benchmarks and some type of trend or tracking polls, but often the frequency of their tracking is less than in statewide contests, and often they do not use all of the specialized techniques and studies. For example, only about 25-30 percent of the targeted congressional races used focus groups in this past cycle; even fewer used any of the new ad testing techniques. These techniques remained, however, the standard fare of the larger budget, statewide campaigns.4
State legislative races are similar to congressional races except their budgets are even smaller and therefore their marginal, competitive races are even harder pressed to afford these modern research techniques and therefore the advice of an experienced pollster. Yet, even in many of these "little" legislative races there are changes occurring which will impact these contests in the future.
The cost of these different research techniques (and other campaign services) are leading state legislative leaders (Speakers, Presidents of Senate, minority leaders) to set up separate funds, raise money, and provide campaign services, including polling services, to the critical group of marginal districts they deem might make a difference in partisan control of these legislative bodies. In some of the larger states these state senate and representative races now use all the same polling techniques and advice provided to a major statewide campaign -- except they use them more like party leaders in a parliamentary system of government.
Local races -- mayor, county commissioner, county treasurer, etc. -- have expanded their use of polling techniques to help plan and operate their campaigns. The variation in use remains fairly wide, however, based on the size of the campaign budget and partisanship in the community. Clearly, mayors' races in all major cities use polling in their campaigns, but the techniques and advice are too expensive and couldn't be used effectively in a county commissioner's race in Saline County, Kansas with a campaign budget of $4,700.
Most American political operatives and consultants think of our modern election campaigns, whether major statewide efforts or local races, as dividing into three particular phases: early, when the campaigns are getting organized, strategy is developed, staff is hired, the candidate is reviewing his or her issue material, but the vast majority of voters are unaware of any candidate or campaign activity; midcampaign, when the candidates attempt to improve their recognition and popularity and each one begins to position himself or herself for the final run; and endgame, which is the period in which the campaigns spend most of their communication dollars, the voters begin to pay attention, and the candidates compare each other and each other's records, while "asking for the order."
During each of these campaign periods, the pollster has become a vital participant in campaign decision-making, but s/he plays a slightly different role, often using different techniques in each period. This relationship of activity to the campaign has changed little over the first two previous eras of political polling -- but in all three time phases the pollster's activity level and importance to the process has intensified. As voters have continued to drift from the anchors which provide consistency in voting behavior, something and someone is needed to interpret which way they will drift -- and whether the tide or the wind will have the most influence on that direction.
Early in the campaign the pollster attempts to become immersed in the campaign itself -- reviewing clip files, studying electoral and demographic targeting, reading any previous strategy memos, and often conducting a series of structured interviews with friends, fundraisers, party officials, and existing campaign or office staff. A most important element of this very early period is for the pollster to strongly urge the campaign to conduct what is now called opposition research -- if this process has not already been started.5 This whole pre-polling process is an effort to (1) provide the campaign with a discipline that information must precede decisions and (2) allow the pollster to become thoroughly familiar with the candidate's record, family and background, and staff, as well as his or her party's officials, opponent, and constituency -- in order to conduct the best benchmark poll possible.
Often the pollster is the first one on board a campaign who has both a gestalt view of the process and a penchant for data-driven decision-making -- and therefore no unique personal agenda within the campaign. This immersion process is critical to moving a campaign at the correct pace and in the right direction. When this process is shortcut, it is usually to the detriment of the benchmark poll and campaign plan.
All of this activity is pointed toward developing, executing and analyzing a benchmark poll for the campaign that is enhanced both by the pollster's experience in other election campaigns and also by sensitizing him or her to the uniqueness of this particular race in this particular constituency. Often a set (two to four) of focus groups will also be conducted before the benchmark. These will be designed to listen to the voters "brainstorm" about the issues, candidates, and their relevance to their lives and to test competing hypotheses or assumptions which have arisen in the immersion process. At a minimum, these focus groups will be used to determine how to best measure the voters' priorities among a number of different competing arguments, messages and hypotheses regarding strategy.
The benchmark poll is the major research product in this early phase. This is a lengthy poll (90-100 items or questions) usually conducted with a fairly large sample, 500 to 1200 likely voters in statewide races and 400-500 in congressional races, depending on the size and heterogeneity of the constituency. After execution of the survey, which includes both open-end and fixed-answer questions on mood, personalities, issues, partisanship, and candidate choice, the polling firm provides a very detailed analysis and makes a full presentation to the key players in the campaign. This is usually the most expensive single research project in the campaign. Below are some future trends products and types of research designs that are or will become a part of the pollster's research repertoire:
Future Trend: Because early in some campaigns means 12-18 months before an election (an incumbent U.S. senator, for example), a number of professionals, including Tom King, a Democratic media consultant and Rob Schroth, a Democratic pollster, believe that this is too early to conduct such an expensive poll -- too far in advance of the campaign developing communication products for the voter. They suggest that conducting a mini-benchmark early with a slightly smaller sample and shorter questionnaire would be sufficient for early strategic planning and then conduct the larger benchmark closer to initiation of actual campaign communication, and therefore closer to the time the voters will be paying attention. The cost saving can be as much as one-third the cost of the benchmark. Everyone, however, agrees that some form of benchmark poll is necessary as early as possible in a campaign.
Future Trend: Some pollsters, including this author, have developed programs using the content analysis technique on voters' actual responses to open-end questions including both quantified analysis and the display of actual verbatim quotes by message element -- as a substitute for early focus groups. While this process gives up viewing the interaction among voters, it has the advantage of a quantified measure, spreads the investigation across a larger portion of the constituency, and, being a by-product of the benchmark poll, it is less expensive than a set of separate focus groups.
The pollster's input to the campaign plan at this early juncture is vital. The existing benchmark poll and these new techniques will guarantee that campaign plan will be based on the most sensitive analysis of voters' opinion about a particular race and set of candidates.
Midcampaign is usually the longest period in the campaign. It begins after a campaign plan is written, discussed, and begins to be executed. Over this period of time there are three different roles which the pollster plays in the campaign: a constant advisor or "reality check" on all campaign decisions which in any fashion might have an impact on the voters -- and this doesn't leave out much in a modern political campaign; a monitor, through trend polling, of the success of the campaign's tactics in achieving their strategic goals -- at separate time phases of the campaign; and an evaluator of different communication elements the campaign produces, from the candidate's stump speech to a tabloid to a television spot.
As already noted, the pollster is a key member of the campaign's strategy team. It is during this building period that the pollster's regular advice on how specific decisions which can affect the electorate is critical -- Should we attend the teachers' convention or not? Should we return the NRA questionnaire (not what answers to give)? Should we hire an out-of-state operative to do press relations? Should we push the existing budget bill to the floor of the legislature for an early vote? These are the types of questions which come up during midcampaign. The pollster should not be the last word on these questions, but he/she should be consulted for insight on how the voters react to the various options posed by these questions. And finally, the pollster often serves as the disciplinarian in the campaign -- the force which attempts to keep the focus of the campaign on the messages set forth in the basic strategy plan -- and not be thrown off track by any of the myriad of decisions that campaigns are forced to make. In major campaigns the pollster is usually in daily contact with the campaign leaders; in congressional or lower level campaigns it is usually the campaign manager or candidate who brings the pollster in on key questions.
The pollster or polling firm usually conducts one or more trend polls during this midcampaign period. A trend poll usually uses a similar sample size as the benchmark, but the questionnaire is much shorter -- usually 25-35 items or questions, meaning the analysis is more direct and brief. The purpose of each trend poll is (1) to simply measure what basic changes have occurred since the benchmark, (2) to "go fishing" in case some issue, mood or message is beginning to move within the electorate, and (3) most importantly, to specifically test the progress toward the goals of the campaign during the interim since the last poll -- goals which specifically relate to the voters' (or the targets') opinions. When the trend poll results are combined with evaluative measures on fundraising goals met, staff hired, volunteers in place, endorsements achieved, etc., the campaign can make a full evaluation of itself.
The second aspect of this evaluator role is the polling firm's ability to work closely with the communications director and the media consultant. Hopefully, the benchmark analysis, open-end comments, special message analyses, focus group reports and targeting information have provided the creators of communication with the initial input on voters' opinion and feelings that they need to design persuasive media. The next step is to test voters' reactions to the concepts, copy, symbols, and specific messages the media producers have operationalized for the press, television and radio. In the early era of polling in campaigns this was not a function performed by the polling firm, but when focus groups came into use in the early 1980s, one use of the technique was to test the potential effectiveness of ads -- at least to have the voters catch glitches or the ambiguities of a symbol or slogan before that communication piece went out to the voters. Focus groups were helpful, but it was still a qualitative tool that was not wholly satisfactory to the pollsters or media producers.
It is in this area where the newest era of political polling has begun much of its innovation.
Future Trend: In 1988 Lee Atwater used standard focus groups in New Jersey to figure out that crime was an underlying fear among many voters, that Democrats were perceived as weak on the issue, and that racism still existed in America. Forgetting morality for a moment, it should not have been difficult to put that analysis together using these basic focus groups. In 1992, Greenberg-Carville-Grunwald faced a more difficult problem for Bill Clinton, and they used a new technique which allows a group of preselected voters (about 50) to push a seven digit dial every ten seconds to demonstrate their feelings while watching a piece of video about their candidate. By later superimposing a quantified trend line on the video, the analysts were able to determine what symbols, issues, facial expressions and body language of the candidate were most pleasing (and not so pleasing) to these voters.
No one should deny that in early June of 1992 Bill Clinton faced a much larger problem with his personal image than even George Bush faced during that same period in 1988. The pollster, who had seen this research product used very occasionally in the preceding cycle, knew that the sensitivity it added to the measurement and analysis of his candidate's problem was what was needed. This technique will see expanding use in major statewide campaigns.
Future Trend: I have adapted a technique from marketing research which entails testing television spots while people view them in their own homes. This requires recruiting a panel of 100 voters to watch a particular TV show on the following evening -- a very inexpensive buy of some rerun program on an Independent station. The campaign would have placed the spot(s) to test in that program. The panel are not told the purpose of the survey but are asked 4 or 5 commercial product questions and 2-3 political questions as a "baseline."
Immediately after the program airs, the panel is recalled and asked a number of "cover" questions about the program and other ads which ran in the program and then a series of questions on the candidate's ads. This allows campaigns to choose between two different styles of ads, find duds, determine which of two symbols best trigger the message desired, etc. And it is done with voters who viewed the ad in a natural setting.
Future Trend: A number of polling firms, with Republican pollster Bill McInturff being the leader, have begun to test radio spots and/or the audio track from TV spots during an actual telephone interview. The interview introduces the poll, asks a few key "pre-ad" questions and then has a computer play the spot for the voter over the telephone. Follow- up questions are then asked in order to measure the voters' reaction to the spot and the potential impact on your candidate and the overall dynamic of the race. This also allows the viewer/listener to receive the stimuli in their natural setting. Being a by-product of a trend or track poll, it is not overly expensive.
Future Trend: Near the end of this midcampaign period many pollsters are recommending specialized polls which include very few questions and very large sample sizes. The purpose is largely to increase the campaign's ability to target more precisely. While early electoral/demographic targeting and benchmark polls have provided some guidance, using enhanced lists (or geo-demographics) late in this midcampaign period allows the pollsters to (a) bring current opinions and issues into the targeting formula and (b) be much more specific in terms of demographics, lifestyles and lifecycles.6
For example, it is important to know that a Republican candidate should target white men in his/her communication, but it is even more important to know whether these should be young men or older men -- well off, not so well off, or across the board in income. These smaller subgroups of white men respond to different messages, live in different parts of town, and watch different news programs and TV shows. Using this increased precision in targeting research greatly improves direct mail strategy, broadcast buys, cable buys and radio segmentation.
The increasingly lower cost of data collection and the shortness of the questionnaire make this modification of a trend poll very feasible for large campaigns with a heterogeneous electorate, and it will be used more and more due to the requests of the media consultants.
As voters begin to focus, campaigns begin to communicate -- or vice versa. With four to eight weeks to go, positioning and image development are usually over in most campaigns, and both party's candidates go into what media guru Bob Squier calls "rock and roll.7" This is the period when both candidates are firing their best shots at their soft voters and the pure persuadables -- not to mention firing at each other. These groups of voters obviously overlap for both candidates and that is where the communications collision exists.
Years ago, the soft vote for each candidate and the pure persuadables only accounted for 10-15 percent of the vote. With the exception of a few races (especially urban mayors' races), this is no longer the case. Now, between 45-50 percent report they made up their mind for whom to vote in October or within five weeks of election day. Most pre- election polls show undecided in the 5-20 percent range, but both candidates' soft vote, in most competitive races, usually runs the totally malleable electorate upward to 50 percent.8 One of the key functions of the polling firm during this period is to define the persuadables in order to pinpoint final targets for the campaign. The other key function is to determine which way the persuadables are moving -- and why?
The volatility at the end of a campaign is largely due to the lack of strength of party identification in the U.S. Party loyalty and strength of commitment to one's party has trended dramatically downward in the past 25 years, and now there is only minimal "structure" to most campaigns -- meaning there are plenty of voters for whom political party identification has little or no meaning. That is why money is spent on repetitive communication at the end of a campaign and why tracking polls have become one of the most important tools of modern day campaigns.
In the 1960s, when party ID meant more, there was less need for tracking, but the political leaders wouldn't have done it anyway. In the 1970s, the in-home personal interview was too expensive and surveys couldn't be turned around quickly enough to have an impact on last minute decision-making. However, improving technology has continually increased the speed of communication.
With the advent of central facility telephone interviewing, completing large samples in one, two, or three days became commonplace. In most contests in the 1970s tracking usually meant conducting a poll once a week for three or four weeks and reporting on each poll once a week. At the end of the decade, however, tracking came to mean rolling averages, continuous tracking which was more costly but provided a more stable trend line and was able to measure the effect of virtually every event or media mix during endgame.
Rolling average tracking means interviewing a set number of likely voters every night, cumulating the number of nights that would provide a full sample, and then tabulating the results. For the next track, the oldest data would be discarded and an equivalent amount of new data from the most recent night(s) would be added -- thus rolling the new data over the old data while keeping the middle data constant. And each wave of data was weighted by a model or demographic estimate of the electorate developed by the pollster in order to hold down variances caused by sampling error on critical demographic variables. This concept was introduced in Mississippi in 1978 (by Lance Tarrance for the Republicans) and Louisiana in 1979 (by Wiliam R. Hamilton for Democrats).
While there are numerous variations of the rolling average technique, major U.S. political campaigns were never the same again. In the early 1980s, campaigns with major budgets used this technique with its twice, thrice, occasionally daily reporting of poll results to help traffic specific television spots, change the candidate's schedule, help determine how to respond to an opponent's charge, determine where to spend extra television money, or measure the impact of a debate. Rather than have a number of specific polls that were three and four days apart and might miss measuring a specific event or media wave, rolling average tracking is available instantly -- it is like the Dow-Jones average of the campaign.
Future Trend: Rolling average tracking will only increase in use by political campaigns. As data collection costs drift downward it will become available to lower level campaigns, and major statewide campaigns will increase their sample sizes and receive more frequent updates.
The product itself is unlikely to change dramatically over the next decade; it will retain the two major determinants -- rolling in new data in place of old and weighting each wave of results by a predetermined model. What may change is that some of the Future Trends noted earlier, principally testing of the TV or radio spots, might be grafted onto the tracking system.
One caveat. While we live in an instant information world, it is difficult to think of the American electorate as so enthralled with politics that they are changing and responding to every little media modification a campaign or its opponent makes. It is possible that with daily tracking a pollster will get movement that is not real but simply a research artifact -- sample surveys of people's opinions can be only so accurate. If, however, due caution is taken by an experienced pollster and major attention is given to the trend line rather than daily changes, the polling profession in America has an immensely powerful tool.
Pollsters are at the heart of this information system -- they creates it, executes it, and develops the reporting/analytical format. This provides the pollster with the potential for enormous influence over both creative and spending decisions at the endgame of a campaign.
There is no doubt that public opinion research has become the central nervous system of the modern American political campaign. Increasing cynicism, alienation, and information overload have created a disconnect between voters and their government -- meaning voters must be polled, probed, and re-polled if we are to stay abreast of what voters are thinking, and how they are thinking about it.
The charges that polling stifles leadership, creates rather than measures opinion, or helps promote mostly negative campaigns, pale compared to the positive attribute of citizen feedback to public officials -- officials who would have very few other ways to obtain this fuel for our democracy.
In this article, I have personified the function of the pollster -- largely because that is the way it is. Almost all strategic research done in the U.S. is done by a small to medium size independent research firm headed by one or more established researcher-gurus. This function and the personification of it is not likely to go away now that it has found the center of the campaign.
|
4201 Connecticut
Avenue, NW, Suite 212 |