Polling and the public

Caucus and Primary season

Combined effect of polls and media coverage of polls is critical during these times for 2 reasons: 1. unable to cover all candidates equally, news media may give more attention to most serious and viable candidates, where viability is determined by which candidates have the most money and where they stand in the polls 2. caucuses and primaries are a sequence of elections in which news media coverage of the outcome in just one state can dramatically affect later polls and primaries -iowa and NH are critical bc winners in these states enjoy a sizable gain in support in the national polls because of the positive and extensive media coverage they receive ; coverage can move a candidate higher in the public opinions polls which in turn enhances candidates media coverage and credibility

scientific polls

use of probability sampling; objective (question wording;sampling)

pseudo polls

web polls, call in polls, push polls, write in polls; use of self selected samples-can be misleading and bias

Truthfulness of voters

-"bradley effect" - some white voters wouldn't reveal their opposition to a black candidate so the polls showed bradley was ahead until he randomly lost

Trial Heat surveys

-a question or series of questions within a survey -pair competing candidates and ask respondents who they'd vote for -"horse race" emphasis on media coverage -much can change between the time question is asked and actual voting on election day -results depend on question construction: party affiliation with less known candidates can make a huge difference

Uses of polls by candidates

-candiadates use positive poll results to generate campaign contributions or to deter contributions to their opponents -if a poll shows a candidate running poorly, the candidate may attack the credibility and relevance of the poll -testing themes in a poll and then driving campaigns off of that can be dangerous bc items in a survey cannot fully capture the actual political and media context of a real world campaign -candidates may manipulate aspects of the polling process to generate poll results that will advance them -candidates may schedule tv commercials in conjunction with party and media sponsored polls -candidates may leak polls (in house polls) -question wording and order to predispose respondents

Deliberative opinion poll

-combines elements of focus groups and standard public opinion poll -brings together a representative group of citizens, provides them with info and the opportunity for discussion on issues,and then polls them on those issues -Fishkin: arged public opinion surveys measure what the public thinks, BUT NOT what they'd think if they could met and become immersed in the issues through discussion -first major deliberative poll was in 1996 election -these polls are expensive and challenging (bringing together a representative group of ppl) -Some are weary that the conclusions from these polls cannot be generalizable to the broader population; Fishkin argues that since the sample was representative, similar attitude changes can be expected in the population at large -critics argue that much of what people experience and learn comes from ads, news coverage etc. & that most ppl don't take time to learn about issues or candidates in a such a thorough way -compared to traditional public opinion polls (which capture immediately whats on voters minds) , deliberative polls reveal what citizens think once they are informed about an issue

Push Polls

-gained way in 1996 elections -not surveys, unethical political telemarketing, phone calls disguised as research that aim to persuade voters and affect election outcome rather than measuring opinion -attack selected candidates, designed to deceive

When and Why election predictions are wrong

-in 1948, the bad call that dewey would beat truman, is attributed to the quota method of sampling in the polls and the fact that polling stopped to far in advance of the election and dint reflect movement of the democrats back to truman -in 1992 election criticisms focused on the tracking polls Gallup conducted at the end of campaign, which received lots of media attention. Generated concerns about the selection of respondents, method by which undecided voters were allocated to candidates, and a shift in poll analysis from registered voters to likely voters -real problem with election polls is not in their accuracy but in how theyre reported-many polls that dominate news coverage create horse race mentality. small changes in poll standing are given too much attention. and poll standing, rather than issues stances, often drives coverage of candidates

Focus groups

-in depth interview with small number of people (10-20) -selected to represent broad demographic groups -effectiveness illustrated in 1988 election when Bush team found through the group the issues it needed to undermine Dukakis campaign -focus group discussions offer insights into what factors motivate citizens, things that a public opinion survey wouldn't reveal -limitations: not mini public opinion surveys , external validity: the results of the group may not be generaziable to any broader population

Exit polls

-interviews with voters as they leave polling places -they ask voters whom they voted form collect info on the issue positions and demographic characteristics of respondents, -most prominent ones are conducted by major news organizations to predict and explain election outcomes -Advantages and uses: a. they are polls of actual voters-avoiding the problem that reelection surveys far of determining who will actually vote b. collected in many states, allowing state by state analysis of the election c. can be tabulated quickly, allowing instant predictions and descriptions of outcomes d. generate rich info that enables journalists and scientists to better understand the factors that shape voters choices -b/c these don't capture early voting poll votes, pollsters must conduct pre election polls in states where theres a lot of early voting to estimate how those voters behaved, then pollsters must combine those estimates with the exit polls to generate accurate predictions----makes it more challenging and raises possibility for errors -2002 election night: a computer programming problem with exit polls led abc, cbs,cnn, fox, nbc, and associated press to disband the Voter News Service (collected exit poll data) and create a new one called the National Election Pool

Estimating turnout in elections

-most difficult task is identifying which respondents will actually vote

Panel survey

-needed if the voltaluty of voters attitudes and preferences is the chief concern -the same individuals are interviewed two our more times; this is more costly and a difficult process b/c a. respondents must be located repeatedly b. respondents may not be willing to participate in multiple interviews c. the experience of being interviews the first time may affect the respondents answers at the next interview -DESPITE these difficulties, provides better info than cross sectional surveys about the dynamics of the campaign and voter decision making -reveall more volatility in voters preferences than cross sectional

Polls in the Presidential selection process

-polls affect campaign fundraising-bad poll results may lead to less money for their fundraising

Tracking Polls

-provide the most up to date info on which to base changes in campaign strategy and media advertising -expensive; rely on rolling samples -polls conducted on a daily basis (200 people interviewed each day leading up) -one danger: any days interview could be highly aberrant (abnormal), be careful not to overreact bc could be statistical blip -in the 2000 campaign, a poll conducted by gallup for cnn and usa today showed many different results in short period of time-due to the fact that when more democrats were surveyed, they voted algore. yet when republicans were, they voted bush -Abramowuitz argues for weighting the number of democrats and republicans in any sample based on a moving average of the proportion of each taken from surveys conducted over several weeks -----this lessens the changes observed from sample to sample

Treatment of undecided voters

-report their numbers but then assume they will split in the same way that the decided already have (reasonable rule when both candidates are equally well known) -if one candidate is known and the other isn't, it becomes harder -1989 contests undecided voters behavior accounted for the failure of reelection polls to mirror the outcomes

Changing political and economic climate

-surveys predict best when their is a normal voter turnout pattern

Timing of polls

-the closer to the election a reelection pol is conducted, the more accurate the results will be -polls would be more accurate if they did a better job of identifying likely voter sand monitoring trends in voters preferences in the end of campaigns -polls conducted over many dats are more accurate than overnight polls (lau) -tracking polls more accurate tan standard polls (lau) -pols interviews on weekdays and weekends are more accurate than weekday only polls (lau)

Cross sectional surveys

-used when the major polling organizations conduct multiple polls over time on an election contest -different samples of citizens are selected for each round of interviewing -problem: they only reveal the net change ; they cannot tell poll consumers about the gross change or about the pattern of the individual changes that produced the net result

Benchmark surveys

-usually begins when candidate decides to seek office -collects info about candidates public image and issues on positions -problem: timing; the earlier it is done, less likely respondents will know anything about the challenger, more likely that the situation will change as the election nears

Presidential debates and the general election

-winning a debate may be a function less of a candidates performance and mores how th media covered and interpreted that performance -one issue where news values and good polling methods clash is the medias appetite for instant polls, which provide an immediate reaction to dramatic events; but the problem is these instant polls have low response rates and have more bias -polls determine which candidates are allowed to participate in presidential candidates debates

3 important pieces of info gathered from benchmark surveys

1. candidates name recognition level 2. candidates electoral strength vs opponents 3. citizens assessments of an incumbent office holders performance

Guarding against push polls (how to spot them)

1. reputable polling provides the name of the sponsor of the research or the organization conducting the research-push polls don't provide that info 2. legit telephone surveys last 5 min or longer, push polls last less than 1 minute 3. thousands and thousands of calls made to citizens about a particular race is a sign of a push poll, bc genuine polls require fewer than a thousand respondents

Accuracy of prediction depends on 5 factors, what are they?

1. timing of preelection polls 2. treatment of undecided voters 3. estimation of voter turnout 4. truthfulness of respondents 5. changing political and economic climate