Measuring Implicit Attitudes

Dotsch and Wigboldus (2008)

Method: measured the effect of the IAT as well as measured skin conductance on a simulation task Result: found that participants primed with stereotypical primes of Morrocan names were more likely to have higher SCLS and maintain their distance from Moroccan avatars in a stimulated virtue bus stop compared to traditional Dutch avatars. Suggestion: They inferred that people held racist attitudes toward Moroccans without being aware of such attributes, especially since the correlation between explicit attitude measures was not significant.

Freise et al 2012

Recently, researchers argued that implicit attitudes would predict voting behavior particularly for undecided voters whereas explicit attitudes would predict voting behavior particularly for decided voters. We tested this assumption in two studies in two countries with distinct political systems in the context of real political elections. Results revealed that (a) explicit attitudes predicted voting behavior better than implicit attitudes for both decided and undecided voters, and (b) implicit attitudes predicted voting behavior better for decided than undecided voters. We propose that greater elaboration of attitudes produces stronger convergence between implicit and explicit attitudes resulting in better predictive validity of both,

Weinberger and Westen (2008): (Exp.2 only relevant):

Results: uncommitted voters were strongly and implicitly influenced by the prime. Limitation: it is difficult to evaluate the extent of these influences on pre-existing political attitudes as negative evaluations of politicians were easier to influence subliminally than were positive evaluations Implication: Therefore, prior to exploring the strength of the subliminal prime it is important to account for pre-existing political perspectives, and the strength of these perspectives.

Greenwald et al (1998)

developed the IAT, they applied it to a study on attitudes toward White and Black targets. Method: To find if participants held prejudiced attitudes toward Black people, they paired a White or Black person face with either positive or negative stimuli, on congruent (White with positive and Black with negative stimuli) or incongruent (vice versa) trials. They measured participants' reaction times of judging a trial as 'good' or 'bad' and computed a valence score in order to make inferences about participant speed of judgment. Greenwald et al (1998) concluded that white participants were slower to react on tasks that paired Black people primes with positive targets while Black participants were slower to react on tasks that paired white primes with positive targets.

Methodological problems with imp tests

-Though perhaps one should distinguish the measuring instrument from the outcome it produces (the "index")Ideally implicit tests should enable psychologists to access these true attitudes

Fiedler (2006) proposed 3 main areas of improvement:

1. pretest the specific stereotypical characteristics of say, Black people or Moroccan people, and apply them to the study. Explanation: This is to avoid the ambiguity of positive and negative target stimuli. It may also rule out alternative evidence that suggest people naturally react positive to in-group members and negatively to out-group members regardless of race (De Houwer, 2009). By using more specific stereotype targets, one could make stronger assumptions about attitudes. 2. additional psychometric measures should be used in tests to support implicit task conclusions. Although, such measures must continue to be researched and developed, they may improve the IATs reliability and validity of attitude constructs. 3. MOST Importantly - the IAT must avoid coupling incongruent and congruent valence scores, a they may reflect separate attitudes or separate degrees in which an attitude is held.

De Houwer et al. (2009), three criteria:

1.It should be clear which attributes causally produce the measurement outcome 2.It should be clear by which processes the attributes cause the measurement outcome 3.Implicit: The way in which the processes underlying a measure are automatic needs to be specified and demonstrated empirically

Reliability of IAT 2

2. it is not clear whether the IAT reflected negative attitudes or negative emotional states that a person holds E.G. a participant may see Moroccan people as friends or have positive attitudes toward them; but recently got into an argument with a Moroccan friend and therefore feels hesitant to be close to them.

Reliability of IAT 3

3. the difference score of valences is not assured to be reflecting the same attitude Explanation: In Greenwald et al (1998) , they assumed that a slow reaction on the paired Black - positive trial and a fast latency reaction or the Black- negative trial both represented a negative attitude toward Black people. But: Stephens and Buscher (2008) found a significant difference in reliability measures for the positive and negative valence tasks; therefore, such latency results should be measured and calculated separately.

Reliability of IAT 4

4. the IAT lacked multi-dimensional assessments to support the inferences made Explanation: - Greenwald et al (1998) used Black and White faces while Dotsch and Wigboldus used Moroccan and Dutch names. They included SCL data, but their results were only significant to a moderate degree.

Reliability of IAT 5

5. Fiedler et al (2006) proposed that IAT tests are quite obvious in what they are measuring which may lead to hypothesis - guessing and biased answers which inhibit the test's ability to measure unconscious implicit attitudes.

Affective Priming Task

Affective priming effect: Interaction between prime valence and target valence on reaction times (RTs) Affective priming effect found when target is evaluated, but also when a lexical decision (word vs. nonword) is required of the target Effects found with subliminal primes (e.g., Frings & Wentura (2003); Custers & Aarts (2007)

Dovidio 2002

As predicted, Whites' self-reported racial attitudes significantly predicted bias in their verbal behavior to Black relative to White confederates. Furthermore, these explicit attitudes predicted how much friendlier Whites felt that they behaved toward White than Black partners. In contrast, the response latency measure significantly predicted Whites' nonverbal friendliness and the extent to which the confederates and observers perceived bias in the participants' friendliness.

Reliability of IAT 1

Fiedler (2006): proposed that there were the following limitations. 1. it cannot necessarily be assumed that reaction times of paired primes and targets directly lead to causal inferences about attitudes. Explanation: Targets tend to be highly ambiguous such as 'flower' for positive stimuli or 'insect' for negative target stimuli (Greenwald et al, 1998) and thus, may not reflect the racial attitudes of someone.

Laurie et al 2011

In Study 1, men who automatically associated women more than men with primitive constructs (e.g., animals, instinct, nature) were more willing to rape and sexually harass women, and to report negative attitudes toward female rape victims. In Study 2, men who automatically associated women with animals (e.g., animals, paw, snout) more than with humans scored higher on a rape-behavioral analogue, as well as rape proclivity. Automatically objectifying women by associating them with objects, tools, and things was also positively correlated with men's rape proclivity. In concert, the research demonstrates that men who implicitly dehumanize women (as either animals or objects) are also likely to sexually victimize them.

Conclusion

The IATs development in 1998 began a revolution in the social psychological field, including opening up a world of psychological implicit traits. However, it does have certain methodological limitations that prevent bold causal inferences to be made (Fiedler, 2006). As the field of attitude research grows, so should the Implicit Attitude test, by improving validity and specificity of the primes and targets, by incorporating a multi-dimensional measurement approach, and by accurately calculating valence scores. Such improvements and more can make the IAT an even better measurement of people's beliefs, expectations and attitudes.

Stephens and Buschner (2008) conclusion

Thus, it was safe for them to conclude that attitudes people held against gay and for heterosexual targets was trans-situational and internal. They also illustrated that with improvements to the IAT, it can be used to measure true implicit attitudes.

Stephens and Buschner (2008) retests

To rule out the possibility that significant positive bias toward 'good' and heterosexual tasks, they did a retest at 1 week and 10 minutes. If the results for IAT held for both1 week and 10 minutes, they could conclude that external influences during the 1week did not change the strong IAT effects and that attitudes were not expressive of emotion or lab-produced beliefs. Stephens and Buschner found that the effects did not significantly differ from 10 minutes or 1 week post-test and results also did not correlate with explicit measures of attitude.

De Houwer et al. (2009):

implicit measure is "the outcome of a measurement procedure that is causally produced by psychological attributes in an automatic manner"

Stephens and Buschner (2008)

measured the valence reaction times to 'good' and 'bad' traits toward heterosexual and gay primes. (Prior to the test, they calculated the validity on each incongruent and congruent measures for both heterosexual and gay constructs and found high internal consistency on four separate measures. ) After participants completed the task, which they included both heterosexual and non-heterosexual participants, they found significant IAT effects.

dosgupta & riviera (2006):

peeops who endorsed traditional beliefs about sexality were friendlier verbally to homosexual confederates but showed more negative non-verbal behaviour