Post on 31-Mar-2015
Do Political Attitudes Exist?
Receive-Accept-Sample Model
By Zhou Yuyang and She Man-Hsuan
From public attitude to ‘simple theory of the Survey Response: Answering Questions versus Revealing Preferences’
Part 1
As we all know, opinion research need to get public attitude or
opinion from public though survey or poll.
Virtually all public opinion research proceeds on the assumption that
citizens possess reasonably well formed attitudes on major political
issues and that surveys are passive measures of these attitudes. The
standard view is that when survey respondents say they favor X they
are simply describing a preexisting state of feeling favorably toward
X.
Public Attitude
However, in the really world, when people are asked the same
question in a series of interviews, their attitude reports are highly
changeable. Many, as much evidence also shows, react strongly to
the context in which questions are asked, to the order in which
options are presented, and to wholly nonsubstantive changes in
question wording.
It shows on two aspects: response instability and response effects.
First, let’s see an instance.
Source: National Election Studies, 1980 Panel Survey
The data in Table 1, based on interviews of the same
persons six months apart, illustrate the problem. As can be
seen from the entries on the main diagonals, only 45% to
55% gave the same answer both times, even though about
30% could have done so by chance alone.
So, according to kind of phenomenon, in some thesis
( Converse ), they take any instability as evidence of a
"nonattitude," was an extreme claim intended to
characterize attitudes only on highly abstract issues.
Response instability
However, can we make sure that political attitude does not
exist? Or there are some other elements lead to the response
instability. Just like Zaller argue in the thesis ‘A Simple Theory
of the Survey Response: Answering Questions versus Revealing
Preferences’
Here, Zaller argue that the fluctuations that appear in people's
overt survey responses are attributed to "measurement error,"
where such error is said to stem from the inherent difficulty of
mapping one's attitudes onto the unavoidably vague language
of survey questions.
Further, not just random response variance (Response
instability ) exists, there also exists systematic variance
from artifactual “response effects”.
For example, in a split-half sample, 37% of respondents
were willing to allow communist reporters in the United
States. Yet when, in the other half-sample, respondents
were first asked whether U.S. reporters should be allowed
in Russia (which most favored), the percentage agreeing to
allow Russian reporters here doubled to 73%.
Response Effects
Thus, the literature on response effects makes it clear that survey
questions do not simply measure public opinion.
Public opinion researchers largely ignore both the longstanding
problem of massive over-time response instability and the newer
findings on questionnaire effect.
In the case of response effects, the patch-up consists of trying to
prevent the problem from becoming conspicuous; in the case of
response instability, the patch-up consists of statistical corrections for
measurement error.
To devise a theory that accommodates both response instability and
response effect. Zaller and Feldman in here give us a new alternative
model of the survey response.
Respondent is in ambivalence - that individuals
possess multiple and often conflicting opinions toward
important issue. They will give temporally unstable
responses in the course of a single conversation.
Current attitude models seem quite irrelevant to these
observation.
Zaller and Feldman persuaded that the basic point
about ambivalence represents an important insight.
An Alternative Model of the Survey Response
A deductive model based on three axioms:
The ambivalence axiom: "Most people possess opposing considerations
on most issues, that is, considerations that might lead them to decide the
issue either way."
The response axiom: "Individuals answer survey questions by averaging
across the considerations that happen to be salient at the moment of
response, where salience is determined by the accessibility axiom."
The accessibility axiom: "The accessibility of any given consideration
depends on a stochastic sampling process, where considerations that have
been recently thought about are somewhat more likely to be sampled."
Analysis bases on data from the 1987 Pilot Study of the National
Election Studies (NES).
The study was conducted in two waves a month apart; 457 persons
were interviewed in the May wave and 360 in the June wave. All had
previously participated in the 1986 National Election Study.
The basic method was to ask people a closed-ended policy item and
then to ask them to talk in their own words about the issues it raised.
The closed-ended items were telephone versions of the standard NES
items on job guarantees, aid to blacks, and government services and
spending.
Date of the Model
In form A, respondents were asked the open-ended probes immediately after
answering the given closed-ended policy item. ("retrospective" open-ended probe )
In form B, interviewers read the items in the usual way, but, without waiting for an
answer, they asked respondents to give their reactions to the principal idea elements
in the question.
("prospective" or "stop-and-think" probe)
The two types of probes are clearly not equivalent. The “retrospective“ probes, which
were posed after people had answered the question in the normal way, were designed
to find out what exactly was on people's minds at the moment of response. The
"prospective" or "stop-and-think" probes, on the other hand, were designed to induce
people to search their memories more carefully than they ordinarily would for
pertinent consideration.
Respondents were randomly assigned to question
form and answered the same type of the question
in each wave of the study. The three test items
and associated open-ended probes appeared near
the end of each wave of the survey. Interviewers
wrote down as faithfully as possible all responses
to the open-ended probes, including incidental
side comments (e.g., "This is a tough one").
Using these date, Zaller and Feldman deduce 17
hypotheses, of which they confirm 16 though their
process.
Instance: Deduction 1 - People who are, in general,
more politically aware have more considerations at the
top of their heads and available for use in answering
survey questions.
Response instability and response effects are also
explained by the model in the edduction.
Tests of the Model
The empirical phenomena for which Zaller and
Feldman’s model offers an explanation may be
grouped under three general headings, generally
explaining all these phenomena.
1. Dependence of attitude reports on probabilistic
memory search.
2. Effects of ideas recently made salient.
3. Effects of thought on attitude reports.
Three headings
1. Dependence of attitude reports on probabilistic
memory search.
Because attitude reports are based on memory searches that
are both probabilistic and incomplete, attitude reports tend to
be (1) unstable over time; (2) centered on the mean of the
underlying considerations; and (3) correlated with the
outcomes of memory searches (Deductions 3-5). This is also
why people who are more conflicted in their underlying
considerations are more unstable in their closed ended survey
responses (Deduction 8).
2. Effects of ideas recently made salient.
The notion that individuals' survey responses can
be deflected in the direction of ideas made
recently salient has been used to explain
question order effects, endorsement effects,
race-of-interviewer effects, reference group
effects, question framing effects, and TV news
priming effects (Deductions 9-16).
3. Effects of thought on attitude reports.
The notion that thinking about an issue, as gauged by general levels
of political awareness, enables people to recall a larger number of
considerations and hence to make more reliable responses has been
used to explain why more politically aware persons exhibit greater
response stability and why the public as a whole is more stable on
“doorstep” issues (Deductions 6, 7). It also explains why more
politically aware persons, and persons especially concerned about an
issue, are able to recall more thoughts relevant to it (Deductions 1, 2).
Finally, the notion that greater thought makes attitude reports more
reliable has been invoked, with only limited success, to explain the
effects of extra thought at the moment of responding to an issue
(Deduction 17).
1 People who are, in general, more politically aware have more considerations at the
top of their heads and available for use in answering survey questions.
2. People who have greater interest in an issue should have, all else equal, more
thoughts about that issue readily accessible in memory than other persons.
3. There should be strong correlations between the ideas at the top of people's minds as
they answer survey items and their decisions on the items themselves.
4. There should exist a fair amount of over-time instability in people's attitude reports.
5. Opinions that are subject to repeated measurement should have central tendencies
that are stable over time, but should fluctuate around these central tendencies.
6. The attitude reports of politically aware persons should exhibit greater over-time
stability than those of less-aware persons.
Seventeen Deductions
7. People should be more stable in their responses to closed-ended policy items
concerning doorstep issues-that is, issues so close to everyday concerns that most
people routinely pay some attention to them.
8. Greater ambivalence ought to be associated with higher levels of response
instability.
9. Raising new considerations in immediate proximity to a question should affect
the answers given by making different considerations salient.
10. People who are ambivalent on an issue should be most affected by
manipulations that raise new considerations in immediate proximity to a question
about the issue.
11. Inserting the name of a prominent politician or group into a question should
affect the public's responses to the question (the "endorsement effect").
12. The race of an interviewer should at least sometimes affect the responses to
questions which he or she asks.
13. Manipulations that raise the salience of a reference group can affect responses to
questions on which the reference group has a well-known position.
14. News reports can "prime" certain ideas, thereby making them more accessible for use
in formulating attitude statements on related subjects (the "priming effect").
15. Question order can "prime" certain ideas, thereby inducing correlations with proximate
related items.
16. Inducing individuals to think about their ideological orientation in close proximity to
questions having ideological content can "prime" ideology for use in answering those
questions.
17. Inducing people to think more carefully about an issue before stating an opinion
should enhance the reliability of the opinion report. (Not confirmed.)
Next Part
From ‘simple theory of the Survey Response: Answering Questions versus Revealing Preferences’ to Receive-Accept-Sample Model
Part 2
This is a model further developed from Zaller and Feldman
1992.
To explain how individuals respond to political information they
may encounter.
Zaller argues that elite-driven communications influence and
constrain public opinion.
That effect, however, is mediated by political awareness
(does the citizen perceive the elite communications?), which
determines the consistency and salience of mass opinion.
Zaller's RAS Model: Receive-Accept-Sample Model
This model is consists of four axioms, about how
individual respond to political information they may
encounter.
These four axioms work as a group.
This model is based on two phenomena: 1. how
citizens learn about matters that are for the most part
beyond their immediate experience; 2. how they
convert the information they acquire into opinions.
In here, Zaller explain some important conception about the model.
First, the consideration:
The public forms "considerations" in response to elite discourse
(political communications) in the mass media. Often, this discourse
consists of multiple, frequently conflicting streams of persuasive
messages. In general, the greater an individual's level of political
awareness, the more likely she is to receive these messages. Also, the
greater a person's level of awareness, the more likely she is to be
able, under certain circumstances, to resist (or accept) information
that is inconsistent with her basic values or partisanship. If
internalized, political considerations become reasons for taking one
side rather than the other on a political issue.
Second, two types of political messages:
Persuasive messages: arguments or images providing a
reason for taking a position or point of view, if accepted by an
individual, they become considerations;
Cueing messages: carried in elite discourse, consist of
“contextual information” about the ideological or partisan
implications of persuasive message, enable citizens to perceive
relationships between the persuasive message they receive and
their political predispositions.
Four axioms
Reception Axiom
The greater the person's level of cognitive engagement with an issue
the more likely he or she is to be exposed to and comprehend — in a
word, to receive — political messages concerning that issue.
Resistance Axiom
People tend to resist arguments that are inconsistent with their
political predispositions, but they do so only to the extent that they
possess the contextual information necessary to perceive a
relationship between the message and their predispositions.
Accessibility Axiom
The more recently a consideration has been called to
mind or thought about, the less time it takes to retrieve
that consideration or related considerations from
memory and bring them to the top of the head for use.
Response Axiom
Individuals answer survey questions by averaging across
the considerations that are immediately salient or
accessible to them.
The “RAS” model
◦ your stated opinions reflect considerations that you have received
(heard or read about)
◦ accepted (if they are consistent with prior beliefs)
◦ and sampled from (based on what's salient at the time)
The Bucket Analogy
◦ Considerations go into your head as if your head were a bucket.
◦ When you express an opinion, you reach into the bucket for a sample of
considerations; those near the top are more likely to be picked.
◦ You then take the average of these considerations, and that's your
opinion (at the moment).
Receive-Accept-Sample Model
The existence of widespread ambivalence, in
conjunction with the Response Axiom, provides a
ready explanation for it.
More aware persons will exhibit less chance
variability in their survey responses, according to
the A1, they are more likely to possess the cueing
messages necessary to respond to incoming
information in critical manner.
Further explain on the “response instability” in RAS Model
Zaller’s analysis of response instability has focused on effect of
awareness and issue concern on random response fluctuation.
However, if less aware persons exhibit greater chance
fluctuation, shouldn’t they also exhibit greater susceptibility to
enduring or systematic attitude change?
After investigation of attitude stability in the 1972-74-76 NES
panel, which attitude measurements were spaced at two-year
intervals.
The argument: Awareness had no effect on susceptibility to
systematic attitude change. (Zaller, 1986)
In RAS model, response variation is rooted in an important
substantive phenomenon, namely the common existence of
ambivalence in people’s reactions to issues.
This ambivalence has numerous implications for such
matters as the priming effect of the mass media, the effect
of survey question order, and attitude change.
The RAS model’s account of response instability is an
integral part of much more comprehensive way of thinking
about public opinion.
A term that refers to cases in which seemingly
irrelevant features of questionnaire design affect
the responses given.
Considering the example in the Part 1’s “response
effects”, it may be readily explained by axiom A3
in this model, which implies that the more recently
a consideration has been activated, the more
accessible it is for use in answering questions.
Further explain on the “ response effects” in RAS Model
Response effects were widely considered to be
“methodological artifacts” that indicated nothing of
theoretical significance about the nature of mass political
attitudes, but have always been given substantive
interpretations - three effects: 1)Race of interviewer,
2)Reference groups, 3) “priming effects” of television news.
These three effects may be counted as additional empirical
regularities for which the RAS model, in particular A3 and
A4, gives an explanation.
Also, RAS model is well suited to explaining priming effects.
It can be used in explaining the case of the Iran-Contra controversy on
ratings of Reagan’s job performance.
Prior to Iran-Contra, most media messages on Reagan focused on his social
welfare policy, which gave people a kind of considerations, as the
gatekeeper entering their heads.
When the Iran-Contra scandal happened, the former considerations was no
longer in people’s head, but the anticommunist or not became a
considerations about pro-Reagan or anti-Reagan.
Considerations that citizens use in evaluating presidential performance or
other types of attitude reports are significantly affected by the way the
topic has been framed in elite discourse.
Attitude change (understood as the a change in
people's long term response probabilities) results from a
change in the mix of ideas to which people are exposed.
Changes in the flow of political communication cause
attitude change not by producing a sudden conversion
experience but by producing gradual changes in the
balance of considerations that are present in people's
minds and available for answering survey questions.
Conclusions of the Zaller’s RAS Model
The psychology view of the survey response
Part 3
Make two main points:
1. Attitudes are preexisting evaluations of some target.
2. They are relatively stable
File drawer model (Wilson and Hodges 1992,p.38)
Three possible alternative sources for answers to attitude questions:
a. Impressions or stereotypes
When we don’t have a very clear sense about an issue, we may fall back on a general impression about the target or the category to which it belongs. Under these circumstances, our responses to an attitude question may reflect this overall impression. (Sanbonmatsu&Fazio,1990)
b. General attitudes or values The public’s the public’s reaction to press coverage of
political issue is determined by such predispositions; these involve basic political values, deeper underlying principles. (Zaller;1992)
Alternative Paths to an Answer
c. Specific beliefs or feeling about the target
Open-Ended MaterialWhen the order of the questions raises the salience of some particular domain or incidents, it tends to affect the rating of overall topic.
Priming StudiesFor example: (an initial item and a subsequent item)Abortion and women’s rightAbortion and U.S. policy toward Central American
Result: The closer the two questions, the more that retrieving considerations for the first question would reduce the time needed to retrieve considerations for the next.
They argue that responses to attitude questions
can be understood as the outcome of a question-
answering process in which people
(1) Comprehension
(2) Retrieval
(3) Judgment
(4) Response (Tourangeau et al. 2000).
The Belief-Sampling Model
Retrieval: Key assumption of the belief-sampling model is that
retrieval yields a haphazard assortment of beliefs, feelings, impressions, general values, and prior judgments about an issue; (cf. Zaller, 1992).
Judgment:
Under many circumstances, multiple considerations about
an issue will come to mind, and the respondent will have to
combine them to produce an overall judgment.
The output from the judgment component is a simple
average of the considerations that are the input to it.
In the equation :
J : the output of the judgement component.
S i : the scale value assigned to a consideration
retrieved from long-term memory.
n : the number of considerations the respondent
take into account.
/n
iJ s n
According to the belief-sampling model,
responses to attitude questions are inherently
unstable because they are based on a sample of
the relevant material, a sample that over
represents whatever considerations happen to be
accessible when the question is asked.
Second source of unreliability is in the values that
respondents assign to the considerations.
Implications for Response Stability
With a few additional assumptions, we can use
Judgment Equation to make quantitative
predictions about the correlation between
responses to the same question over time.
The key assumption is that the response actually
given to an attitude is a simpler linear
transformation of the Judgment described by
Equation.
According to the model, three key parameters
determine the level of the correlation between
responses to the same item on two occasions.
The first is the reliability of the scaling process,
measured by the correlation between the scale
values assigned to the same consideration on two
occasions.
The second parameter that affects the correlation
between answers overtime is the degree to which any
two considerations retrieved by the same respondent
are correlated.
The final parameter that affects the correlation
between responses over time is the degree of overlap
in the sets of considerations taken into account on
different occasions.
Reliability over two occasionsThis Equation shows the expected level of the correlation between
responses on two occasions as a function of these three parameters:
n1: Number of considerations sampled at time 1
n2: Number of considerations sampled at time 2
ρ1: Consistency in assigning scale values (scaling consistency)
ρ2: Homogeniety in pool of considerations(homogeniety)
q: Overlap between samples, expressed as proportion of
n2(overlap)
1 2 1 2 2 1 2
12 2 21 2 1 2 2 2 2 2
(1 )
(1 ) (1 )
n n n q
n n n n
Attitudes is a kind of memory structure that
contains existing evaluations, vague impressions,
general values, and relevant feelings and beliefs.
On any given occasion when we think about an
issue, some subset of these contents will come to
mind.
Conclusions
Thank You!