Romain Walker* and Lee Jussim
Department of Psychology,
Rutgers University, New Brunswick New Jersey 08901
*Rutgers Undergraduate Research Fellow and Henry Rutgers Scholar
This paper reports the results of two studies assessing the reliability and validity of a new scale, the Political Correctness Scale (the PC Scale), which is intended to assess the tendency to lie to appear unprejudiced. The scale was found to be reliable and valid. People with high PC scores lied more on a social desirability scale and reported lower prejudice on the Modern Racism scale. In addition, two experiments showed that people had higher PC scores when placed in a situation where there was pressure to hide prejudices and lower PC scores when there was situational pressure to tell the truth. A valid PC scale can be a useful measure for researchers studying the nature of prejudice.
Prejudice and discrimination are at the center of many modern social issues. Schools give tests and interview potential faculty to try to weed out those whose prejudices might be detrimental to students. Juries also are questioned about their beliefs and are instructed to reveal any prejudices that they have. These are definitely steps in the right direction. However, a new problem is then raised: do people lie to appear unprejudiced? If people are aware of the undesirable stigmas attached to being prejudiced, might they then hide their prejudices to get that teaching job or stay on the jury panel? To examine this question, we performed two experiments that tested whether people do lie to appear unprejudiced and if it is possible to measure that tendency.
A thought experiment
Consider a thought experiment: pick several people off a busy street and tell them you want to find out how prejudiced they are. Most would probably say that they consider themselves to be moral people and therefore do not see themselves as being prejudiced. Such responses, when taken at face value, would be very heartening. The world would indeed seem to be becoming a better place.
Then choose another set of people and ask them if it is sometimes necessary to lie to protect their interests. The response would most likely be an overwhelming "yes". But then compare the responses of the two sets of people. If they agree that people lie sometimes, how does one determine when they are lying? Surely, in today's society admitting to be prejudiced against minorities would be painting a very unflattering picture of oneself.
Declining prejudice in surveys
Despite the implications of the thought experiment one would be hard pressed to argue that the level of prejudice in the U.S. has remained constant over the past century. The country is undoubtedly more egalitarian at the beginning of the 21st century that it was at the beginning of the 20th century. Studies designed to measure whether people continue to hold negative beliefs and prejudices against minorities have shown that prejudice has indeed declined in the U.S (Karlins, Coffman, and Walters, 1969; Katz and Braly, 1933; Madon, et al, 2000).
Recently, however, researchers have begun to wonder if prejudice has declined as much as some surveys suggest (Devine, 1989; Plant and Devine; 1998; LePore and Brown 1997; Wittenbrink, Judd, and Park 1997). The most serious objection raised against the survey research questions the use of self-report scales to assess prejudice. The problem is that research has shown that people have a strong tendency to hide personal characteristics that might be socially undesirable (Jones and Sigall, 1971; Paulhus, 1991; Paulhus, in press).
Because prejudice is often socially stigmatized, people might also sometimes try to hide their prejudices. In the present study, such a pattern is called "politically correct responding" (PCR). PCR may create serious problems for members of stigmatized or marginalized groups. PCR also creates problems for researchers studying stereotypes and prejudice via questionnaires and other self-report methods. If people try to deny or hide their bigotry, studies may find less evidence of stereotypes and prejudice than really exists.
The present research aims to address this problem by attempting to develop a scale, named The Political Correctness Scale (PC Scale), for assessing the degree to which people lie on questionnaires in order to appear unprejudiced. Two experiments tested the validity of the PC scale by determining its relationship with prejudice and socially desirable responding, and by assessing whether PC scores could be influenced by social pressures to appear unprejudiced or to avoid being caught lying.
Implicit vs. explicit measurements
Researchers have long suspected that self-report ratings ("How much do you like Blacks?" "How assertive is Jane Smith?") have the potential to be polluted by all sorts of motivations to appear unbiased. As a result, researchers have developed an impressive array of measures designed to bypass participants' potentially misleading responses and get to the true nature of their attitudes and beliefs. Priming, lexical-decision tasks, verbal association tasks and observing behavior are examples of these methods (Devine 1989; Jones and Sigall, 1971; Lepore and Brown, 1997; Weitz, 1972; Winntenbrink, Judd and Park, 1997). The results of these studies consistently suggest that prejudice is, indeed, alive and well in modern culture.
An early study by Weitz (1972) compared attitudes held by White college students on the subject of acceptance of African-Americans versus their behavior toward African-Americans. To compare the attitudes to the behaviors participants were given a description of their partner (e.g. black lawyer, white gas attendant, white lawyer, black gas attendant) to read while they waited in a room. The descriptions were manipulated to create four experimental conditions; Black gas station attendant, Black law student, White gas station attendant and White law student. An attitude measurement scale, asking questions such as "How comfortable would you feel around this person?" was then given. The participants' decision on 5 behavioral measures based on the partner information given to them was compared with responses from the attitude measurement scale. Voicing was also measured while participants were asked to rehearse instructions that they would later give (which they did) to their partner.
Results showed that participants who claimed the most favorable attitudes toward blacks, also had the least friendly voice tone and behavior toward Blacks. This disparity suggests that they were overcompensating for their negative attitudes towards Blacks by rating them higher on explicit attitude scale.
Presumably, if people were completely forthcoming with their beliefs about all participants then there would be no disparity between an implicit measure of those beliefs and an explicit measure. However, as these studies show, there does exist a disparity. While these studies are not direct evidence that people lie to appear unprejudiced, they do show that at the very least self-report methods of assessing prejudice are not necessarily valid measurements of people's actual prejudices.
Socially desirable responding and the bogus pipeline
The tendency for people to give answers to make themselves look good is called socially desirable responding (SDR). SDR can manifest when measuring any psychological construct that relates to morality or competence. Researchers have been aware of this phenomenon for over 50 years and have developed various methods to attempt to correct, or at least detect this potential problem.
One method of controlling for SDR, a derivative of which we will be using in this study, is called the Bogus Pipeline (Jones and Sigall, 1971). Participants are hooked up to a machine that they have been led to believe is able to assess the veracity of their answers through involuntary physiological reactions. In experiments using this technique, people were more willing to express anti-social or immoral attitudes on topics such as sex and prejudice. Essentially, they opted to risk the possibility of being perceived as perverts and bigots rather than as liars, perverts and bigots.
In this study we used a variation of the Bogus Pipeline to achieve the same end. Participants were not hooked up to a pseudo lie detector. However, they were informed that the phenomenon of participants' lying on questionnaires, while somewhat irritating, is not at all new and therefore the experiment would employ procedures for detecting when participants lie to create a favorable impression of themselves. It was expected that this manipulation would replicate the results from Jones and Sigall (1971) and Sigall and Page (1971).
A need for the pc scale
The PC scale is designed to assess the degree to which people lie to appear unprejudiced. Such a scale is needed for several reasons. 1) The potential problem of participants lying on self-report questionnaires designed to assess prejudice has vexed researchers for some time; 2) a valid and reliable scale that assesses that tendency can be an invaluable tool to researchers using questionnaires to study prejudice; 3) it can also provide valuable insight into the nature of expressions of prejudice and stereotyping; and 4) no such scale exists.
Although no equivalent of the PC scale exists, several scales designed to assess motivation to appear unprejudiced do exist (Dunton and Fazio, 1997; Fazio, Jackson, Dunton and Williams, 1995; Plant and Devine, 1998.) The distinction between these scales lies in the fact that these other scales measure motivation to respond without prejudice whereas the PC scale measures the act of lying to appear unprejudiced.
Another advantage that the PC scale has over motivation to appear unprejudiced scales is that whereas the motivation scales focus entirely on prejudice toward African-Americans, the PC scale assesses prejudice toward a number of socially stigmatized groups. As prejudice toward African-Americans may not be equivalent to prejudice toward other groups, the PC scale may be useful to researchers studying prejudice against any of a variety of groups.
The overall aim of this research is to test the validity of the PC scale. We hypothesize that the more people are aware that their prejudices are being assessed the more they lie to attempt to cover them up. Presumably, if they are unaware that prejudice is being assessed they will not feel that their egalitarian image is being threatened and therefore will be more forthcoming with their beliefs and attitudes (i.e., they will express more prejudice). Also, we hypothesize that people will reveal more prejudice when they believe they might be caught lying. Therefore, PCR should be highest in the experimental conditions where prejudice is emphasized and lowest in the conditions where there is pressure to tell the truth.
In this attempt to assess the validity of the PC scale, two additional hypotheses are assessed. First, if PCR occurs when people want to avoid the stigma of prejudice, then perhaps PC scores will correlate with people's general tendency to lie to create a positive impression. In other words, higher PC scores should be associated with higher scores on a scale assessing impression management and socially desirable responding.
Second, if people who score more highly on the PC scale are more likely to lie to appear unprejudiced, then they should also be more likely to lie on scales assessing prejudice. Therefore, PC scores should negatively correlate with self-report measures of prejudice (higher PCR less self-reported prejudice)
The first study assessed the validity of the PC scale in two ways: 1) by assessing whether Politically Correct Responding increases with situational pressures to appear unprejudiced and decrease with situational pressures to tell the truth; and 2) by assessing the relationship of Politically Correct Responding with other scales measuring related constructs.
In this study we set up three different situations (conditions) under which different participants answered the same questions. One condition creates a situation in which the pressure to appear unprejudiced is high (Prejudice Obvious); another condition creates a situation in which there is pressure to tell the truth (Pseudo Bogus Pipeline); and a third control condition creates no pressure either to tell the truth or to appear unprejudiced (Camouflage).
Ninety-three (93) (39 males and 54 females) undergraduate psychology students were recruited from the Introductory Psychology pool. The ethnicities of the participants were: 38 White, 11 African-American/Black, 5 Latino and 29 Asian. Ten students classified themselves as "other". Participating in the study was one way to obtain course credit.
For all three conditions we used the same scales and the same filler questions, which were composed primarily of questions addressing political and economic issues. The measures for each scale are presented in Table 1A.
PC Scale. This 21-item scale is designed to catch people in the act of lying on anonymous questionnaires to appear unprejudiced. The questions on the scale do not directly ask participants if they lie to appear unprejudiced. Instead they asks questions such as "I have never noticed a person's race when I first met them" (Denial of awareness of obvious group differences) and "I am always friendly when I encounter a homeless person." (Exaggerated liking of a member of a stigmatized group). The scale uses a 1-7 range for answering. High PC scores should indicate more lying and low PC score should indicate less lying.
Modern Racism Scale. The MRS is a 6-item scale that assesses racist attitudes toward Blacks (McConahay, 1986). The scale uses a 1-5 range for answering. The scale is frequently used to assess racism and is a reliable measure (a = .79)
Impression Management Scale. The 20-item IM scale is a sub-scale of the Balance Inventory of Desirable Responding. It is designed to measure the degree to which people try to create an overly favorable impression of themselves on questionnaires. The most recent version of the IM scale uses a 1-5 responding scale and has been found to be a reliable measure of SDR (a= .70). Responses are transformed to "1" s "0"s and then summed. Only responses of 1 or 5 (reverse coded) are transformed into "1"s, all other scores are coded as "0" for analysis. This coding is done because only extreme scores on the IM scale constitute socially desirable responding.
This experiment uses a one-way between group design. There are three conditions: Prejudice Obvious, Camouflage, and Pseudo Bogus Pipeline.
Prejudice Obvious condition. Participants were given a questionnaire packet with an introduction similar to the verbal introduction given by the experimenter. This version of the questionnaire packet was divided into sections each with a brief introduction designed to make the participants aware that their prejudices were being measured.
To make it obvious that prejudice was being assessed, the first set of questions was entitled "The General Bigotry Scale" and was composed of the PC scale and the Modern Racism Scale. The introduction to this section emphasized the importance of erasing prejudice because of its harmful and unfair effects on minority group members, stating that
"Affirmative action programs, diversity programs and multiculturalism have all been instituted in order to help people overcome their biases and in order to reduce unfair discrimination against minority groups, women, gay men, lesbians, etc. This scale assesses certain prejudices that you may or may not hold."The "General Bigotry Scale" immediately followed these instructions. The purpose of creating this version of the questionnaire packet was to make the participant very aware that his/her prejudice was being assessed. For reasons stated earlier, when people are aware that their undesirable characteristics are being assessed, especially using a questionnaire, they may adjust their responses to appear unprejudiced. The purpose of this framing of the questionnaire was to evoke politically correct responses.
The next set of questions was entitled "Social Behavioral Assessment Scale" and was composed of the Impression Management scale and other social monitoring scales. These questions were used only as filler (to render prejudice assessment less obvious) in other versions of the questionnaire and therefore were also included in this version to equalize the length. The brief introduction for this section stated that the,
"This scale asks you some questions regarding both attitudes towards personal experiences and also toward a very broad range of social situations. It is designed to capture characteristics that are the most influential in decision making, in order to predict how people will react in various situations."The third and final set of questions was entitled "General Political Attitudes Scale" and was composed of filler questions, which were used in the other versions of the questionnaire, asking beliefs and opinions on various controversial political and economic issues. The instructions stated that
"This study assesses beliefs, opinions, and attitudes on political, social and economic issues of the day. The purpose of the scale is twofold. First, it assesses knowledge of political issues on the subject. Second, based on the responses, it predicts how strongly the average person would react to a sudden change in political structure."Camouflage Condition. In the Camouflage condition all the main scales (PC, MR, IM) used in the Prejudice Obvious version were interspersed with the filler questions. The introduction to the Camouflage condition stated that
Social psychological research often focuses on people's attitudes toward the social and political issues of the day. Attitudes are extremely important because they can be important influences on people's behavior. People often hold attitudes regarding a great many issues, although they may hold some attitudes more clearly or more strongly than others.
Pseudo Bogus Pipeline Condition. The Pseudo Bogus Pipeline condition uses ideas derived from the bogus pipeline paradigm created by Jones and Sigall (1971). The introduction attempted to convince the participants that their beliefs would be known through sophisticated methods developed by psychologists. Specifically, The introduction stated that
Social psychological research often focuses on people's attitudes toward the social and political issues of the day. Attitudes are extremely important because they can be important influences on people's behavior. People often hold attitudes regarding a great many issues, although they may hold some attitudes more clearly or more strongly than others.
"For example, consider the following question:
Participants were seated in a room in groups of 5-20. They were randomly assigned a questionnaire that placed them in one of the three experimental conditions; Prejudice Obvious, Camouflage or Pseudo Bogus Pipeline. The experimenter gave a brief rehearsed verbal introduction to the study, informing them that the study "aims to understand people's attitudes toward certain social situations." There was no mention of prejudice in the verbal introduction.
They then completed the 192-item questionnaire, which took about 40 minutes. After completing the questionnaire they completed a post-experimental questionnaire, which asked participants if they felt as if they were deceived and also what they thought the purpose of the experiment was. The most common complaint was repetitive questions (some questions from each scale are worded similarly to each other). However no one outside the Prejudice Obvious condition reported prejudice as being the focus of the study.
After completing both the questionnaire and post-experimental questionnaire, all participants were thanked, debriefed on the study, and given credit for participating.
Results and discussion
Concurrent validity of the pc scale
The concurrent validity of the PC scale was examined by assessing its relationship with socially desirable responding and racism. Impression Management Scale was used to measure SDR and the Modern Racism Scale was used to assess participants' racial prejudice.
Correlations were consistent with predictions. SDR scores positively correlated with PC scores (r(93)=.35, p<.01). People who scored higher on the PC scale were generally more likely to lie to create a positive impression.
Scores on the Modern Racism Scale correlated negatively with PC scores(r(93)= -.57 p<.01). This relationship was expected. Those who have greater tendency to lie to appear unprejudiced (high PC scores) also claim to hold less prejudice toward Blacks (low Modern Racism scores).
PC scores in the three conditions
ANOVA revealed that differences between the groups approached significance (F(2, 93) = 2.96, p < .057, Eta = .25). The means for PC scores differed in the 3 conditions according to the predicted outcomes (Prejudice Obvious = 90.7, Camouflage = 86.2, Pseudo Bogus Pipeline = 79.8). The mean in the Prejudice Obvious condition of 90.7 was significantly higher than the mean of 79.8 in the Pseudo Bogus pipeline condition (t(1,90) =2.41, p<.02) However, there was no significant difference between the Camouflage (86.2) and Pseudo Bogus Pipeline (79.8) conditions (t(1, 90) = 1.46, ns) or between Prejudice Obvious and Pseudo Bogus Pipeline (t(1, 90) = 1.01, ns). Test of Linearity showed that the difference between groups was almost entirely linear (F(1, 93) = 5.85, p<.02). There was no significant difference between means after accounting for linear relationship (F(1,93) <1, ns).
These results generally confirmed the hypothesis that people are likely to lie to appear unprejudiced when under situational pressure to do so. Analyses showed that the means between the groups differed suggesting lying on the questionnaire. This discrepancy gives evidence that the PC scale is capturing people's tendency to lie to appear unprejudiced. The relatively weak effect size (Eta =.25) and the limitations of this study that may have contributed to it are addressed in Study Two.
Study One provided evidence that people lie to appear unprejudiced. Furthermore, it provided evidence supporting the PC scale as a valid and reliable measure of lying to appear unprejudiced. Study Two further assesses the validity and reliability of the PC scale by addressing some of the limitations of Study One.
The first, and most obvious limitation of Study One was the relatively small effect size of .25. Although the pattern of means between the three conditions was consistent with predictions and the linear effect of the experimental manipulations was significant, the overall ANOVA was only marginally significant. Similarly, although the Prejudice Obvious mean differed significantly from the Camouflage mean, neither significantly differed from the mean of the control group.
There are two possible reasons for the relatively weak effect size of Study One. The first is the small N (93). Although the pattern in means was consistent with the hypothesis, the sample size may not have been large enough to render the effect significant.
The second possible contributing factor to the weak effect size may have been that the experimental manipulations were not strong enough in each condition. Study One used written introductions and relied on the participant to read and to respond based on the instructions. It is possible that some did not read the introductions carefully.
Study Two addresses both these possible limitations. The sample size was more than double that of Study One (from 93 to 194). Also, to maximize the strength of the experimental manipulations, introductions were given verbally by the experimenter along with a written version summarizing the verbal instructions. Finally, to strengthen the manipulation for the Prejudice Obvious condition (i.e., create more pressure to appear unprejudiced), participants were told that their responses on the prejudice questions would be discussed openly with those in their group. We hypothesized that people are aware of the negative view with which society views prejudice and discrimination and will not readily openly admit to it. Therefore, immediately after being reminded of these stigmas and then told that they would be discussed, they would be more likely to make a greater effort to avoid being perceived as prejudiced. This manipulation should result in higher PCR.
Study One also established a negative correlation between PCR and racism. This showed that people who have higher PC scores also report having less prejudice toward other groups. However, whether PC scores predict expressions of prejudice toward other groups remain unknown. Therefore, in addition to attitudes toward Blacks, attitudes toward two more stigmatized minorities, gays and lesbians, were examined in Study Two.
Also, there are different aspects to socially desirable responding that were not addressed in Study One. Lying to create a more favorable impression to others is only one kind of SDR. People may also try to deny to themselves that they possess unfavorable characteristics. A wider range of measures assessing SDR was therefore used in Study Two.
One hundred ninety-four (73 males and 121 females) Rutgers University introductory psychology students participated in this study as one way to obtain course credit. The ethnicities of the participants were: 99 White, 16 African-American/Black, 29 Latino, 39 Asian and 10 classified themselves as "other".
This study used the same scales as the previous study (PC, MR and IM) in addition to several new scales. The complete measures for all scales are listed in Table 1B. The new scales are described next.
Attitudes Toward Gays/Lesbians (ATG/L). The 6-item ATG/L is a self report scale measuring people's attitudes towards gays and lesbians (Herek and Capitanio, 1996).
Balanced Inventory of Desirable Responding (BIDR). The 40-item BIDR contains the Impression Management sub-scale as well as the Self-Deceptive Enhancement scale (SDE). The IM scale provides information on the tendency of some respondents to respond in such as way as to make themselves appear favorably to whomever interprets their results. The SDE scale provides information on the tendency of some respondents to provide agreeable self-profiles that are due to an overly confident, yet inaccurate, view of their abilities (Paulhus, 1991; Paulhus, in press). The scale has a high reliability (alpha = .83).
Self-Deceptive Denial Scale. The Self Deceptive-Denial scale also measures socially desirable responding. However, whereas the SDE captures people's exaggerated belief in their abilities, the SDD measures their tendency to describe themselves as overly moral and truthful.
The experimental design for Study Two is nearly identical to that of Study One. Study Two also uses a one-way between group design and also uses the same 3 conditions as Study One (Prejudice Obvious, Camouflage, and Pseudo Bogus Pipeline) with modifications in the experimental manipulations.
Manipulations. The structure of the questionnaires was identical to those of Study One. Each questionnaire contained a written introduction similar to a verbal intro (described next) given by the experimenter before the questionnaire was handed out. In addition, participants were given written introductions that summarized the points made in the verbal introductions. The written instructions were the same as those used in Study One.
The experimenters were trained to give a different verbal introduction to participants in each of the three conditions. In addition to the introductions for the three conditions, a general introduction was given to all participants.
The general introduction introduced the study as the "personal attitudes" study. It gave instructions on completing consent forms and told them that the study would take approximately 45 minutes.
The verbal introduction for the Prejudice Obvious condition highlighted the stigmas attached to prejudice and the efforts to remove it from society by stating that
Prejudice, discrimination, and racism are important social problems. Diversity programs, multiculturalism and affirmative actions programs have all been instituted in order to overcome social problems such as racial profiling by the police, sexism and homophobia.The experimenter also told them that they were about to "complete a questionnaire assessing your prejudice against various groups such as gays, blacks, lesbians, and other minority groups" to make it very clear that their personal prejudices were being measured. Also, to justify the presence of the filler questions, participants were told that they were also going to answer "questions concerning issues other than prejudice such as political issues, your views on international economics, and some of your own day-to-day experiences. The reason for this is to get an idea of the kind of person you are."
Participants in this condition were also told, untruthfully, that there would be an "open discussion of your answers on the prejudice questions" after everyone was finished with the questionnaire. We did so to enhance the situational pressure to appear unprejudiced.
In the Camouflage condition, participants were told simply that the study measured "people's attitudes in various social situations" in an attempt to "better understand people's behaviors." They were told that they would complete a questionnaire that asks their "opinions and how strongly you hold those opinions on many social issues, including international and national politics, economics, personal social interactions, and various other controversial issues." The condition was designed to appear that the study was measuring general attitudes rather than any specific construct.
The Pseudo Bogus Pipeline introduction did three things. First, it told participants that we were aware that people have lied on questionnaires and "psychologists have known for over 50 years that people sometimes lie to portray themselves more favorably than they deserve" (we know you're going to try to lie). They were then told, "In that time, some very sophisticated techniques have been developed to catch people in the act of lying" (we've figured out how to catch you lying). Finally, they were told that because the questionnaire is "peppered" with such questions, some obvious and some not, the experimenters would be able to "statistically adjust your answers so that they more accurately reflect the real, and considerably less favorable, you" (implying that since we know you're going to try to lie and we can spot a lie, you might as well tell the truth). The point of this manipulation was not only to convince them that their beliefs will be known, but also to convince them that it would be in their best interests to answer honestly.
Fifteen (15) to twenty-five (25) participants per session were seated in the main room. Once all participants were present the general introduction was given and consent forms were completed. All participants were then randomly assigned to one of the three conditions. Each of the three groups was then instructed to follow the experimenter assigned to their group to a separate room. Those assigned to the Prejudice Obvious group remained in the original room.
Once all participants were seated in their respective rooms, the experimenter of each group delivered the verbal introduction. All groups were given a similar set of basic instructions for completing the questionnaire (e.g. "Please read the instructions carefully", "If you have a questions, raise your hand instead of asking out loud", "Do not talk until everyone has finished and the questionnaire has been collected"). They were then given the questionnaire to complete.
As in Study One, after completing the questionnaire all participants completed a post-experimental questionnaire asking if they felt deceived and what they thought the purpose of the study was. Also as in Study One, no one outside the Prejudice Obvious group responded that the study was about prejudice. Again, the most common complaint was of repetitive questions. They were then thoroughly debriefed by their experimenter, thanked, and given credit for their participation.
Concurrent validity of the PC scale
The concurrent validity of the PC scale was established by examining its relationship with self-report measures of prejudice and socially desirable responding. The general findings were that people who score high on the PC scale are also more likely to engage in socially desirable responding and less likely to rate themselves high in prejudice on self-report scales.
Correlation Between PC and Prejudice Scores. Consistent with hypotheses, PC scores correlated negatively with scores on the Modern Racism scale (r(194)= -.46. p<.01). However, there was no significant correlation between PC scores and scores on the ATGL scale, possibly because open dislike toward Gays and Lesbians is not as stigmatized as open dislike toward Blacks. If this is the case, then people do not feel pressured to hide their prejudices toward Gays and Lesbians. However, since there is a well-known stigma attached to being prejudiced toward Blacks, people are more likely to lie to hide their prejudices accounting for the negative correlation.
Correlation between PC and SDR Scores. PC scores correlated positively with all measures of socially desirable responding. The Self-Deceptive Enhancement scale had the lowest correlation with PC scores (r(194)= -.31. p<.01). The Impression Management scale correlated highly with PC scores (r(194)= -.64. p<.01), as did the Self-Deceptive Denial scale (r(194)= -.70. p<.01). The IM and SDD scales likely correlated more strongly with the PC scale than did the SDE because these scales all ask questions related to morals and the treatment of other people. The SDE scale, on the other hand, while measuring a form of SDR, asks questions that capture a person's overconfident belief in his or her abilities or competence.
Scores in the three conditions
PC scale. Analysis of the means for PC scores confirmed predictions. Prejudice Obvious condition had the highest PC score (90.8), Camouflage had moderate PC scores (83.1) and Pseudo Bogus Pipeline had the lowest PC scores (73.6). ANOVA revealed a significant difference in means between the groups (F(2, 194) = 18.53, p<.001). Test of Linearity showed that the difference in means was almost entirely linear (F(2, 194) = 36.95, p<.001). There was virtually no deviation from linearity in means across the three conditions (F(1, 194) = .109, ns). Multiple comparisons of means also revealed that means for each condition differed significantly from each other. Pairwise contrasts showed that the Prejudice Obvious condition had significantly higher scores than the Camouflage condition (t(1,192) = 2.63, p<.03) and Pseudo Bogus Pipeline condition (t(1,192) = 6.07, p<.001). The Camouflage condition also had a significantly higher mean than the Pseudo Bogus Pipeline condition (t(1,192) = 3.24, p<.005). These results provide strong evidence of the validity of the PC scale because they show that people are more likely to lie to appear unprejudiced when there is situational pressure to do so and less likely to lie to appear unprejudiced when there is pressure to tell the truth. The PC scale was found to have high internal consistency (Cronbach's alpha = .80).
Modern Racism. ANOVA revealed an overall difference in means between all three conditions (F(2, 194) = 3.32, p<.05). Means for the Modern Racism scale were almost identical in the Camouflage and Pseudo Bogus Pipeline conditions (13.08 and 13.65, respectively) but significantly less than in the Prejudice Obvious condition (11.81) (t(1,192) = 2.53, p<.05). However, there was no significant difference either between the Prejudice Obvious and Camouflage conditions (t(1,192) = 1.68, ns) or between the Camouflage and Pseudo Bogus Pipeline conditions (t(1,192) = .075, ns). Test of Linearity also showed that there was a significant linear relationship between means for all three conditions (F(2, 194) = 6.35, p<.02). These results generally confirmed the hypothesis that people are less likely to reveal their prejudices when they are aware that the prejudices are being assessed.
Attitudes Toward Gays/Lesbians. The means for the Attitudes Towards Gays/Lesbians scale were almost identical in the Camouflage and Pseudo Bogus Pipeline conditions (17.4 and 17.8, respectively) and slightly lower in the Prejudice Obvious condition (16.8). ANOVA revealed neither significant difference between the conditions (F(2, 194) = .55, ns), nor a linear relationship between the means (F(2, 194) = 1.07, ns). Again, the lack of difference between the conditions when measuring attitudes towards gays and lesbians may be due to the possibility that prejudice toward gays and lesbians is not as stigmatized as prejudice toward Blacks. Therefore, the pressure to lie to appear unprejudiced toward gays and lesbians may not be as strong as the pressure to appear unprejudiced to Blacks.
Although interest in prejudice and stereotypes has been high in recent years, a large portion of the research has focused primarily on which groups people are prejudiced and the effects of prejudices and stereotypes (Herek, 1996; Fazio, Jackson, Dunton and Williams, 1995; Devine, 1989). Also, while there is research that somewhat suggests that people may lie to appear unprejudiced (Plant and Devine, 1998; Wittenbrink, Judd and Park, 1997; Jones and Sigall, 197l), this is the first study that seeks to find direct evidence. Though there are currently scales that deal with assessing people's prejudices (Plant and Devine, 1998; McConahay, 1986), the PC scale is the only known scale that captures people's tendency to lie to appear unprejudiced.
Do people lie to appear unprejudiced?
This research addressed two fundamental questions; 1) would people deliberately lie on questionnaires to appear unprejudiced? and 2) is it possible to develop a valid and reliable scale to measure the degree people to which lie to appear unprejudiced? Both Study One and Study Two provided evidence that suggests that the answer to both these questions is "yes".
Many people are very sensitive to the stigmas that have become associated with being prejudiced and may be highly vigilant against being perceived as prejudiced. The results of both studies strongly suggest that when placed in a situation where it is known that prejudice is being assessed, people are unlikely to be completely truthful about their prejudices. Only when they were not aware that their prejudices were being assessed or when confronted with the possibility that their true beliefs will be known were they more likely to be truthful about their prejudices.
The results confirmed the hypothesis that people who would lie to appear unprejudiced do so because they are worried about the perception that others will have of them (i.e., being perceived as bigots). This need to appear desirable manifested itself not only in responses on the PC scale, but also on a number of scales capturing socially desirable responding.
The Impression Management and Self-Deceptive Denial scales had the strongest relationship with the PC scale. This might be because prejudice is a moral issue. Presumably, people lie to appear unprejudiced to avoid being perceived as immoral or unkind. While the Self-Deceptive Denial and Impression Management scales do not measure prejudice, they do capture the tendency for one to portray himself as unrealistically moral or altruistic. It is the fact that these measures of socially desirable responding share this trait with the PC scale that explains the strong relationship.
On the other hand, the Self-Deceptive Enhancement scale does not measure a person's desire to appear overly moral. Rather, it captures a willingness to appear more competent and able than is realistic. This scale moderately correlates with the PC scale because, while it is a measure of SDR, it does not address the morals of the person.
Validity of the PC scale
The current research also provided compelling evidence supporting the validity of the PC scale. The strong discrepancy between the means of PC scores across the three conditions suggests that the scale does capture the tendency to lie to appear unprejudiced. Furthermore, the relationship between the PC scale and the SDR and prejudice scales established construct validity of the PC scale. Overall, results suggest that the PC scale is a valid measure with high internal consistency.
Moreover, since the scale is intended to capture lying to appear unprejudiced, then, by definition, those who score high on the PC scale should also report less prejudice on other scales. The relationship between Modern Racism scores and PC scores showed exactly that. Study Two, however, did not find any relationship between Politically Correct Responding and scores on the Attitudes Toward Gays and Lesbians scale. Indeed, no relationship between ATG/L and any other measure was found. There may be multiple explanations for these findings. For example, there may not be especially strong stigmas placed on prejudice against homosexuality. Being gay, unlike being an ethnic minority, may be viewed as a conscious choice to pursue a lifestyle that is in opposition to accepted norms and is therefore more open to criticism. Therefore, there is less pressure to appear unprejudiced towards gays, which would explain the lack of a relationship to either the SDR scales or the PC scale.
Limitations and future research
Future studies will need to be conducted to further validate the PC scale by addressing some of the limitations that were present in these studies. First, the conclusions drawn from this research are based entirely on data gathered from college students. This obviously limits the validity of the scale by either excluding or not controlling for people from other backgrounds. Second, research on prejudice has shown that there are differences in participants' responses when experimenter race is controlled (Fazio, Jackson, Dunton, and Williams, 1995). It is necessary, then, for future research to test the relationship between PC scores and experimenter race. One can only speculate on what the findings of such research will be.
The current research also provided evidence suggesting that situational factors play a part in PC-lying. It might then be both interesting and beneficial for research to be performed addressing specifically what other situations might be, the degree to which they influence PC-lying and who is most likely to be influenced in his/her responses to these situations.
Conclusion: The PC scale as a tool
A complete and valid PC scale is a potentially useful tool in prejudice and stereotype research. Lying to appear unprejudiced is a potentially major methodological problem in many studies of stereotypes and prejudice. It is also a potentially interesting and important theoretical phenomenon in its own right. PC-lying may explain why individuals report less prejudice toward minorities. It may also help to explain the differences in implicit and explicit responding on issues relating to prejudice. Some research on prejudice finds that Whites often evaluate a Black target more favorably than an identical White target (Jussim, Coleman, and Lerch, 1987; Linville and Jones, 1980). Though there is no evidence that there is lying involved, administering the PC scale would, at minimum, rule out one explanation for these puzzling findings.
Socially desirable responding scales have become a commonplace check on assessing the validity of self-reports. The PC scale has the potential to become a routine and useful check on research assessing prejudice and stereotypes.
This research was funded in part by a grant from the Henry Rutgers Scholars Program and is based on a thesis in the fulfillment of the requirements of the Henry Rutgers Scholars Program. We are grateful for the outstanding contributions and dedication of the following individuals: Melissa Filippone and Pat Seunarine for their invaluable assistance in data collection, data entry and data analysis; Eric Malek, Carolyn Sharkey, Li-Yun Chen, and Renee Prince for their help in the long and tedious tasks of data collection and data entry.
Devine, P.G. (1989). Stereotypes and Prejudice: Their automatic and controlled components. Journal of Personality and Social Psychology, 56, 5-18.
Dunton, D. C. and Fazio, R. H. (1997) An Individual Difference Measure of Motivation to Control Prejudiced Reactions. Personality and Social Psychology Bulletin, 23, 316-326.
Fazio, R. H, Jackson, J. R, Dunton, B. C. and Williams, C. J. (1995) Variability in automatic activation as an unobstrusive measure of racial attitudes: A bona fide pipeline? Journal of Personality and Social Psychology. Vol 69(6), 1013-1027.
Herek, G. M. and Capitanio, J. P. (1996) "Some of my best friends": Intergroup contact, concealable stigma, and heterosexuals' attitudes toward gay men and lesbians. Personality and Social Psychology Bulletin. Vol 22(4), 412-424.
Jones, E. E. and Sigall, H. (1971) The Bogus Pipeline: A new paradigm for measuring affect and attitude. Psychological Bulletin, 76, 349-364
Jussim, Coleman, and Lerch, L. (1987). The nature of stereotypes: A comparison and integration of three theories. Journal of Personality and Social Psychology, 52, 536-546.
Karlins, M, Coffman, T. L. and Walters, G. (1969) On the fading of social stereotypes: Studies in three generations of college students. Journal of Personality and Social Psychology. 13(1), 1-16
Katz, D. and Braly, K. (1933) Racial stereotypes of one hundred college students. Journal of Abnormal and Social Psychology. 28, 280-290.
LePore, L. and Brown, R. (1997). Category and Stereotype Activation: Is Prejudice Ineveitable? Journal of Personality and Social Psychology, 72, 275-287.
Linville, P. and Jones, E, E. (1980). Polarized appraisal of out-group members. Journal of Personality and Social Psychology, 38, 689-703.
Madon, S. J. (1999) The Best Guess Model of stereotyping: Information seekers versus cognitive misers. Dissertation Abstracts International: Section B: the Sciences and Engineering Vol 59(12-B), Jun 1999, 6517, US: Univ. Microfilms International.
McConahay, J. B. (1986) Modern racism, ambivalence, and the Modern Racism Scale. [Chapter] Dovidio, John F. (Ed); Gaertner, Samuel L. (Ed). Prejudice, discrimination, and racism. (pp. 91-125).
Paulhus, D.L. (1991). Measurement and control of response bias. Measures of personality and social psychological attitudes.
Paulhus, D.L. (in press) Socially desirable responding: Evolution of a construct. To appear in H. Braum, D., E. Wiley, and D.N. Jackson (Eds) Under construction: The role of construct validity in psychological and educational measurements. Hillsdale, NJ: Erlbaum.
Weitz, S. (1972). Attitude, Voice and Behavior: A Repressed Affect Model of Interracial Interactions. Journal of Personality and Social Psychology, 24, 14-21.
Wittenbrink, B., Charles, J.M. and Park, B. (1997). Evidence for Racial Prejudice at the Implicit Level and Its Relationship With Questionnaire Measures. Journal of Personality and Social Psychology, 72, 262-274.
Table 1a. Descriptive statistics for scales in study one
Table 2a. Correlation between scales for study one
Table 3a. ANOVAS on main dependent variables in study one
*Note: Means that do not share at least one superscript (a, b) differ at p<.05.
Table 1b. Descriptive statistics for scales in study two
Table 2b. Correlations between scales in study two
Table 3b. ANOVAS on main dependent variables in study two
*Note: Means that do not share at least one superscript (a, b, c) differ p<. 03 or less.
Copyright 2002 by Lee Jussim
Current URL: http://rutgersscholar.rutgers.edu/volume04/walkjuss/walkjuss.htm