These notes were from a lecture given at
ATTITUDE
CHARACTERIZATION AND DEFINITION
What
is an attitude?
Gordon Allport: (1935) 'The concept of attitude is probably
the most
distinctive and indispensable concept in contemporary American social
psychology.'
Still arguably true, even now, on this side of the
it's a good intuitive starting point. Everybody sort of knows
what an
attitude is.
Three colloquial notions of an attitude:
1. Opinion, viewpoint, position, belief, reaction,
evaluation, judgment,
disposition, tendency, inclination
2. Striking a pose: an angle or
slant on things - partiality and bias
3. Taking a firm and forthrightly articulated position
Can't devise a definition
that captures all of it. It always leaves
something out.
However, in science, theoretical definitions are handy, so
you can agree
on what you are studying, although theoretical definitions
can undergo
revision in the light of research too.
Broad definition of attitude:
"A psychological tendency that is expressed by
evaluating a particular
entity with some degree of favor or disfavor." (Eagly & Chaiken, 1993).
=> Basically, an *evaluative* response: positive or
negative.
An attitudinal response can be cognitive (thinking),
affective (feeling), or
behavioral (stuff you do), or any combination thereof. Called the
tripartite
view, because it says attitudes have three components (doh!).
Attitudes are also said to vary in terms of their
*properties*. Osgood,
Tannenbaum, & Suci (1967) found that attitudes are made up of 3
basic
independent dimensions: positivity-negativity (by far the most
important)
, strength-weakness, and
activity-passivity.
Many important properties of attitudes are related to their
*strength*,
e.g.,
extremity, resistance to persuasion, persistence over time,
predictiveness of behavior.
Other examples of attitude properties include: certainty,
importance,
knowledge, accessibility, ambivalence, involvement. There are little
cottage industries of research revolving around each of them.
An alternative narrow definition of attitude, now commonly
accepted, is:
"A summary psychological
evaluation of an object". No more reference to
thinking, feeling, and behavior, just *evaluation*: the positivity-
negativity dimension is given pre-eminence.
The above suggests a *minimalist* interpretation of what
attitudes are:
"An attitude is an association between and object and
an evaluation".
The strength of
this association indicates attitude strength, also called
*attitude accessibility*. It is indexed by how quickly people make an
evaluation of an attitude object, e.g., the speed of their responses
classifying a named object as good or bad, or giving it a rating
(Fazio , 1986).
(So, attitudes = Object -> Evaluation associations,
e.g. Osama Bin Laden -> Bad.
By a similar logic, Beliefs = Object-Attribute
associations,
e.g., Osama Bin Laden -> psycho)
******************
ATTITUDE
MEASUREMENT
How do you measure an attitude? On possibility is to assess
open-ended
responses, like "What do you think about the war in
Afganistan?". You get
*unstructured protocols* from
this method.
Analysis of the content of these
unstructured protocols is then done by coders, who rate them along
emerging
dimensions of interest. This technique provides lots of information,
and
imposes few preconceptions, but such protocols are quirky, unique,
and
difficult to analyse. A more efficient way to measure attitudes is
to
put them on a *metric* or *scale*.
How do you put a psychological quantity, like attitudes, on
a metric
or scale?
No obvious physical referents, unlike measures of volume,
velocity,
or viscosity, which can be measured by instruments like a
ruler, a ruler + clock, drip-timing apparatus (!?). But what do you use for
attitudes?
********************
SELF-REPORT
MEASURES
Thurstone (1928) - Famous paper entitled: "Attitudes
can be measured".
Devised ways of capitalizing on the subjective judgments of
respondents
so as to draw reliable inferences about their attitudes and
express those
attitudes on a scale, with intervals corresponding to *equal
graduations* of attitudinal extremity.
The equality of graduations is important: imagine measuring
a distance with
a ruler consisting of unequal graduations!
That is, Thurstone devised sophisticated ways of mentally
*scaling* attitudes.
(Here is his method,
in case you're interested, but you will not be examined
on the details.
Thurstone's method of
'Equally Appearing Intervals': Large pool of statements, expressing both
positivity and negativity towards target
object. Group of judges then *classifies* each statement into 11
positive
and negative categories. [Or, make all possible pair-wise
comparisons between attitudinal statements made by a sample of respondents,
convert resulting proportions to z-scores, and sum these for each item derived.
Whew!]
Establish scale value
of item by taking mean or median of group ratings.
Select statements that
best reflect each of the 11 intervals. Then
administer to new individuals: mean or median value of the statements
with which they agree qualifies
Drawback: It's a pain
in the ass. Extensive preparation is necessary.
You need to use
separate samples for developing and testing the measure.
Three other methods, though not as rigorous, are a whole
lot easier:
(1) Likert (1932) scales, using the method of *summated
ratings*:
Again, get a large
pool of statements (half positive in character,
half negative in character, to discourage the acquiescence
("saying yes to
everything") bias.
Respondents indicate
the extent of their agreement or disagreement with
statements using a 5-point scale.
Sum responses to get total scores. Correlate total scores with item scores
Throw out the items
that don't correlate with the total (they don't give you any useful
information)
Advantages: you don't
have to pretest, if you get lucky
Disadvantages: the
scale intervals are not necessarily equal (e.g., a score
of 20 does not necessarily mean twice as extreme as a score
of 10)
(2) Osgood (1957): the semantic differential
Rate an object on bipolar scales anchored by pairs of
adjectives reflecting
an evaluative differential.
Advantages: No pretesting needed. The same anchors can be
used for a variety
of objects, permitting maximum flexibility. Also, scale
intervals tend to be
equal.
Disadvantages: Abstraction may limited
to behavioral predictiveness of attitudes
(3) Just use a single item!
e.g. What do you think of Tony Blair?
Advantage: Avoids all
scaling complications
Disadvantage: A single
item is less reliable (doesn't give the same response
over time), and is more prone to a variety of structural biases
(see below)
In
general, a psychologist should ensure that any self-report they use is
valid
(is measures what it is supposed to measure, and not something else)
and
reliable (it give the same score over time).
There are three types of biases that tend to creep into all
self-report
measures:
1. Structural:
depends on the format of, and items in, a questionnaire
2. Epistemological:
People don't know how they really feel.
3. Motivation:
People:
(a) don't want to admit their true
attitude to others (social desirability);
or
(b) don't want to admit it to
themselves (self-deception)
To get round these problems, *indirect
measures of attitude* may be used:
********************
INDIRECT
MEASURES OF ATTITUDE:
(A)
Honesty-promoting:
1.Flip-a-coin method:
If heads, say 'like' to an item, regardless of your
attitude.
If tails, then give your real attitude, 'like' or
'dislike'.
For any item, it cannot be known for sure if the response
is real or not
However, across many subjects, you can still estimate what
the group attitude
is.
2. Bogus Pipeline:
With appropriate stage-management, you can fool
participants into believing
that a machine can really tell their 'hidden' attitudes,
reducing the
perceived utility of telling lies in participants' eyes (Jones &
Sigall, 1971).
(B)
PROJECTIVE TESTS
1.Thematic
Apperception Test:
Participants tell open-ended stories told about a set of
pictures. Emerging
themes can reflect attitudes.
2.
Information Error Test:
Relies on bias. Attitudes, especially strong ones, produce a distortion
about
unknown estimated facts. So pro-lifer partisans may overestimate
number of
abortion complications, pro-choice partisans the number of women
who regret
having kept their babies.
(C)
Behavioral:
1. The "lost-letter" technique: (Milgram, Mann,
& Harter, 1965).
Postage-paid letters
clearly addressed to the (apparent) post-office boxes
of various politically contentious organizations left lying
around. The
post-office boxes are in fact rented by social
psychologists, who tally the
numbers of letters they receive, then infer social attitudes
towards the
organizations. A variant is the "lost
wallet" technique. The rate of return
of the wallet is assessed as a function of the race, gender,
etc. of the
person losing it, who leaves his or her address inside.
2. Various unobtrusive measures of actual behaviour from
which attitudes can
be plausibly inferred (Crosby, Bromley, & Saxe, 1980)
These include amount of eye contact, seating proximity,
warmth of handshake.
(D) Bodily: NOT NEEDED FOR
AS but interesting anyway!
1. Event-related
potential measures: (ERPs)
Some of our brainwaves (P300s) go 'blip' when unexpected
information is
encountered.
So if an item
towards which a person has a positive attitude is embedded
in a series of items towards which a person holds a negative
attitude
(or vice versa)
2. Micro-contractions in facial muscles:
Smiling activates those in the cheek, mainly *zygomatic*
muscles
Frowning activates those in the brow, mainly corrugator
muscles
Liking can be inferred from subtle twitches, invisible to
the naked eye.
(E)
Cognitive: NOT
NEEDED FOR AS
1. Attitude accessibility (Fazio, 1986):
The speed with which you classify an attitude object as good
or bad
indicates the strength of the object-evaluation association in
memory.
2. Implicit measures
(Greenwald & Banaji, 1986): (You don’t need to
worry about this!!)
One prime example --
Sequential Evaluative Priming:
If two stimuli, a
prime and target, are presented in quick succession, if
they match (are both positive or both negative), as opposed to
mismatch (one
is positive, the other negative), responses (where a
respondent has to classify the targets as being positive or negative) are speeded
up as opposed to slowed down. For example, if unknown X and Y primes, preceded
P(ositive) and N(egative) targets, and responses were faster for X-P and Y-N
pairings than for X-N and Y-P pairings, then a more positive attitude towards X
than Y could be inferred.
One problem with most
indirect measures of attitude is that they tend to be
more "noisy", that is more prone to random error,
than self-reported measures
of attitudes.
********************
FUNCTIONAL
MODEL OF ATTITUDES (IMPORTANT for AS)
Why are we interested in attitudes anyhow? Three reasons:
1: We want to *know* what's on people's minds per se
2: We want to be able to *predict* how they think and
behave
3: We want to *change* how they think and behave.
Consider voting: ask a representative sample in a poll how
they will vote.
1: It lets us know what people are thinking -- interesting
in itself.
2: It predicts how people think about related issues.
(e.g., if you like New Labour, you'll be biased in their
favour --
"spin-doctoring is harmless" -- and if you
dislike the Tories, you'll be
biased against them --
"baldy men make bad prime ministers")
3:
It predicts how people will actually vote.
(for either New Labour or the
Tories)
In the case of 2 and 3, attitudes are supposed to have
*causal power*. That
is, they dispose people to do or think, or not to do or
think, something.
But which comes first, attitude or behaviour? Consider the
analogy of a car:
is attitude the fuel and steering wheel, or just the compass
and speedometer?
Either way, attitudes predict behaviour (correlate with the
speed and
direction of the car) but only in the former case do they have
causal power.
This is a chicken-and-egg problem we explore in greater
detail in Lecture 3.
Let's suspend judgment for the moment however.
******************
Attitudes seem to
promote *biases* in attention, perception, and memory.
These are:
1. Selective
attention:
Our pre-existing
attitudes affect what information we choose to expose
ourselves to.
We pay attention to
supportive information, and turn attention away from
unsupportive information (a "congeniality
effect")
True?
Not always, but
overall, yes (Frey, 1986). Especially when:
(a) we
have previously freely committed ourselves to a opinion, or to a
course of action implying an opinion
(b) when
the information presented looks credible and plausible
However, people often
seek our unsupportive information when:
(a) They have *very
strong* pre-existing attitudes
-- "Go on, just try to prove me wrong!"
-- "bullish"
approach
(a) They have *very
weak* pre-existing attitudes
-- "Better watch
out, I might be wrong, so best to look at both sides"
--
"scaredy-cat" approach
So the congeniality
effect is strongest with moderately strong attitudes.
But let's not forget
that people are also interested in information that is
useful, which depends on the situation. Sometimes that is
information that
refutes our attitudes.
2. Selective
perception and judgement:
Our pre-existing
attitudes affect how we perceive and evaluate information.
We tend to perceive
and welcome supportive information, and not perceive or
decry unsupportive information (a "congeniality
effect")
True?
Yes.
People assimilate
(minimize the dissimilarity) of information that is
relatively in keeping with their own attitudes, whereas they contrast
(exaggerate
the dissimilarity) of information that is relatively at odds
with their own attitudes (Hovland & Sherif, 1952; Judd,
Kenny, & Krosnick,
1993).
Also, perceptions of
debates between politicians by rival supporters: each
thinks the other performed better, depending on the direction and
strength
of their pre-existing attitudes (Bothwell & Brigham,
1983;
Fazio & Williams, 1986).
Also, partisans on
both sides of the issue tend to see the media as hostile
to their point of view (e.g., both pro-war and anti-war
sympathizers think
that the media coverage does not give their side a fair
hearing)
(Vallone, Ross, & Lepper, 1985).
Not just passive
selective perception and judgment: participants may also
spend longer actively thinking about how to refute uncongenial
information
(Edwards & Smith, 1996).
3. Selective memory:
Our pre-existing
attitudes affect how we remember information. We tend to
remember supportive information, and forget unsupportive
information
(a
"congeniality effect").
True?
No. It very much
depends.
For one thing, people
may have adopted an active or a passive approach to
dealing with non-supportive information.
Adopting the passive
approach means taking attention away from, and not
processing, negative information, in an attempt to *avoid* it, which
tends
to make it less well remembered.
Adopting the active
approach means focusing on it, and processing it, in an
attempt to *refute* it, which tends to make it better remembered.
Also, memory may
depend on the nature and polarity (positive or negative) of
one's attitude-relevant knowledge (Pratkanis, 1989). A lot of
positive
knowledge will facilitate memory for positive information, a lot of
negative
knowledge will facilitate memory for negative information
("schema-consistent
recall") -- matching knowledge is better remembered because
it fits.
However, mismatching
knowledge may also be better recalled because it jars
with expectations, especially when people are interested in
getting at the
truth (Stangor & McMillan, 1992).
******************
ATTITUDE
STRUCTURE (IMPORTANT
FOR AS)
(1)
Tripartite or 3-component model:
Thinking,
feeling, doing
·
Cognitive (belief),
·
affective (moods,
feelings, emotions, physiological changes),
·
behavioral (behavior,
or intentions to behave)
Should
be some overlap, but not too much (just like different items on a
questionnaire).
The
Tripartite model is useful conceptually, because the categories make
intuitive
sense. But in practice the distinction may not be useful or feasible to make
Example:
Your lecturer.
Your think he's really smart, you have warm feelings when you
hear
him speak, and you really want to attend all his lectures. (Any
resemblance
to lecturers living or dead is purely accidental.)
Imagine
how the components stick together. No imagine how they could be at
odds
with one another (ambivalence). How do you resolve the matter?
Some testing
(Structural Equation Modelling, which allows you to test how
different correlational models suit a particular pattern of data)
shows that
3-factors suit the
data better than a single factor.
Breckler (1984). The 3 factor model worked with a live
snake, not a merely
imagined snake, in a key experiment. It might depend a lot of what
the
attitude object is.
One interesting
finding to emerge from this research is that past behavior
predicts future behavior independently of everything else. This
suggests
that conscious self-report leaves something out.
One compromise
solution: conclude that attitudes, as *summary evaluations*,
can be based on, or inferred from, all three sources of
information.
******************
(2) Expectancy-value
(EV) approach (Fishbein, 1967):
Attitudes are the
additive result of evaluative beliefs held about the
attitude object.
Beliefs are
object-value associations. Att = Sum(B(i) x E(i))
Sum of the expected
values:
sum of b(i) (= subjective beliefs about the likelihood that an
object has
attribute, i) weighted by e(i) (= subjective evaluation of the
mertis of
attribute, i)
---> equal final attitude.
Example:
Attitudes towards
I am 50% sure that
S.U. is good academically (8 out of 10 on a 10-point scale)
I am 66% sure that
S.U. is good socially (9 out of 10 on a 10-point scale)
I am 75% sure that
S.U. is bad food-wise (4 out of 10 on a 10-point scale)
Attitude: = (0.5 x 8)
+ (0.66 x 9) - (0.75 x 4) = (4 + 6 - 3) = + 7
And so on...
Here, feelings,
intentions, and behavior are the *result* of summed expectancy values about an
object, not a separate component.
Advantages of EV:
links beliefs to attitudes, suggests concrete ways you can
intervene to change attitude.
Disadvantages of EV:
Feelings left out a separate contributor to attitude.
Can't these serve as a
basis for attitude (a summary evaluation)?
You can derive more
general algebraic models for linking beliefs to attitudes
(I spare you the
details). These fall under the heading 'information
integration' models (
be replaced by other factors.
Averaging rules can be used in preference to
additive ones {which we used above), and indeed these have proven
more useful in social psychological research.
********************
FUNCTIONS
OF ATTITUDES: IMPORTANT
FOR AS
We don't just hold attitudes on empirical grounds. Sometime
they are useful
to hold. That is, they have a function for us. What could
those functions be?
Two obvious attitudes:
To hold socially approved attitudes: => social-adjustive
To hold valid attitudes : =>
reality orientation
Katz (1960)
List of four functions of attitudes:
Knowledge (object-appraisal):
It is useful to make the discrimination between good and
bad things. We also
summarize complex phenomena into simpler schemas: we mentally
structure our
social worlds so that they are manageable.
"I like you because it makes you easier to
understand"
Utilitarian (instrumental):
Attitudes serve self-interest -- they secure rewards and
avoid punishments
"I like my boss because he will help me get to the
top"
Value-Expressive:
Shaping one's own identity, involving the motivation to
clarify, define, and
express oneself. Core values are reflected.
"I hate non-organic food because I am committed to healthful
living"
Ego-defensive:
One holds an attitudes that
enhances or protects one's ego.
"
Synder and DeBono (1987):
For high self-monitors, attitudes serve a social-adjustive
function
For low self-monitors, attitudes serve a value-expressive
function
FUNCTIONS
OF ATTITUDES:
We don't just hold attitudes on empirical grounds. Sometime
they are useful
to hold. That is, they have a function for us. What could
those functions be?
Two obvious attitudes:
To hold socially approved attitudes: => social-adjustive
To hold valid attitudes : =>
reality orientation
Katz (1960)
List of four functions of attitudes:
Knowledge (object-appraisal):
It is useful to make the discrimination between good and
bad things. We also
summarize complex phenomena into simpler schemas: we mentally
structure our
social worlds so that they are manageable.
"I like you because it makes you easier to
understand"
Utilitarian (instrumental):
Attitudes serve self-interest -- they secure rewards and
avoid punishments
"I like my boss because he will help me get to the
top"
Value-Expressive:
Shaping one's own identity, involving the motivation to
clarify, define, and
express oneself. Core values are reflected.
"I hate non-organic food because I am committed to
healthful living"
Ego-defensive:
One holds an attitudes that
enhances or protects one's ego.
"
Synder and DeBono (1987):
For high self-monitors, attitudes serve a social-adjustive
function
For low self-monitors, attitudes serve a value-expressive function