2. Social desirability
Attitudes, feelings and associations may be outside our conscious
awareness
Remedies for measuring sensitive attitudes:
Racial resentment scale to “smoke out” prejudice by providing cover
behind traditional values. Problems?
Racial stereotypes take the difference between stereotypes of Blacks
andWhites. Problems? (reactive: some people motivated by social
desirability bias))
Survey experiments to measure discrimination (random assignment
of respondents to different question wordings, asking about Black vs.
Latino vs.White or Male vs. Female candidates) to reduce consistency
bias.
Unobtrusive measures (List experiment) take care of social
desirability. Unobtrusive: respondent can’t tell the purpose of the
question.
Self-monitoring scale
Implicit attitudes
3. High SM are chronically concerned with the appropriateness of their interpersonal
behavior. They carefully regulate their self-presentation with regard to social norms
and contexts, and thus are highly responsive to social and interpersonal cues. L
Low SM scale are relatively less concerned with – and less capable of managing – how
well their behavior fits a situation. They are guided, instead, by their inner attitudes,
emotions, and dispositions; as a consequence, low (but not high) self-monitors manifest
consistency between their private attitudes and public actions across a range of social
domains (see Snyder 1987).
4. The association between gender stereotypes and
support for affirmative action is greater among Low
SMs than High SMs
Stereotypes of Latinos and support for the death
penalty is stronger among low SMs.
5. Some people may (consciously) edit their
responses to disguise their true attitudes.
Group biases may exist an implicit level (e.g.,
System 1 associative memory).
The grab-bag of remedies for measuring sensitive
attitudes are unlikely to work here.
6. Processes occurring outside of awareness and
control
Suggests limitations of conscious self-
reports, like focus groups and surveys
Could you report:
▪ whether a pro-life advertisement showing a partial birth
abortion influenced your feelings?
▪ whether a pro-gun control advertisement showing the
carnage atVirginiaTech influenced your attitudes?
7. Most easily studied in the lab with careful
controls and timed responses
Political science needs general knowledge
applied to general and diverse publics, not
just college sophomores
Solution: Internet-based survey methods--
the best of both worlds!
Control
Generalizability
9. In 1995, then-Republican House Majority
Leader Dick Armey (now a leader in the “tea
party” movement) famously referred to openly
homosexual Congressman Barney Frank as
"Barney Fag" in a press interview. Armey
apologized and said it was "a slip of the tongue".
Frank did not acceptArmey's explanation, saying "I
turned to my own expert, my mother, who reports
that in 59 years of marriage, no one ever introduced
her as Elsie Fag."
10. DELIBERATIVE
Deliberative processes are
cognitively effortful,
demanding of attention,
time consuming, and
presumed to be based on
an intentional memory
search for relevant facts
and considerations
Examples:
Answering a survey question.
AUTOMATIC
Involuntary, fast, immediate, top of
the head, and unlike conscious
processes can be activated even
when the individual’s conscious
attention is focused elsewhere.
Examples
the immediate activation of cognitive
associations (e.g., Bush is a Republican)
the spontaneous activation of feelings
(Republicans are evil; Democrats are dumb),
habitual actions that operate “mindlessly”
amount of time ABC anchorman Peter
Jennings smiled when reporting on Reagan
over Mondale in the 1984 presidential race
The racial cues in theWillie Horton ad
attacking Michael Dukakis in 1988.
“RATS” ad
12. Explicit attitudes measured with verbal self-
reports in a survey can be limited in two ways
Social desirability bias (editing & censoring “true”
attitudes)
People may not be aware of their implicit attitudes &
associations
13. EXPLICIT ATTITUDES
Measured directly through
verbal self-reports (e.g.,
racial resentment, racial
stereotypes)
Formation and change is
through a process that is:
Conscious
Deliberative to some degree
Controlled
Can be adjusted for new
information
IMPLICIT ATTITUDES
A mental association in
memory between an attitude
object and an evaluation that
occurs:
Outside of conscious awareness
or control
Automatically
Must be measured indirectly
(IAT, AMP)
Formed automatically
through repeated
associations between object
and evaluation or two objects
14. EXPLICIT
Properties:
People are fully aware that a self-
report of their attitude is being
requested.
Measures:
QuestionDeliberationExplicit Attitude
Examples of prejudice measures:
▪ Stereotypes, symbolic racism
Advantages
▪ Face validity
Disadvantages
▪ Social desirability bias
▪ Lack of awareness of “true” or implicit
attitudes
IMPLICIT
Properties:
automatic evaluation
outside our awareness
beyond our control
Measures:
StimulusAssociationsImplicit Attitude
Examples:
▪IAT, Payne’s AMP
Advantages
▪More unobtrusive (we can’t control our
responses to the measure)
Disadvantages
▪Can be hard to interpret (e.g., IAT)
15. Explicit attitudes of negative campaigns ads
find that people become more negative
toward the source but the target
Implicit attitudes become more negative toward
both the source and the target
16. How well does the IAT measuring implicit
attitudes toward Latino immigrants predict
immigration policy attitudes—net of
ideology, demographic factors,
authoritarianism and ethnocentrism?
17. Two blocks of sorting exercises, with 40 trials in each block
where people are sorting a stimulus item (e.g., Black,White).
Block 1: subjects quickly classify stimuli using a pair of combinations where" Black"
and "Good" go together, and "White" and "Bad" go together. Stimuli consist of facial
images of Blacks, facial images ofWhites, words with negative valence, and words
with positive valence.These stimuli appear randomly and individually on the center of
the computer screen, and as each stimulus appears, subjects use the 'E' or “I” key on a
computer keyboard to quickly sort the word into the appropriate pairing (e.g., Black +
Good IWhite + Bad).
Block 2: subjects perform a similar sorting task while using a slightly modified pair of
combinations, where "White" and "Good" go together, while "Black" and "Bad" go
together.The stimuli are the same as before, and again, subjects use the 'E' and “I”
keys to rapidly s ort these items into the appropriate categories as they randomly an d
individually appear o n the c enter o f the computer screen.
IAT = Block 1 – Block2: If subjects display f aster average sorting times for the first
block (i.e., Black + Good IWhite + Bad) than for the second (i.e.,White + Good I Black +
Bad), one infers that subjects possess a positive implicit attitude toward African
Americans (relative toWhite Americans).
20. Explicit attitudes toward Latino immigrants is measured with a feeling
thermometer rating of “Latino immigrants”
0 to 100
Very unfavorable Very favorable
21. 1. To what extent are voting decisions
influenced by explicit vs. implicit attitudes?
2. How much do voters know about their own
preferences?
3. Has racial prejudice faded?
4. Did it play a role in 2008 presidential
election?
22. If we ask people outright, hardly anyone
would say they would discriminate against a
qualified Black, Hispanic, Female job
candidate or presidential candidate (see
figures)
But we know, based on careful audit studies
that discrimination still occurs
Question: Perhaps group prejudice or
negative associations toward groups can’t be
captured fully with explicit measures?
25. ImplicitAssociationTest (IAT)
Anthony Greenwald
Online: 122 research reports of 184 independent
samples, 14,900 subjects
Implicit and explicit attitudes not highly
correlated & do not necessarily predict the same
types of behaviors
Affect Misattribution Procedure (AMP)
Keith Payne
26. Good reliability
Large effects
Difficult to
complete
Ambiguous
interpretation
•The most well-known implicit attitude measure
•The example here is an implicit attitude toward a candidate
27. Simple to complete
Simple to interpret
High reliability
Payne, Cheng, Govorun, & Stewart
(2005).
Payne’s measure: Large effects and easier to
interpret
29. “Your job in this task is to make simple judgments about the Chinese
symbols while avoiding distraction from the photos.”
“Pleasant photos can make you judge the symbols more positively than you
otherwise would. Likewise, unpleasant photos can make you judge the
symbols more negatively than you otherwise would… Please try your
absolute best not to be influenced by the photos…”
The Logic of the warning: If someone is explicitly trying to do one thing,
and systematically does another, you have good evidence that their
responses are automatic, not controlled.
Impact of the Warning: Telling people NOT to let the pictures influence
their judgments has no impact on their ratings of the Chinese symbols.
Conclude: this measure of implicit attitudes is not subject to awareness
or control.
30. Payne, Cheng, Govorun, & Stewart (2005). JPSP.
Large effect d = 2.4
Alpha = .81
Proportion pleasant responses
31. Payne, Cheng, Govorun, & Stewart (2005). JPSP.
Large effect d = 2.4
Alpha = .81
Proportion pleasant responses
32. Time Series (address based sampling)
Computer-assisted in-person interviews
Pre-election interview September – November 2008
Post interview November – December
N = 1,933 completing all measures
Panel Study (random digit dialing sampling)
Internet survey
Surveyed each month January – November 2008
Prejudice measured September – October
N = 1,056 completing all measures
37. Explicit prejudice in Survey 1
(the Panel Study)
Feelings toward Blacks
Sympathy for Blacks
Admiration for Blacks
Too much influence
Explicit prejudice in Survey 2
(theTime Series)
Feelings toward Blacks
Symbolic Racism
Stereotypes of IQ and work ethic
Controlled for:
Party,
Conservatism,
Race, Gender, Age,
Education, Income
38. Obama voter vote for McCain
Non-voter vote for McCain
Obama voter vote for neither
39. Directly, independent of explicit
Indirectly, serving as input for deliberate
reasoning (explicit attitudes)
Vote for
Obama
40. Direct and indirect influences of implicit prejudice on voting for McCain. The figure shows unstandardized regression coefficients with
all variables standardized on a scale from 0 to1. Coefficients in parentheses are values after controlling for explicit prejudice (p < .05,
one-tailed; p < .01; p < .001).
41.
42.
43. Both explicit and implicit attitudes had a significant
impact, though in different ways
Implicit attitudes have a direct impact on voting, especially for
Obama.
Implicit attitudes also have an indirect effect on voting for
Obama and McCain by shaping explicit attitudes
Conclude: Amazing that Implicit attitudes have any effect after
controlling for Explicit attitudes, which includes Symbolic
Racism as well as Racial stereotypes and thermometer ratings.
44. A new published article replicating the Payne et al
research using the AMP concludes that:
Implicit attitudes do not have a strong or consistent
direct impact on opinions toward or voting for Barack
Obama versus Democrats in general.
But, they do find that Implicit attitudes shape attitudes
toward Democrats, which may be due to the party’s
political support forAfrican Americans.
Explicit prejudice has a strong impact on voting forObama
in 2008.
Thus, implicit attitudes may shape political attitudes, but
they did not appear to influence voting for Barack Obama
in 2008 independently of explicit prejudice measures.