Huntingdon College | Political Science | Courses | What's New?

PSC 311: Voters, Parties & Elections
Traugott & Lavrakis, Voters' Guide to Election Polls, 2/e
Students' Outlines
Revised 19 Sep '06, compiled by Jeremy Lewis.

- A poll or a survey is a method of collection information from people by asking questions
-polls involve standardized questionnaires
-polls are a collection of information from a sample of people rather than the entire population
-polls provide candidates with information on what voters think and how they are inclined to vote
-media conduct polls to collect information for news stories and to form judgments on what kind news coverage to provide
-the end of the campaign, media use polls to project the winners of the race
-surveys is a data- collection technique that involves a questionnaire administrated to a group of individuals usually sampled from a larger population
-questionnaires are multiple items or questions
-surveys are conducted by academic researchers and government researchers
-polls come from the usage meaning ascertain preferences polling places is where people go to cast their votes on Election Day
-self administered questionnaires are polls that are conducted by telephones
-face-to-face interviews are the most expensive to conduct; they are limited to academic and government research project
-cross- sectional survey is a single sample of respondents are interviewed once and ask a set of questions
   Polls do have an effect on those who see or hear the results.  Some people change  their attitude and opinions based on what they think other people believe.

* Polls begin with the selection of a sample of people to be interviewed, who are called respondents. The
respondents are asked a series of questions in a standardized form, called a questionaire.

* Some election polls are conducted for candidates to help them develop strategy, organize their
campaign, and raise funds.

* Most election polls are conducted during the campaign period leading up to Election Day; they are
called pre-election polls. Some polls are conducted on Election Day with voters who are leaving the
places where they cast their ballots; these are called exit polls.

* Pre-election polls provide substantial content for pre-election news coverage. Most poll based
reporting focuses on who is ahead or behind. Exit polls are used to project the outcome of races on
election night.

* Modern election polls confront four critical problems:
  1. generating a good estimate of who is going to vote
  2. capturing the volatility that sometimes appears in the electorate
  3. allocating the preferences of those who say they are "undecided" when they are asked for whom they will vote
  4. increasing number of voters are castin their ballots the traditional Election Day, and their preferences have to be
     accounted for in making estimates of the outcome of the election

* Voting intention is a combination of candidate preference and likelihood of voting for that person.

* Candidate support is measured through a trial heat question. First, people are asked for whom they
are going to vote. Those who express an initial preference for one candidate or another are then asked
how strongly they feel about that choice.

* A strong commitment to a candidate is also a good indicator of intention to vote. So, the "strong
supporters" form the core or base electorate.

* Many polling organizations use an allocation method to assign the undecideds in different ways. One
way is to divide them equally between the two candidates. Another way is to allocate them according to
party identification.

* 2 different impacts of exposure to election projections
  1. The bandwagon effect suggests that people tend to support a candidate they believe is going to win.
  2. The underdog effect suggests that people tend to support the candidate they know is trailing.

* Election night projections of the the outcomes of races are based, in large part, on exit polls. Exit polls
are important because they provide explanation.
 

II. How Do Candidates Use Polls?
* They use polls to learn what the important issues are in the minds of voters.  These poll results provide a broad picture of the theme that the campaign will take.
* Polls can show the strengths and weaknesses of a candidate and their opponent.
* Polls can also be used to see how people view the position that a candidate has taken on certain issues.
* The candidate can use this information collected to influence the strategy of the campaign.

III. Other forms of Collecting Information During a Campaign?
* Analyzing historical voting patterns in precincts and cities,
* Reviewing successful advertising strategies from past campaigns
* Keeping track of news coverage of the campaign.
* Focus Groups – group of 8-10 people formed to discuss a particular topic

o These are unlike polls in that they focus on a broad range of discussion on limited set of topics whereas polls people are all asked the same question with limited responses.
o Campaigns use focus groups to find out what people are thinking about, but they are not uses to predict majority opinion
o They are many times used to evaluate the effectiveness of ads.

V. Do the results of the Campaign Polls Depend on How They are Conducted?
* Poll results can be manipulated by altering the data-collection methodology
* Wording of questions can make a huge difference in the level of support that a candidate receives.
* An Open End question like “Who would you like to see win the presidential election?” would serve a candidate with good name recognition.
* A question like” Out of these candidates (list) who would you like to see win the presidential election?”  Would serve a lesser known candidate better.

VI. Strategic Uses of Campaign Polls?
* Main purpose of campaign polls is to provide information on what the campaign should do or how effective its actions have been.
* Monica Lewinsky scandal – Clinton administration used poll results to determine that the public would accept Clinton’s adultery but not perjury.
* The release of poll results can be used to harm an opponent or make an opponent seem more favorable.

VII. Can Candidate Polls be Misleading?
* When campaigns release poll info especially in small pieces, the results can be misleading.
* Push Poll – a series of hypothetical question about their opponents are asked.  These questions could concern fact from their personal life or positions on certain issues.
* Candidates use this information to find out what they should disclose about the other candidate during the campaign in order to “push” support away from them.

VIII. How do Political Parties Use Polls?
* They use polls to check their standing with the electorate
* Use polls in the GOTV drives Get out the Vote Campaign
* Reform Party used mail survey in 1996 to elect their presidential nominee.

IX. Why Do Special Interest Groups use Polls?
* To maintain contact with their members and to find out where they stand on a particular issue
* To collect info from the public to see where it stands on a particular issue.
* To collect and distribute information to public policymakers.
* Problems can occur when theses polls are made public because they can be extremely biased.
* An unaccepted practice of pseudo polls is sometimes used to solicit new members or for fund raising. This is known as Soliciting Under the Guise and Fund Raising Under the Guise.
* Key to poll credibility is that there is a fair sample, full method is provided, and group conducting the poll is identified.

X. Do Elected Officials Pay Attention to Polls?
* They do pay attention to polls but are usually pretty good at determining whey polls should be considered and when they should be discounted.
* Research shows that most individuals will discount polls when they disagree with the views that person already holds.
* An example of this occurred when the Republican held congress wanted to impeach Clinton despite the fact that polls supported that the public did not support impeachment.
 
 

Candidates like to have as much information as possible at the most reasonable cost.

Most campaigns use polls before the going into an important stretch of the campaign trail. (Pre G.O.P convention poll, Post convention poll for an example)

Campaigns use polls to learn what the important issues are in the minds of the voters.

Also they are used to find out the strengths of the candidate and the opposition in terms of the positions they have taken on these issues.

Candidates use polls to evaluate the popularity of different or alternative positions they might take on an issue, or to evaluate the impact of different ways of discussing the issues.

The most obvious reason is to decide whether strategy should be changed, or should more effort be put into other areas.

Analyzing historical voting patterns in precincts and cities across the constituency, reviewing successful advertising strategies from past campaigns, and keeping track of the news coverage of campaigns, are ways in which campaigns rather info other than polls.

Another important figure in collecting info from places other than polls, are the use of “focus groups”

Eight to ten people are assembled for a group discussion organized around a particular topic; campaigns use the group to see how things are going in their particular races.

Early in the campaign, the best known candidates will generally have higher favorability rating, this rating usually tells who will do better in the trial heat questions that evaluate the candidates’ relative standing.

Poll results reflect a more reliable assessment of how the candidates are doing in their respective race.

The results of campaign polls, just like any other kind of poll, are highly sensitive, (most polls that are taken only show a small part of the registered voters) poll results can be manipulated by altering the data-collection methodology.

Campaign polls provide information on what the campaign should do or how effective its actions have been, so campaign polls have always been used strategically.

The release of poll results can be used strategically to foster a candidacy or to harm an opponent.

Campaign polls often evaluate strategic alternatives, the release of results, especially in small snippets of information, can be misleading, most campaign use “push polls”.

Push polls are polls in which a few hypothetical questions are asked.

Media results can hurt a candidate, when results can come out days before in the election; tendency shows that people are more likely to vote for the incumbent.

Political parties conduct their own polls to check their standing with the electorate and to augment the information available to their candidates.

Special –interest groups use polls to maintain contact with their members and to find out where they stand on a particular issue; or to collect and distribute information to public policymakers.

Interest groups also include pseudo polls with highly biased questions in their mass mailing campaign.

Interest groups try to project there ideas on people, so polls conducted by these groups could be tainted. It is useful to always find out who sponsored a poll and what questions were asked, who they were asked to.

 Pre-election Polls – questions about the issues
 “trial heat” focusing on candidate preference. Who is ahead & candidate standing.
 Election Day exit polls- helps predict outcome and analyzes why voters cast ballots the way they did

1. Like to have editorial control over the content and timing of surveys in order to use own judgment on news decisions and values.
2. To inform and structure their subsequent coverage
3. Enjoy professional prestige that comes from peers acknowledgement of the quality of polls. (Other news organizations citing another news organization’s poll.)

- News Organizations subscribed to syndicated services such as George Gallup or Louis Harris who usually used face-to-face polling
- Newspapers received 1 or 2 prepackaged news stories each week from syndicated services

- When cheap telephone service came about, news organizations started own polling
- All major news sources today conduct own polls. Of those who do not is because of financial reasons.
- T.V. stations attempt to have own polls but use inadequate methods that lead to inaccuracies

- Provides an independent perspective on citizen’s views about the candidates and issues involved in elections campaigns.
- Well conducted surveys provide another perspective on the dynamics of the campaign and popular assessments of the candidates.

- To share the costs.
- Partnerships began during the 1976 election when the New York Times and CBS worked together. Other partnerships include:

? ABC news and the Washington Post
? NBC and Wall Street Journal
? CNN with USA Today and Time

- T.V. presents results first but often a short simplistic explanation.
- Newspapers follow a few hours later but in greater detail.

- Yes, because the news is a “product” that has to be produced in a relatively short field period.
- One consequence is response rates are lower in short field time surveys as opposed to long field surveys.

- Telephone interviews
Samples of phone numbers are purchased
Computer Assisted Interviewing preferred

Closed-ended questions (most used)- respondents asked to choose from answers given
Open-ended questions (least used)- respondent answers the way he or she feels

Call-in – readers/viewers asked to call 800 or 900 number
Mail-in – readers asked to complete questionnaires inserted in magazine
or newspaper
Log-in – people go to a website
(CRAP) Computerized Response Audience Polls- data collected from
large number of computerized calls to samples that produce very low response rates

- A short field time means a higher probability of bias and inaccuracy
-  Surveys based on close-ended questions can also be biased especially when answers choices do not reflect public opinion.

- All polls analyzed with the use of statistical software.
- Media polls are analyzed by looking at the marginals or counts for each question in the survey and then by running cross tabulations by looking at the joint distribution of the responses of 2 questions. (example of cross tabulation: provides info as the percentage of men compared to women who approve of Bill Clinton.)

- Television- V.O. on file footage
- Newspaper- longer & more involved

- Reasons they can be different
? 2 polls use different samples or questions
? polls conducted at different times
? some polls use registered adult voters while other use anyone 18+
? order of questions (ex: asking about the economy before the presidential approval.)
? Why are there so many polls reported in the media?

Some of them purchase data from market research firms – others subscribe to a nationally  syndicated service such as Gallup or Harris polls.

New York Times and Washington Posts, have their own polling units.

Media Organizations do different polls at different stages of campaign:
1) Early in the campaign (pre-election)
2) “Trial-heat” (Question measuring candidate preference)
3) On Election Day (news organizations sponsor exit polls used to estimate the outcome of the race based on interviews with voters leaving their balloting places).

Election coverage is a “good story” for news organizations.

Elections have high impact on their audience members.

3 reasons why media conduct their own polls:

1) they like to have editorial control over the timing and content.
2) Use poll results to inform and structure their subsequent coverage.
3) Enjoy the professional prestige that comes from their peers.

All major newspapers use polls in reporting ….. those that don’t is because of the financial aspects of it.

With arrivial of low-cost telephone surveys, news organizations began to conduct their own polls at reasonable cost.

Main reason why media organizations collaborate on polls is to share the costs.

Pressure effect media polls, because the news is a “product” produced on a 24 hour cycle.  Therefore, in order for poll result to be “news,” they have to be fresh, meaning collected only in a couple of days of interviewing.

The short time results in lower response rates.  Less time to re-contact potential respondents who were not home, or refused to interview the first time.

Standard Data collection methods used in media polling:
- telephone interviews (samples of phone numbers are purchased)
- Computer assisted telephone interviewing (CATI)  interviewer reads questions from their monitor and records the answer directly into the database.  When interview completed , the data record for that respondents is immediately ready for analyses.
- Readers or viewers can call in to “800” numbers.
- Mail-in polls
- Log- in poll ( people can go to a website and register their views)

How are media polls analyzed and disseminated:
- with use of statistical software written for computers.
 Look at “ marginals”, percentages of men and women or percentage of approval and disapproval.

News organizations are “ data rich but analysis poor”  (because of time limitations)

One of the most important decisions that every pollster has to make is how to allocate the money available for a survey.

There are 2 critical components to a survey: the number of people to interview and the number of questions to ask.

Sampling is a technique for selecting a subset of units from a population in order to produce an estimate of some attribute or characteristic of the population at a reasonable cost. The practice of sampling actually involves 2 steps: 1) design of a scientific sample and 2) implementation of the sample in a valid and reliable way.

3 steps are involved in designing a sample: 1)define the target population, 2)select an appropriate sampling frame, and 3) select an appropriate probability method.

2 important principles are associated with the size of a sample: 1) the larger the sample size, the more precise the estimate and 2) the precision of a sample is typically not related to the size of the population….

Sample designs differ in 3 ways:
1) need to based on the laws of probability in order to take advantage of concepts like the margin of error,
2) vary in terms of the precision of the estimates they produce, and
3) vary in the ease with which they can be implemented.

2 conditions of a probability sample: 1) everyone in the sampling frame has a chance of being selected and 2) the pollster can determine the actual probability of each person being selected.

A nonprobability sample is when a sample of voters are selected without a clear specification of the target population and also when probabilities are unknown.

A random sample involves a sampling frame chosen at random.

A simple random sample uses a set of "random numbers" that are typically generated directly from a computer program….

A systematic sample is a special form of random sampling with a design that forces the random selection to take place over the entire sampling frame.

A stratified sample is another form of random sampling that strives to assure that the sample is also representative.

Mode of data collection
     1.   Face to face: the preferred interviewed method for many years.
      a.  Advantages
                         i.   Linked to the social realities of having the interviewers
       ii More complicated questioned formats can be used
       ii If a question requires to look up an answer in  their house
            iii.    While visiting a location, can unobtrusively and correctly code info about
          the person, house, and neighborhood.

      b.  Disadvantages
            i.      The lack of constant interview supervision
            ii.     Difficulty to find interviewer
            iii.    More respondents do not let strangers in their houses.

     2.   Telephone interviewing
               a.   Started late 1970s, and used almost every pre-election poll since 1980s
               b.   RDD (Random Dial Digit)
               c.   Advantages
                         i.   Speed in obtaining data
                                   (1)  CATI (Computer Assisted Telephone Interviewing):

Data can be recorded simultaneously in a computer file

     3.   Mail Services : Self-administered Questionnaires
               a.   Advantages
                         i.   More convenient to be able schedule the time
                         ii.  Relatively low cost
                         iii. Opportunity to retain the respondents anonymity
      b.  Disadvantage
            i.      Low response rate
            ii.     The timing necessary to conduct them: usually several weeks
            iii.    Produces incomplete or ambiguous answers
Unscientific mode of data collection
 People who conduct them have no way of knowing whether the resulting data are an
accurate reflection of the population they are supposed to represent

     4.   Interviewer
               a.   Demographics of skilled interviewers are generally unrelated to the data exception
          of asking substantive questions i.e. gender issues.
               b.   Need to be skilled, get well trained and well supervised
     5.   Interview
               a.   Introductory statement
                         i.   Helps to gain the cooperation of the person
                         ii.  Goal: to provide a minimum, yet adequate, amount of info about the study
               for the respondent to make a reasonably informed decision about whether
               or not to participate
               b.   Length of time
                         i.   Usually takes 10~20 mins, and between 50~100 questions
                         ii.  Contingency question sequences the questionnaire contains ( not everybody
               is applied) and how many of these sequences an individual respondent’s
               answers will invoke.
               c.   If initial contact was unsuccessful
                         i.   Try again different time of days, different days of week for respondents’
               convenient time.
                         ii.  Refusal conversation can yield somewhat 20~40 percent of success rate.
                         iii. Replacing or substitute another household or person can create serious
               problems of bias in the results

Questions-
There is no set amount of questions that have to be in a questionnaire.
These questions should be chosen and worded carefully b/c the questions are the
source of the data that pollsters later analyze.
These questions measure peoples:
    #1. Opinions
    #2. Knowledge of candidates and policies
    #3. Voting intentions
    #4. Whether or not they voted in last elections
    #5. Background Info (age, gender, education's, party identification)

There are two main types of questions:
    #1. Open-end: allow person to answer question in their own words for as long as
they like
        (i.e. What is the most important problem with the economy?)
        Problem with open-end question:
          A. Answers must be transformed into meaningful categories
    #2. Close-end: Gives the person choices for their answers
        (i.e. Do you strongly agree, agree, disagree, or strongly disagree that the U.S
should send troops to support NATO's mission in Kosovo?)
        Problem with close-end question:
          A. More than likely to end up with artificial and simplistic answers

Sometimes very in depth questionnaires contain double barreled questions.  These
questions often try and cover two very extensive topics. (i.e. How do you think the
President is handling foreign and domestic affairs?)

Double barreled questions are a good example of why pollsters must be careful with
the questions that are presented in their questionnaires.  Questions should be
worded so that every person, no matter what education level, can understand.

Purposes of questionnaires:
>#1. Describe- approval or disapproval of a certain topic
>#2. Predict- how a politician will handle certain situations
>#3. Explain- why something might happen or why it did happen

Questionnaires hold a valuable place in politics.  They can collect important data,
predict certain trends, and provide a general opinion of the voting public.

*Each question is converted into a variable or into variables that are then analyzed by statistical software.
*Software allows predictions about the poll(s) to be made at any time during the polling.

a.      Small sample sizes
b.      Presentation of biased data
c.      Statistical data that is unintelligible to the common citizen

a.      Who sponsored the poll

i.      One of the most important clues to know a poll’s accuracy
ii.     Academic institutions are the most trustworthy of sponsors

b.      Who conducted the poll

i.      Important to know in case you want to follow up on details of a poll

c.      How the poll was conducted

i.      Specific details about the questions asked – provides indication of level of the questions and allows evaluation of potential bias
ii.     Who the respondents were
iii.    How the respondents were sampled
iv.     Dates of when the poll was conducted

III.    Standards for reporting poll results

a.      National groups of professionals who have developed standards

i.      American Association for Public Opinion Research (AAPOR)

1.      members are individuals who produce and use survey data
2.      Code of Professional Ethics and Practices

ii.     National Council of Public Polls (NCPP)

1.      association of organizations that conduct polls
2.      Principles of Disclosure to guide the publication of poll results

b.      Agreed upon items that should be disclosed in reporting results

i.      Sponsorship
ii.     Who conducted
iii.    Field dates for interviewing – How current is the info?  Did any intervening events occur since or during the field dates?
iv.     Method of obtaining the interviews
-telephone interviews have higher response rates than mail surveys and are more timely
v.      Population sampled – probability design is most reliable
vi.     Size of sample – larger sample = smaller sampling error
vii.    Complete wording of questions asked
viii.   Percentages on which conclusions were based

-Polls can be inaccurate in that not every person who responds to poll questions is informed

-This can lead to measurement error, as respondents answer questions inaccurately because they do not want to appear uninformed

-The sponsor of each poll can word questions to find the information they want.  Poll questions can create “desired findings” which create a misleading result that enters the news stream

-Surveys are sometimes manipulated in the case of push polls, which manipulates the opinions of voter for strategic purposes

-Polls inserted into newspapers can be used to measure opinion’s reliably and accurately because of the use of better polling organizations, such as gallup

The following factors should be considered before participating in a survey:
-An explanation of who is conducting and sponsoring the study
An explanation of the purpose of the study
An assurance that your answers will be kept confidential
An explanation of who is being sampled and how the sample was chosen
A telephone number that you can call if you have any questions or want to check on the
 legitimacy of the study
A questionnaire appearing to have been crafted in an unbiased fashion with questions that are easy to understand and response choices that logically fit the questions being asked
The option to be called back at a convenient time
An assurance that cooperation is entirely voluntary and that response will remain totally
confidential.
 
 

Polls don’t always measure “Real” attitudes
Sometimes answers to poll questions do not reflect a considered opinion. Measurement error is produced by respondents who answer a question inaccurately because they don’t want to appear uninformed.

Polls can be designed to find whatever the sponsor wants
Poll questions can be written to lead to “desired” findings and the misleading result is then entered into the news stream.

A “survey” is not always a survey
Push polls are used for the purpose of manipulating the opinions of voters for strategic purposes. “SUGing” is used by telemarketers to solicit under the guise of polling.  “FRUGing” is the practice of fund raising under the guise of survey research.

Audience Call in polls are usually inaccurate
Call in polls might be considered legitimate vehicles for building audience interest in a radio or television program or among a newspaper’s readership, but they have no value for anyone who wants to sample voters’ opinions or preferences with any accuracy.

Editors and producers want to generate their own data at the lowest cost possible.
They only get what they pay for

It is the responsibility of journalists to serve as gatekeepers for data quality
This requires additional training in basic polling methods and the elements of data analysis for most of them.  This requires money.

Polls inserted into newspapers and magazines can not be used to measure people’s opinions and preferences reliably.

You should consider the following basic factors when deciding whether or not to participate in a survey:
Is there:
     An explanation of who is conducting and sponsoring the study
     An explanation of the purpose of the study
     An assurance that your answers will be kept confidential
     An explanation of who is being sampled and how the sample was chosen
     A telephone number that you can call if you have any questions or want to check on the
     legitimacy of the study
     A questionnaire appearing to have been crafted in an unbiased fashion with questions that
     are easy to understand and response choices that logically fit the questions being asked
     The option to be called back at a convenient time
     An assurance that cooperation is entirely voluntary and that response will remain totally
     confidential.