Measuring Attitudes Day 2 & 5 01/08/2014
Day 2 January 8 h
Semantic Differential Technique (Osgood, Suci & Tennenbaum, 1957)
Asks individuals to rate something on a scale from strong to weak, good to bad, active to passive, fast to
Evaluates attitudes that people have on objects> view how people feel about them and change the product
E.g Pontiac brand> slow, old; 1960s they came out with a faster, sportier car
Empirically Determined Scales (Thurstone’s Equal Appearing Interval Scale, 1928)
Collect ~100 statements reflecting different opinions or evaluations of attitude object.
Many judges sort statements into piles, depending on degree of attitude.
Number on each statement depending on the degree of positivity or negativity.
Scale value of item= median of judges’ assignments.
Look for items that have a narrow spread, doesn’t change much> becomes relatively negative or positive.
Pick a few of items of that numerical judgment which are presented to an audience to evaluate them>
E.g “too much love will kill you” is rated at a 1 (negative) and “love is all that matters” is rated at an 8.5
Calculate the average of the statements presented to them that were rated.
Likert’s Summated Rating Scale, 1932
Collect statements reflecting different opinions or evaluations of attitude object.
To what extent do you agree or disagree with this statement?
Separate judges with 25% of the most negative and most positive separately> chose those statements that
best discriminate between the positive and negative evaluations.
These statements are then included in the final survey. Popular scale because it is useful and easy to use.
Reversed scored when the statements are negative (instead of 1 to 5, to 5 to 1).
Explicit Measures: Advantages & Disadvantages
Simply asking people their opinions
Quick and easy to administer
Quantitative results for statistical analysis
Scored by computer
Assumptions & Disadvantages:
Assume conscious access to attitudes> assumes that people can access these opinions on things.
Assume cognitive storage of attitudes> assume that someone has an attitude on something
Sometimes our attitudes are created on the fly
Sensitive to context effects> the ways the questions are worded, our mood, where you are, what you are
Prior experience (immediately)
Perceptions about expectations of researcher
Presence of other people
Cultural, social expectations about appropriate responses
Steps in Answering Context Effects
Decode and understand questions
Decide what the questions is asking (may be different depending on the person; “rate the effectiveness
of your instructor.”) Retrieve relevant information
Information that is retrieved will depend on the information retrieved from previous questions.
Life Satisfaction> Marital Satisfaction (correlation of about +.32)
Marital Satisfaction> Life Satisfaction (+.67)> because thinking first about satisfaction it their marital life
will be included in the life satisfaction section.
Respondents elaborate the first option ‘easier’ rather than harder.
On rating scale, choose first response in ‘latitude of acceptance’
E.g “Should divorce be easier to get, or harder to get?” OR “Should divorce be harder to get, or easier to
Acquiescence effects> tendency to say yes because people prefer to be agreeable.
Respondents tend to agree with any statement
One study found correlation of only .22 between mutually exclusive options.
Krosnick et al: Across 10 studies, 52% agree with the statement, but only 42% disagree with its
More common in lower IQ individuals, lower educated, more difficult questions, near end of long
questionnaires, in phone vs face to face interviews.
Fit response to alternatives provided
Our attitude may not be accurately reflected by the possible answers.
Scale items and numbers can bias responses> 010 vs 5 to +5> 5 is considered to some catastrophic
events happening and being very unsuccessful and will make there answer higher.
People will choose the categorical alternative closest to theirs, even if theirs is not present.
Schuman & Presser (1981):
62% “ to think for themselves” as most important skill for children to prepare for life when it is listed as an
5% list it when question is openended
People will choose an option even if they have no opinion.
Giljam & Granberg (1993): Nuclear Power Plant 15% choose “No opinion” when available
4% do not respond when “No Opinion” unavailable
Reliability is higher when all scale points are labeled rather than just the extremes.
Reliability decreases beyond 7 alternatives: hard to make such fine distinctions.
“Is my teacher an 8 or a 9?”
Day 5 January 15 – missed class
Structure and Function of Attitudes
Believed that we have stored representations of our attitudes
How are related attitudes stored?
Theory: we access information relative to an attitude object and we come up with an attitude on the spot.
Context effects may be relevant then because they predict the way that we determine our attitudes on the
If attitudes are stored, what part of them is stored and where are they stored and what elements are stored?
What is the relationship between implicit and explicit attitudes?
Each different model of attitudes suggests different storage and representation of attitudes> by testing how
attitudes are stored we may be able to agree about WHAT an attitude really is.
Can also tell us of how to change attitudes
Different storage of attitudes> tell us how these attitudes are formed> determine behaviour.
Some attitudes, based on how they are formed or their importance to us may be stored or
Structures of Attitudes Consistency
Consistency across> same attitude at one time compared to another, more often consistency
between the components of an attitude.
An attitude is influenced by emotions> feel certain ways towards attitude object(s)
Beliefs about things
Behaviours towards attitudes
Ambivalent attitudes> positive and negative feelings towards a specific attitude objects (cognitions).
Inconsistency within the cognitive domain
May also have ambivalent feelings about an attitude object> cognitive and emotional.
May have a positive cognitive (nice person), negative emotions (dislike) or vise versa.
Speed at which we can access attitudes> where the attitude is stored, importance, etc may effect our
ability to access these attitudes.
What determines accessibility?
Assume that strong attitudes are more readily accessible> really important attitudes, extreme or intense
How attitudes come to be strong attitudes? Are they stored differently which determines the strength of
Value of the attitude, to what degree does it shape one’s behaviour.
Models of Attitude Structure and Storage
Fishbein & Ageson (1970s): The Expectancy Valued Model
Attitudes are nothing but cognitions with varying degrees of strength.
A series of qualities, beliefs about certain attitude objects> quality is expressed as a probability of the
extent to which the individual possesses this quality. All qualities are not equally relevant or important to us
Attitude> collection of these qualities
Attitudes shape our emotions and behaviours but does not necessarily determine it.
Zanna & Rempel (1980s): One Factor Model weighted versions of all three factors (ABCs> Affect,
The attitude importance is determined by the extent to which it is able to help us reach our goals.
To what extent does this attitude represent positive or negative emotions, tendency to approach or avoid an
object> cognitions about an object> attitude.
(1980s): Two Factors
Factors: affect and cognition are involved in attitudes
Collection of cognitions and emotions and the relationship between them is the attitude.
No behavioural components> these two combined shapes behaviour.
Three Factor Model (1950s/60s):
Three parts to an attitude> cognitive (beliefs), affect (emotions) and behaviours (past behaviours towards
Correlations between all three> affect to behaviour, cognitive to behaviour.
Less correlated these factors are, the more ambivalent our attitudes are> less ambivalent= more
Combination of affect, accessible cognitions or beliefs and past experiences ARE the attitude> what we
store and report as attitudes.
How do we decide which model is correct?
Validity> the relationship between the model and reality
Convergent validity: measures of attitudes are correlated are similar to those of the model.
Do not get high correlations between the different components of attitudes (affect, behavioural, cognitive).
Measures within one component are correlated (e.g different measures of the cognitive component).
Discriminate validity: one particular model is more plausible than other models (statistically). One study showed that the Three Factors model was better than the One Factor, other study said that the
One Factor model was better than both the 2 and 3 Factor model.
Different modes of storage> different models may be better at explaining the storage of a particular attitude
better than another model. Measuring Attitudes Day 3 01/08/2014
Day 3 January 10 h
Edit or censor response
May not censor answers when any answer to these questions can be socially or culturally accepted.
More common when question is threatening or of a sensitive nature> racial,
ethnic, religious prejudice, political, sexual activities, etc.
Edit to make ourselves sound more desirable or socially acceptable.
More common in facetoface interviews
Reduced by assuring confidentiality