SP: Measuring Attitudes
Sample Exam Questions
1. When asked directly about their attitudes towards sensitive topics, people sometimes seem to lie. How can this problem be solved by using implicit or indirect measures, rather than by asking people? 2. Discuss measurement problems that researchers interested in attitudes have to deal with.
Explicit measurement problem: validity
A further measurement problem faced when using explicit structured measures is that the type of question asked can influence responses which lowers the validity of the measure. Responses to the same question can depend on whether the question is an 'open' question and therefore has no response alternatives or whether the question is a 'closed' question which requires respondents to choose their answer from several listed options.
Explicit measurement problem: social desirability bias
Actual attitudes or their socially desirable attitude: Explicit measures rely on respondents expressing their attitudes openly and honestly. This is hard to guarantee when using both structured and unstructured measures, especially when asking people about socially sensitive issues such as religion, politics, race or violence and personal issues such as smoking, drug use or cheating (Grimm, 2010). When asked about such issues, people often display the 'social desirability bias' whereby if they hold a socially undesirable attitude, they will try to hide or deny this and therefore will not respond with their true attitude (Campbell, 1963).
Implicit measurement problem: actual attitude
Another measurement problem for researchers interested in attitudes when using the IAT is that the automatic associations obtained do not necessarily reflect a person's actual attitude towards the target concepts. For example, even individuals who have no negative attitudes towards black people when assessed by priming measures and even black people themselves sometimes show strong associations between 'black' exemplars and negativity on the IAT.
Explicit measurement problem: scales
Another measurement problem researchers need to consider when measuring attitudes using closed questions in self-report measures is that responses may be influenced by the specific scales used. Again this makes it unclear whether people's true attitudes are being obtained. Uni-polar scales which typically range from 0 to 10 often elicit different responses than bi-polar scales which typically range from -5 to +5.
Measurement of attitudes
Attitudes can be measured explicitly or implicitly. Explicit measures, which use self-report techniques such as questionnaires, reveal attitudes that people are aware of and want to express overtly. Implicit measures elicit people's automatic evaluations of certain attitude objects that they are possibly unaware of and that may be different to the attitudes elicited by explicit measures. There are various implicit attitude measures, however, currently the most widely used is the Implicit Associations Test (IAT). Researchers rely on the responses of their participants to learn about their attitudes however there are several measurement problems of both explicit and implicit measures which reduce the validity of the responses.
Definition of attitude (Eagly and Chaiken, 1993)
Attitudes have been defined as: "a psychological tendency that is expressed by evaluating a particular entity with some degree of favour or disfavour" (Eagly and Chaiken, 1993). Attitudes are of interest in many fields such as social psychology, politics, health policy and marketing because they are one important factor that influences people's decisions and behaviour (Azjen and Fishbein).
Explicit measurement problem: validity Schuman and Presser (1981)
For example, Schuman and Presser (1981) found that when people were asked what they think "the most important thing for children to prepare them for life" is, when response alternatives were provided, 61.5% of respondents chose the response "to think for themselves". However, when this question was presented as an open question, only 4.6% of respondents gave this answer. Evidence such as this makes it unclear whether or not responses from closed questions do in fact reflect the respondent's true attitude which is problematic when drawing conclusions from research. It is possible that by providing participants with response alternatives, attitudes are placed in the minds of respondents and they respond with an answer that they may not have otherwise considered.
Explicit measurement problem: scales (Schwarz, Knäuper, Hippler, Noelle-Neumann and Clark, 1991)
Further research showed that when participants were asked about their attitudes towards their success in life on a rating scale ranging from "not at all successful" to "extremely successful", only 13% of a representative sample reported high success when the scale used ranged from 0-10 compared to 34% when the scale used ranged from -5 to +5. Thus the specific scales used in attitude measures can quite significantly change the attitudes people report.
Implicit Measures: no conscious access
Furthermore, implicit measures can elicit attitudes that people have no conscious access to and therefore are useful when researchers want to know people's preferences towards an attitude object. When the outcomes of implicit and explicit measures are compared, the attitude revealed towards the same attitude object can often be different (Fazio and Olson, 2003). This raises the question of whether one measure reveals a person's 'true' attitude or whether the attitudes revealed by explicit and implicit measures are qualitatively different.
Implicit measurement problem: awareness of target items
Furthermore, the implicitness of this measure is questionable as people are aware of the categories they are responding to. Interestingly, it has been shown that responses to the IAT can be faked because of this awareness. If people are told to appear neutral on an IAT they are able to slow down their responses on congruent trials and some evidence has shown they are actually able to speed up their responses on incongruent trials. Thus in some cases, it is possible that implicit measures do not reflect participants' true automatic responses.
Implicit measurement problem: one attitude object
However, one measurement problem that researchers interested in attitudes have to be aware of when using the IAT is that it does not allow the measurement of automatic attitudes towards just one attitude object. As it relies on the associations between two different categories, it may be problematic for researchers who purely want to measure attitudes to one attitude object.
Implicit Measures
Implicit measures which do not explicitly ask participants about their attitudes rather they use more subtle measurement techniques, are becoming more widely used, partly because they overcome problems of explicit measures.
Explicit measurement problem: scales (Schwarz, Grayson and Knauper, 1998)
Later experiments investigated reasons for this difference and found that the interpretation of the label "not at all successful" was different when it was presented as '0' compared to '-5'. When presented as '0' participants interpreted it to reflect a lack of outstanding achievements however when presented as '-5' participants interpreted it to reflect the experience of explicit failures in their life (Schwarz, Grayson and Knauper, 1998). Therefore the conclusions from attitudinal research using such measures may reflect respondents' interpretations of the scales used as opposed to their true attitudes.
implicit vs explicit attitudes
Researchers have to deal with the possibility that people's stated attitudes, especially when assessed by explicit measures, can be significantly influenced by both the desire to give socially desirable responses as well as by certain aspects of the measure itself, such as the types of questions asked and the rating scales used. Even though implicit measures such as the IAT overcome most of these problems, responses may again not accurately reflect a person's attitude towards the target concept. The measurement technique of the IAT elicits associations between categories which may reflect prior knowledge of participants as opposed to their true attitudes.
Explicit Measures
Researchers use explicit unstructured measures such as interviews as well as structured measures such as surveys, self-report questionnaires, attitudes scales and opinion polls with the hope of providing accurate and valid data about how strongly people feel and react towards various attitude objects (Dawes and Smith, 1985).
Explicit measurement problem: scales Schwarz (1999)
Schwarz (1999) has show that when rating a particular politician, many more people report a negative evaluation when using a uni-polar scale compared to a bi-polar scale indicating that people appear less inclined to use negative numbers.
Social desirability bias: suggestions
Several methods, however, have been suggested to address this problem. Some explicit measures can be administered without a researcher being present such as postal questionnaires which may help avoid the bias somewhat (Grimm, 2010). Additionally, measures that guarantee anonymity may help reduce the bias; participants may be more inclined to give honest answers if they know their responses will not be able to be traced back to them (Smith and Mackie, 2007). Another way to avoid the bias is to use the 'bogus pipeline' in which participants can be persuaded to tell the truth by convincing them that their 'real' attitudes are being measured physiologically even when that is not the case.
The IAT: Greenwald et al. (1998)
The IAT developed by Greenwald et al. (1998) is one of the most widely used implicit attitude measures and indirectly measures the strength of associations between two concepts e.g. race and positivity/negativity (Greenwald et al., 1998). The task requires participants to respond with specific response keys to four types of stimulus exemplars, such as common names of white and black people and the valence of different words (pleasant or unpleasant). There are several blocks to the task to learn to respond to the different stimuli with specific keys.
The IAT: Greenwald et al. (1998) II
The critical phases of the IAT involve simultaneous sorting of the stimulus items with two response options. The logic of the IAT is that this sorting task should be easier when the two concepts that share a response are strongly associated than when they are weakly associated. Greenwald et al. (1998) found that when 'black names' were paired with unpleasant, responses were faster than when black names were paired with pleasant. This indicates that participants' found it easier to associate the attribute unpleasant as opposed to pleasant with the with the target concept black. Such associations are inferred to reflect a person's automatic attitude towards the target concept. Again, this overcomes the problem of explicit measures in which people may be extremely reluctant to give a socially undesirable response when asked about their attitudes towards a certain racial groups.
Implicit Measures vs explicit measures
The literature suggests that explict and implicit measures are qualitatively different; explicit measures reveal explicit attitudes which are often a reflection of the evaluation people wish to endorse, implicit measures however reveal people's automatic attitudes which often reflect the positive or negative associations people have towards an attitude object. Interestingly, some research indicates that implicit measures are at times more useful in predicting behaviour than explicit measures however implicit measures have problems of their own.
Explicit measures: intend to interpret
The measurement problems of explicit measures explained above suggest that respondents use the features of the self-report measures to try to interpret the exact meaning of the questions in the measure. Reported attitudes may depend on how concerned an individual is with providing socially desirable responses. They may also depend on the specific rating scales and types of questions used among other factors, which is a potential problem for researchers as they may not be able to conclude that they have obtained participants' true attitudes towards the attitude object in question.
Implicit measurement problem: actual attitude explanation
This may be because of the knowledge of the historical negative portrayal of black people in American society which comes to mind easily when the 'black' exemplars and negative responses are congruent in the IAT thus facilitating their responses to such trials. Again this may be a possible reason why a higher percentage of a sample of white college students appear negative towards black people when assessed by the IAT compared to when priming measures are used. Thus the IAT may be problematic for researchers interested in attitudes if it does not actually measure actual attitudes towards the target concept. Findings such as these need to be taken into account when drawing conclusions from IAT research.