Measuring Biases, Stereotypes, and Heuristics
The Properties of Psychometrically Sound Measures
Psychometrics is a field in psychology associated with measuring psychological attributes. Psychometric research is concerned with two main tasks; constructing the instruments for measurement and developing the procedures for measurement.
One property of a psychometrically sound measure is reliability. A measure has to be able to measure a construct across time, situations, and a number of individuals in a consistent manner in order to be described as reliable. This means that there should be no significant change of results if the test is taken today or eight months from now. This is known as the test-retest reliability. Other types of reliability that psychometrically sound measures should demonstrate include internal consistency. The premise here is that all the items on the test should be measuring the same construct both in theory and in practice (Michell, 1999).
Another concept of psychometrically sound measures is validity. This means that the measure should be able to measure what is was created and intended to measure without fail. To ensure validity, reliability is necessary but not always sufficient in some instances. The types of validity in psychometric tests include face validity, faith validity, content validity, empirical validity and construct validity.
Construct validity is the degree to which the test measures an established trait or construct. Empirical validity, also known as predictive validity, refers to the relationship between the scores on the test and some criteria of performance obtained from the test. Content validity is also known as logical validity and it measures the degree to which a measure represents all the elements of the construct. Faith validity is the conviction that a selection test is valid while face validity refers to whether an instrument is measuring what it is supposed to be measuring.
Methods used to measure Biases, Stereotypes, and Heuristics
Social psychology is concerned with the study of attitudes, stereotypes, and biases. Many people think that bias is a prejudice, or negative bias. However, there are many instances where one has a positive bias towards a particular object or subject. Bias is a cognitive phenomenon that is usually measured by studying certain behaviors that result from an underlying cognitive bias.
The premise behind the heuristics and biases program is that judgment or decision-making carried out under uncertainty rests on simple heuristics rather than extensive complex algorithmic processing. The theory helped to explain human error without ever invoking the need to categorize it under motivated irrationality (Gilovich, Griffin, and Kahneman, 2002).
According to Gawronski and De Houwer (n.d.), self-reporting is the most representative research tool when trying to measure heuristics and biases in the field of social science and psychology. To measure people’s attitudes, biases, and other personality characteristics, it seems obvious that the best way is to simply ask them. Traditional self-reporting is regarded as an explicit measure. Explicit measures use self-reporting instruments to assess the attitudes that people say they have. These measures are deemed appropriate to measure explicit biases because people are usually aware that they hold these biases and are thus able to report them.
However, there times when individuals are unwilling, unable, or afraid to report accurate information of their own psychological attributes. Scenarios such as these occur when they are asked to self-report on socially sensitive matters such as abortion or ethnic prejudices. The responses on the self-report measures are usually altered by concerns regarding social desirability and self-presentation. With the ongoing clamor for political correctness, self-report measures are no longer the most accurate methods of assessing the biases that individuals have. This is because of the reactivity that may be invoked by the social desirability concerns.
To overcome this problem, psychologists developed an alternative measure described as implicit. Implicit measures reduce the respondent’s ability to control his own response to the assessment. The measures also do not require introspection for the assessor to be able to assess clearly the respondent’s psychological attributes. One important implicit measure is the Implicit Association Test.
The test measures how quickly people make associations between differing target concepts as well as when pleasant or unpleasant words are mentioned. The difference in speed in making these associations is what is used to deduce the measure of implicit attitudes towards target concepts. The underlying principle is that people will make a faster association/link with related concepts, but take a much longer time to make the association between unrelated concepts. For instance, individuals will quickly link flower terms with positive words and take a much longer time to associate weaponry terms with nice words.
The decreased reactive time for word pairing consistent with stereotypes occurs probably because the pairings occur or are prevalent in many of the stereotypes that abound in society. The more frequent occurrences of stereotype consistent pairings compared to the stereotype inconsistent pairings leads to the quicker processing of the stereotype consistent pairings leading to faster response times on these pairings.
Due to the differences between implicit and explicit biases, it is often assumed that the two have no relation. Explicit bias is largely considered conscious or aware bias while implicit bias is considered by many to be unconscious and unvolitional. On the surface, his or her distinction implies there is no relation. These assumptions support the belief that IAT along with other implicit measures can measure biases that operate independently from explicit biases.
However, there are experiments that have led psychometricists to believe that there could be some relation between implicit and explicit biases and therefore, their measures could be related. Studies conducted by Snowden (2005) on the predictors of criminal verdict tendency have shown that explicit and implicit measures can be used concurrently.
Cunningham, Preacher, and Banaji (2001) have shown that the IAT and a priming window method, both of which are implicit measures, have a significant relation to the Modern Racisms Scale, which is used to measure explicit bias against African Americans. Results from the studies show that explicit and implicit bias measures are usually related to one another. These relationships are not always between the bias measures or between the bias measures that are measuring the same construct.
Snowden’s studies reveal that bias measures such as MRS and IAT showed various degrees of positive relations while others such as the PJBQ-Race were not related to the IAT bias measure at all. In addition, the IAT bias measure was related to a number of other legal bias measures that did not directly measure racial bias. This implies that the IAT measure, an implicit measure, does not always measure the individual’s beliefs about racial bias but rather the degree to which he acknowledges, is aware of, and is influenced by the society’s attitudes towards precarious topics such as the society’s attitude towards ethnic minorities.
While the existence of a relation between the two types of bias measures, explicit and implicit, points to an overlap between implicit and explicit biases, the two categories are still studies and tested separately. The relations between implicit measures tend to be stronger than the relations between implicit and explicit measures meaning that both measures should be used to measure different constructs.
Gawronski, B., & De Houwer, J. (n.d.). Implicit measures in social and personality psychology.
Gilovich, T., Griffin, D., & Kahneman, D. (2002). Heuristics and biases: the psychology of intuitive judgment.
Michell, J. (1999). Measurement in Psychology. Cambridge: Cambridge University Press.
Snowden, J.L. (2005). Explicit and implicit bias measures: their relation and utility as predictors of criminal verdict tendency.