Public Opinion Methodologies

1
Public Opinion Methodologies

• Main Options

2
4 principal methods

• survey research or polling
• focus groups
• experimental research
• analysis of mass media content

3
The Survey

• Key elements of the survey include
• Population of people that the survey is to
  provide data about (usually a constant)
• The kind of sample of individuals drawn from that
  population
• The method of data gathering used personal,
  phone or mail interviews
• The kind of questionnaire or interviewing
  instrument used
• The kinds of analyses done and inferences drawn.
  

4
Kinds of survey

• Polls are a special kind of simple, descriptive
  survey. More sophisticated surveys seek to
  explain why characteristics of a population are
  as they are or how they relate to others.
• Census surveys attempt to count everyone in the
  nation.
• Sample surveys take a sample group out of the
  total population.
• Probability sampling methods use statistical
  methods to choose who will be included in the
  sample.
• Non-probability surveys may be useful in helping
  to develop concepts and measures.
  Person-in-the-street or mall-intercept are
  examples. Surveys are typically cross-sectional.
  
• Panel surveys interview people at different
  times.
• Surveys may be a part of field experiments in
  which samples may be divided into experimental
  and other groups, but are they representative?

5
Problems with surveys

• If people dont want to be interviewed/surveyed.
• Telephone interviews are cost-effective, but
  limited to rather easy-to-read-and-comprehend
  question and response categories.
• In-person interviews (good rapport at great
  cost, therefore low numbers
• Mail-administered surveys (cheap but horrific
  response rates may be good for smaller,
  specialized populations that have an interest in
  the subject-matter) they allow more complex
  questions).
• Other fax and email, Internet (counting hits).

6
Problems with surveys questionnaire design

• Trade off between complexity of real-life and
  getting the general message across to broad
  population
• opinion items are often difficult to ask, and
  produce the least reliable responses (best to
  test how knowledgeable people are first, or give
  them some information first).
• Hypothetical questions produce poor answers.
• Questions can be leading (invite a given
  response) or double-barreled (cover more than
  one issue).
• Questions need to avoid respondents wanting to
  give socially-acceptable answers.
• Question order effects can also be a problem
  early questions can produce an attitude
  commitment that affects answers to later
  questions.
• Sometimes, multiple indicators are needed for
  complex, subtle or hidden attitudes.
• Interview training is critical.

7
Problems with polling surveys

• Pollsters can ask questions in biased or
  distorted ways.
• Push polls or telephone surveys on behalf of
  one candidate, in which views of opposing
  candidates are purposively misrepresented.
• Too much reliance on polls produces a leadership
  vacuum, reduces innovativeness, and unpopular
  actions
• Political polls may lead to candidates telling
  people what they want to hear.
• Polls focus too much attention on the aggregate,
  not enough on individual motivation and activity.

8
Focus Groups

• Carefully planned discussions designed to obtain
  perceptions on a defined area of interest in a
  permissive, non-threatening environment
• Are revealing of the role of social interaction
  in opinion formation and expression
• Allow for probing and depth
• Are not reliant on the dynamics of the
  respondent-interviewer relationship
• Are a check against the researchers bias,
  open-ended flexibility.
• Participants talk in their own language.
• They produce data and insight that would not be
  available without the group.
• They are good supplements to other methodologies
• They demonstrate opinion formation, a process of
  continuous construction through a myriad of
  complex schemata.
• Focus groups reveal the extreme contestability of
  many public issues, and they help reveal peoples
  dependence on the media, and their ambivalence
  about that dependence.
• Limitations settings of discussion are less
  natural than in participant observation results
  less easily analyzed than in survey research not
  quite like real life participants feel on the
  record.

9
Experimental Method

• Experimental method gives control. Experimenter
  randomly assign populations to experimental and
  control groups, subjects experimental group to
  the experimental stimulus, organizes pre-tests
  and post-tests, while holding all extraneous
  factors constant.
• Gets at variables that it would be very difficult
  to research in uncontrolled conditions. It is
  possible to test sophisticated hypotheses.
• Limitations difficult balance between internal
  validity and convincing simulations of the
  real-world they can usually only cope with a
  limited number of variables at a time.
• Subjects may second-guess the researcher.

10
Content analysis

• Mass media products must resonate with the public
  to do well, so they must be representative, in
  some ways, of public opinion.
• Is unobtrusive and yields rich data.
• Limitations coding schemes may be biased
  content analysis focuses on manifest rather than
  latent content. It doesnt deal well with visual
  content. Very labor intensive and costly.