Implementation of a Computer Based Implicit Association Test as a Measure of Attitudes toward Individuals with Disabilities.

In 1990, the Americans with Disabilities Act (ADA) was passed to assist individuals with disabilities in securing jobs and to improve the treatment of job incumbents by employers and employees. The United States Census Bureau reported in 1994-1995 that individuals with disabilities were less likely to be employed and that earnings were likely to be lower than earnings by individuals without disabilities. At that time, 23% of individuals with a work disability (as defined by the ADA) and 73% of individuals with a severe work disability were not in the labor force (U.S. Census Bureau, 1995). Although the ADA was enacted over fifteen years ago, the employment of individuals with disabilities has not increased in more recent years. According to the 2000 U.S. Census, 74% of working age (16 to 76 years old) individuals with work disabilities (as defined by the ADA) were not in the labor force. Furthermore, for that same age range, 91% of individuals with a severe work disability were not in the labor force. For those with a work disability that were able to find gainful employment, evidence indicates that they may not have been treated equitably once in the workplace. For example, in 1999, the mean earnings of individuals with a work disability were only $19,745 compared to mean earnings of $32,000 for individuals without a work disability (U.S. Census Bureau, 2000a, 2000b).

As indicated by the above numbers, individuals with disabilities face at least two key obstacles in the workforce: (a) access to jobs and (b) treatment as a job incumbent. In reference to access to jobs, individuals with disabilities may have physical obstacles that prevent entry into the workforce (Feldman, 2004; Nietupski & Hamre-Nietupski, 2000). Additionally, recruitment practices and selection procedures may unfairly eliminate individuals with disabilities from jobs for which they are otherwise qualified (Drehmer & Bordieri, 1985; Hernandez, 2000; Satcher & Dooley-Dickey, 1992; Stone, Stone, & Dibpoye, 1992).

As job incumbents, individuals with disabilities may also be at a disadvantage when compared to individuals without disabilities. Various researchers have found that individuals with disabilities receive lower pay and benefits (U.S Census Bureau, 2000a), receive fewer opportunities for training (Reyna & Sims, 1995), and may receive biased performance appraisals (Colella, DeNisi, & Varma, 1997). Also, individuals with disabilities when compared to individuals without disabilities have lower promotion rates (Bordieri, Drehmer, & Taylor, 1997) and shorter job tenure (Colella, 1994). Socially, individuals with disabilities have been found to have fewer relevant role models (Jones, 1997), and may even be subject to out-group membership status (Jones). Furthermore, stigmatization associated with their disability may lead individuals with disabilities to feel self-conscious about how they are perceived and about their behaviors in social situations, which may in turn lead them to avoid the development of social relationships (Livneh, Lott, & Antonak, 2004; Stone et al., 1992). By isolating themselves in such a fashion, individuals with disabilities may experience depression and anxiety. Finally, individuals with disabilities may have higher rates of attrition in organizations than individuals with no disability (Lerner, Adler, Chang, Lapitski, Hood, Perissinotto, Reed, McLaughlin, Berndt, & Rogers, 2004). Taken together, these various difficulties create consequences for individuals with disabilities, for organizations, and for society at large.

Given the unique set of challenges faced by individuals with disabilities in the workplace, it is important to accurately measure and identify potentially biased attitudes toward individuals with disabilities. That is, one way to begin eradicating these obstacles is to focus on measuring the attitudes that interviewers, supervisors, co-workers and subordinates may have toward individuals with disabilities. Along those lines, several instruments using direct response techniques have been developed to specifically measure "explicit" attitudes toward individuals with disabilities. Some examples of direct paper-and-pencil measures are the Attitudes Towards Disabled Persons Scale (ATDP; Yuker, Block, & Campbell, 1960), the Disability Factor Scales (DFS; Siller, Ferguson, Vann, & Holland, 1967), the Scale of Attitudes Toward Disabled People (SADP; Antonak, 1982), and the Interaction with Disabled Persons Scale (IDP; Gething, 1994).

Although these direct response methods have been widely used, they are susceptible to numerous threats to validity (Antonak & Livneh, 2000). Cannon and Szuhay (1986) found that rehabilitation counseling students, when asked to fake the ATDP, had significantly higher scores than those asked to respond honestly. In addition, Yuker (1986) concluded that some people in certain conditions, such as when individuals have a motive to provide socially desirable answers, have done so on the ATDE It is important to note that these direct response limitations are not constrained to the most often used scale, the ATDP. Thomas (2001) and Gething (1994) have both found other commonly used scales, such as the IDP and the SADP, are likewise susceptible to these same validity issues.

Measuring attitudes through an indirect response method may provide a way to reduce and potentially eliminate these threats to validity found with direct response methods. Traditional examples of indirect response methods are projective techniques (e.g., Ford, Liske & Ort, 1962), disguised measures (e.g., Comer & Piliavin, 1975), behavioral observations (e.g., Cacciapaglia, Beauchamp, & Howells, 2004), physiological methods (e.g., Zych & Bolton, 1972), and the randomized response technique (e.g., Antonak & Livneh, 1995; Antonak & Livneh, 2000). While these indirect methods have been successful in reducing the impact of socially desirable responding and faking in the laboratory, many of these methods are impractical to apply to interviewers, supervisors, and coworkers in an organizational setting.

Another recently developed indirect method of measuring attitudes toward individuals with disabilities that seems feasible to conduct in the real world is the Implicit Association Test (IAT; Greenwald, McGhee, & Schwartz, 1998). The IAT measures implicit attitudes of a rater. Implicit attitudes, as defined by Greenwald et al. (1998), are automatic evaluations activated without the individual's awareness. These automatic evaluations are evidenced in actions and judgments, but unlike direct measures of these attitudes, are thought to be outside of the person's conscious control. In short, the IAT attempts to assess the increased cognitive processing time that occurs when non-compatible target concepts and attributes are placed together (e.g., Cancer and Pleasant) versus when compatible target concepts and attributes are placed together (e.g., Cancer Free and Pleasant). It is thought by most researchers in the area of implicit attitudes (e.g., Greenwald et al., 1998; Jajodia & Earleywind, 2003; Teachman & Woody, 2003, etc.) that this increased cognitive processing time may be due to the underlying bias the individual has toward the target group (individuals with Cancer in this particular case). This mean difference in response time between congruent and incongruent categorizations, measured in milliseconds, is technically referred to as the IAT effect (Greenwald et al., 1998). The larger the IAT effect, the more potential the rater has to be biased against individuals with disabilities.

The Implicit Association Test Procedure

Stated more technically, the IAT administration consists of five major steps in which participants are presented with either the discrimination category of interest (e.g., for disability: Cancer and Cancer Free; for gender: female and male, etc.), the evaluative attribute dimension (e.g., Pleasant and Unpleasant), or a combination of both a discrimination category and an attribute dimension (for example, using hypothesized incongruent pairings: Cancer--Pleasant and Cancer Free--Unpleasant). The participant is then instructed to use a computer keyboard to designate the stimulus items into one of the categories or dimensions by pressing appropriate keys with his/her left or right forefinger as quickly as possible. Each side of the discrimination category (e.g., Cancer and Cancer Free) or attribute dimension (e.g., Pleasant and Unpleasant) is shown on opposite sides of the screen. The stimulus item that needs to be categorized is presented in the middle of the screen. The participants then press the appropriate key to either assign the neutral stimulus item to the discrimination category (or attribute dimension) on the right-hand side of the screen using the right forefinger or the discrimination category (or attribute dimension) on the left-hand side of the screen using the left forefinger (Greenwald et al., 1998).

The first of the five major steps in the IAT administration introduces the discrimination category (e.g., Cancer and Cancer Free) to the participant. At first, a screen appears with the word Cancer on the top left-hand side of the screen and the words Cancer Free on the top right-hand side of the screen. One at a time a stimulus word such as weak or sick appears in the center of the screen and the participant assigns the word to either the Cancer category by pressing the key assigned to the left-hand side of the screen or assigns it to the Cancer Free category by pressing the key assigned to the right-hand side of the screen.

The second step is the introduction of the attribute dimension (e.g., Pleasant and Unpleasant). At first, a screen appears with the word Unpleasant on the top left-hand side of the screen and the word Pleasant on the top right-hand side of the screen. One at a time a stimulus word such as flower or thorn appears in the center of the screen and the participant assigns the stimulus word to either the Unpleasant category by pressing the key assigned to the left-hand side of the screen or assigns the word to the Pleasant category by pressing the key assigned to the right-hand side of the screen. Thus, the only real difference between steps 1 and 2 is that step 1 involves assigning neutral words to the discrimination category of interest while step 2 involves assigning neutral words to attribute dimensions.

In the third step, the discrimination category (Step 1) and attribute dimension (Step 2) are combined into a compatible formation (e.g., Cancer and Unpleasant on one side and Cancer Free and Pleasant on the other). That is, all four words appear on the computer screen simultaneously, with Cancer and Unpleasant appearing together on the left-hand side of the screen and Cancer Free and Pleasant appearing together on the right-hand side of the screen. Once again, the stimulus items are then presented one at a time in the middle of the screen and the participant is asked to use their right hand or their left hand to place the stimulus words into their appropriate dimension or category.

In the fourth step, the participant learns the reversal of hand response assignments for the discrimination category (i.e., the previous hand assignments for Cancer and Cancer Free are switched). Note that in this step the attribute dimensions are omitted so that only Cancer and Cancer Free appear on the sides of the screen. This is similar to step 1 but it is important to understand that in step 4 Cancer would appear on the right-hand side of the screen if it was presented on the left side of the screen in step 1.

Finally in step 5, the attribute dimension, which has no change in hand response assignments, is combined with the reversed discrimination category into a non-compatible formation on each side of the screen (i.e., Cancer Free and Unpleasant on one side and Cancer and Pleasant on the other side). That is, all four words appear on the screen as in step 3 but now they are placed in non-compatible pairs. The stimulus words are presented one by one and the participant is asked to assign these words to either the Cancer condition or the Cancer Free condition. However, as Cancer now is paired with Pleasant and Cancer Free is now paired with Unpleasant, it often takes participants longer to assign the stimulus words because of the presence of the noncompatible attribute dimension. It is thought that the more biased an individual is toward individuals with cancer, the longer it will take that individual to place a positive word in the Cancer condition and the longer it will take that individual to place a negative word into the Cancer Free condition. This measure is calculated over several stimulus words in steps 3 and 5 and is recorded in milliseconds. The difference in the average time it takes to correctly assign stimulus words to a discrimination category in step 5 and the average time it takes to correctly assign stimulus words in step 3 is the actual measure of the IAT effect.

Evidence for Utility of IAT for Measuring other Attitudes

Evidence for the validity of the IAT in measuring attitudes toward gender, race, and other groups has been found by several researchers. Greenwald et al. (1998) found results indicating the usefulness of the IAT in measuring implicit attitudes. They concluded that the IAT is probably more resistant than explicit measures to validity threats such as social desirability bias and the evaluation apprehension effect. Greenwald et al. (1998) also found evidence for convergent and divergent validity of the IAT when used to measure attitudes toward race. Cunningham, Preacher, and Mahzarin (2001) demonstrated convergent validity for the IAT as well. They used a confirmatory factor analysis with three implicit attitude measures, including the IAT, to demonstrate the convergent validity of the technique. Not only did Cunningham et al. find that the implicit attitude measures correlated with each other, but they also found that the implicit attitude measures formed a single, overall, latent construct.

In addition, other researchers have found that the IAT does not simply capture the same information as the more traditional direct measures of attitudes toward gender, race, and other groups. Rudman, Greenwald, Mellott, and Schwartz (1999) found the IAT measures and self-report measures of the same attitude were not related to each other. Greenwald and Farnham (2000) demonstrated the stability of the IAT in measures of self-concept and self-esteem. They found evidence for construct validity in the form of known-groups validity and divergent validity. Evidence of known-groups validity was found in that the IAT had high sensitivity to known differences between genders in self-concept of masculinity and femininity. There were low correlations between the IAT measures and more explicit measures that again indicated evidence of divergent validity of the IAT when compared to more direct measures of attitudes.

Overall, the potential validity of the IAT and its potential utility over traditional measures has been demonstrated in its correlation with other implicit measures of attitudes (Greenwald et al., 1998) and its lack of relationship with explicit measures of attitudes (Gawronski, 2002). Despite the overwhelming popularity and robustness of the utilization of various forms of the IAT in attitude research, it has only recently begun to be adapted for the purposes of disability research (Pruett &Chan, 2006). Pruett and Chan developed a paper-and-pencil version of the IAT to measure attitudes toward individuals with disabilities called the Disability Attitude Implicit Association Test (DA-IAT). The DA-IAT measures the number of correctly categorized items in 20 second blocks in a paper-and-pencil format. Traditionally, however, researchers have found (Greenwald, 2004) that computer based IATs have produced larger and more robust findings than paper-and-pencil IATs. Pruett and Chan, in fact, found the DA-IAT to be only marginally related to a battery of demographic and psychosocial variables and called for the development of a computer based version of the IAT.

In summary, using the IAT may reveal attitudes and other automatic associations that individuals would prefer not to express and probably would not express on a more direct measure, particularly when there is an incentive to respond in a more socially acceptable fashion (Greenwald et al., 1998). The IAT has been used successfully to assess implicit bias for racial attitudes (McConnell & Leibold, 2001), gender attitudes (Aidman & Carroll, 2003), self-esteem (Greenwald & Farnham, 2000), self-concept (Asendorpf, Banse, & Muecke, 2002), religious beliefs (Rowatt, Franklin, & Cotton, 2005), ethnicity (Rudman, Greenwald, & Mellott, 1999), age (Castelli, Zecchini, & Deamicis, 2005), nationality (Greenwald, et al., 1998) and smoking behavior (Swanson, Rudman, & Greenwald, 2001). The current study was designed to develop a computer based IAT to measure attitudes toward individuals with disabilities. By nature, this group is one for which raters may be particularly reluctant to share their true attitudes. This reluctance has been evidenced in the numerous studies finding socially desirable responding on the more direct measures of attitudes toward individuals with disabilities (Antonak & Livneh, 2000; Hagler, Vargo, & Semple, 1987; Loo, 2001). Experiment 1 assessed the basic psychometric properties of the computer based IAT. In addition to the development of the materials, the goal of experiment 1 was to test to see if individuals would differentiate different disabilities with the IAT and to see if there were relationships between race and gender with the IAT developed herein. With the use of direct measures, gender has been found to be correlated with attitudes toward individuals with disabilities, with women holding less negative attitudes than men (Yuker, 1988; Jones & Stone, 1995). Antonak and Livneh (1988) likewise reported that on the most commonly used measure, the ATDP, females displayed greater acceptance of physical disability than did males. In addition, the more direct attitude measures have also shown differential responding due to the race of the participant (e.g. Satcher & Dooley-Dickey, 1992; Sciarra, Chang, & McLean, 2005). It remains to be seen if these gender and race effects would exist on a more implicit measure.

Pilot Study

Method

A pilot study was conducted to develop the discrimination categories for the disabilities to be used for the administration of the Implicit Association Test. These are the words that go on the sides of the screen in step 1 (and that eventually get paired with either compatible or non-compatible attribute dimensions in steps 3 and 5, respectively). Students enrolled in psychology courses at a medium-sized Mid-Western university were given a list of six disability conditions and six non-disability conditions in order to generate potential stimulus words with each disability. They were instructed to generate one-word adjectives or descriptors about the particular disability or non-disability condition. A total of 12 disability exemplars had resulted from a thorough literature review (Janicki, 1970; Jones & Stone, 1995; Lyons & Hayes, 1993; Schmelkin, 1984; Thomas, 2000; Tringo, 1970). The 12 disability conditions that were selected were familiar to the general population. These conditions included the following: asthma and no asthma, heart disease and healthy heart, blind and non-blind [sic], drug addicted and drug free, mental illness and mental health, amputee and non-amputee, alcoholic and non-alcoholic, cancer and cancer-free, deaf and non-deaf, paralyzed and non-paralyzed, HIV positive and HIV negative, and learning disabled and non-learning disabled. [Please note the authors felt the use of less handicapping language such as individual with alcoholism, individual with paraplegia, etc. would be problematic given the IAT involves reaction times to stimuli.]

Forty-four students completed the pilot study, and, to prevent fatigue and boredom, each participant completed materials for six of the 12 disability conditions and six of the 12 non-disability conditions. For all 12 conditions, the words from all participants were combined into a master list. A group of four subject matter experts (two psychology graduate students and two psychology professors) then judged the viability of the 12 disability exemplars and the words generated by the participants for each of the disability exemplars. Five disabilities were judged to have a sufficient number of common words listed by the participants to be viable discrimination categories. That is, given the need to provide numerous trials of stimulus words to be categorized; only five potential discrimination categories were found to have enough seemingly usable words for the typical IAT administration.

The pilot study resulted in an adequate set of stimulus words for the following five disabilities: (a) Alcoholic, (b) Cancer, (c) HIV Positive, (d) Mental Illness, and (e) Paraplegic. Further investigation, however, demonstrated that HIV Positive was not a satisfactory discrimination category to include in the IAT administration because of the inverse meaning of HIV Positive and HIV Negative (again, this caused problems with lengthening reaction times as did the use of more politically correct, longer, language). Thus, the four other disabilities were adopted as discrimination categories and the stimulus words provided by the participants of the pilot study were adopted for use in the alpha version of the computer based IAT. The complete list of stimulus words can be obtained from the first author. As one example, however, typical stimulus words associated with the discrimination category "Alcoholic" were weak, abusive, addicted, careless, immature, and compulsive while the typical stimulus words associated with "Non Alcoholic" were sober, stable, controlled, dependable, disciplined, and responsible.

Experiment 1

Method

Participants. Participants in Experiment 1 were 129 students at a medium-sized Mid-Western university enrolled in psychology classes that participated in this study for extra credit. Of those reporting the following information, the sample consisted of 48 (39.3%) males and 74 (60.7%) females. The mean age of participants was 20.2 years (SD = 3.0). One hundred and three (84.4%) of the participants were Caucasian, while 10 (8.2%) were African American, three (2.5%) were Hispanic/Latino, two (1.6%) were Native American, two (1.6%) were Asian American, and two (1.6%) identified themselves as "Other."

Materials

Informed Consent Form. The informed consent form explained the major purpose for and the key steps in the experiment. Participants were informed of any potential risks and benefits associated with the experiment. The informed consent form also informed participants that they could refuse to participate or withdraw from the study at any time.

Biographical Data. Participants provided their gender, race, and age.

Dependent Measures. The dependent measures for the IAT administration were the participant's response time difference for each of the four separate IAT tests (i.e., Cancer, Mental Illness, Paraplegic, and Alcoholism). The response time difference was calculated for each of the four LAY tasks by subtracting the mean response time to categorize the stimulus items into a set of compatible groups (e.g., Cancer and Unpleasant versus Cancer Free and Pleasant) from the mean response time to categorize the stimulus items into a set of non-compatible groups (e.g., Cancer and Pleasant versus Cancer Free and Unpleasant). Thus, for all four IAT tests, participants received a response time difference score in milliseconds as typically done in such research. Additionally, an overall average IAT score was computed by adding the four IAT tests (i.e., Cancer, Mental Illness, Paraplegic, and Alcoholism) together and dividing the total by four.

Procedure

First, the general purpose and procedure of the study were explained to the participants; they were then asked to read and sign an informed consent form. The participants completed the biographical information form. Then, the participants completed the four IAT measures of disability conditions on a computer situated in a laboratory environment under direct supervision of a trained graduate student.

IAT Administration. The four IAT tests were administered individually to each participant. The participants were seated at the computer and then given verbal instructions regarding how to complete the IAT. The researcher initiated the first IAT for each participant. At the beginning of the IAT, two categories appeared on opposite sides of the computer screen, and a series of stimulus items randomly appeared in the middle of the screen. The participant was instructed to use the 'A' key for left-handed responses and the '5' key on the number pad for right-handed responses. If the stimulus item belonged in the category on the left side of the screen, the participant was instructed to push the 'A' key and if the item belonged in the category on the right side of the screen, they were instructed to press the '5' key. If the participant attempted to place the word in the wrong category, a red X appeared at the bottom of the screen. This procedure was followed for all trials. Each IAT test consisted of seven total trials. Following the five step procedure laid out above, steps 1 and 2 were conducted twice so that participants would have a better opportunity to learn the associations between the stimulus words and the discrimination categories (step 1) and between the stimulus words and the attribute dimensions (step 2). That is steps 1 and 2 were repeated so that they became trials 1, 2, 3 and 4. Then trial 5 was step 3 from above with a compatible combination of category and dimension (e.g., Cancer and Unpleasant and Cancer Free and Pleasant). Trial 6 was step 4 from above with the reversal of the discrimination category hand response. Trial 7 was step 5 from above with an incompatible combination of category and dimension (e.g., Cancer and Pleasant and Cancer Free and Unpleasant). After the participant completed the first IAT (Paraplegic), the researcher initiated the next IAT for them (Alcoholic). Again, the participants then completed the seven trials associated with the second disability condition. Each participant completed a total of four different IATs and all participants did so in the following order: Paraplegic, Alcoholic, Cancer, and Mental Illness.

Results

As mentioned above, it was possible for participants to incorrectly classify stimulus words to discrimination categories. Some participants had error rates that suggested excessive inattention to responses or outright incorrect pairing of stimulus words to discrimination categories. The decision was made to delete those participants who had a congruent pair error rate greater than 27%. Error rate was calculated as the total number of incorrect pairings of stimulus words to discrimination category across all four conditions during the congruent situation divided by the total number of stimulus word assignments completed during the congruent situation. Or in other words, the percentage of items incorrectly classified by the participant during step 3 (trial 5). This decision rule was a compromise between numbers previously reported in existing IAT research. Greenwald and Farnham (2000) chose to delete participants with error rates greater than 20 % while McFarland (2003) chose to delete participants with error rates greater than 33%. Based on a 27% decision rule, a total of seven participants were deleted, leaving one hundred and twenty two participants.

Descriptives and Correlations of the IAT Administrations. For each of the four IATs (i.e., Paraplegic, Alcoholic, Cancer, and Mental Illness), each participant received a response time difference score. The mean for each condition is the average reaction time difference in milliseconds between the compatible pairing of dimension category and attribute dimension (step 3) and the incompatible pairing of dimension category and attribute dimension (step 5). As the IAT was developed as an exploratory tool, the Pearson product moment correlations were computed to determine the bivariate relationships between the four IAT tests. Additionally, as this is the first use of the IAT for this purpose and norms might be informative, the means, standard deviations, and correlations among the disabilities are provided in Table 1. All four disability conditions were significantly correlated with each other. As one would also expect, all four IAT scores were significantly correlated with the overall IAT score.

Analysis of the IAT Administrations. A Repeated Measures 2-way ANOVA was conducted with race and gender as between-subjects factors. The scores received for the four different IAT tests (i.e., Paraplegic versus Non Paraplegic, Alcoholic versus Non Alcoholic, Cancer versus Cancer-free, and Mental Illness versus No Mental Illness) was the within-subjects factor. In the analysis of the effect of race, the sample was examined using only Caucasian and African American participants due to the low number of participants in each of the other groups. The Repeated Measures ANOVA indicated that there was no significant effect due to either race, F (1,109) = 2.2, p > .05, or to gender, F (1,109) = 0.0, p > .05. Furthermore, no significant interactions were found between race and gender, F (1,109) = 0.2, p > .05. The interactions between gender and IAT score, F (3,327) = 1.7, p > .05, and between race and IAT score, F (3,327) = 1.7, p > .05, were likewise non-significant. These results indicated that neither race nor gender produced any effect whatsoever on scores of the four IATs.

The four administrations of the IAT were found to be significantly different from each other however, F (3, 327) = 6.2, p <.001. Again, the mean IAT effect for each of the four disabilities are presented in Table 1. Tukey's Post Hoc analysis found Paraplegic to have a significantly higher IAT effect than Mental Illness and Alcoholic. Additionally, Cancer was found to have a significantly higher IAT effect than Mental Illness. The full results of the Repeated Measures ANOVA are presented in Table 2.

Discussion

The IAT is thought to assess automatic evaluations of the respondents without their conscious awareness by comparing the cognitive processing time required to categorize compatible concepts and noncompatible concepts. First, the current study found that it did indeed take participants longer to categorize stimulus words correctly when the discrimination category and attribute dimension were incompatible than when they were compatible. Second, the results of this study indicate that some disability exemplars (i.e., Paraplegic) required more cognitive processing time for the non-compatible grouping when compared to the compatible grouping than do other disability exemplars (i.e., Mental Illness and Alcoholic). This finding might imply that some disabilities have greater automatic association with unpleasant ideas than other disabilities. However, the general ordering of the "preference" for different disabilities is not in agreement with traditional work in the field. Some thirty years apart, Tringo (1970) and Thomas (2000), and many other researchers during that time frame (Deal, 2003; Harasymiw, Home, & Lewis, 1976; Jones & Stone, 1995; Lyons & Hayes, 1993; Schmelkin, 1984) have demonstrated a consistent "hierarchy of preference" for disabilities. This hierarchy invariably has found communicable diseases to be least preferred and physical disabilities (i.e., Paraplegia) to be most preferred by raters with ratings for mental illnesses falling somewhere in between communicable diseases and physical disabilities.

In the current experiment Paraplegic had the highest IAT score indicating the largest potential for a biased attitude toward individuals with that disability. Mental Illness had the lowest IAT score indicating the least potential for a biased attitude toward these individuals. The most obvious explanation for this result is the order in which the disability exemplars were given. Over the course of the experiment, with the exception of the result for Cancer, the average IAT effect went down. That is, the participants were exposed to a fairly unusual task. Even with the four practice trials, it appeared that participants simply got better at resolving the conflict presented in step 5 (trial 7) over the course of the experiment. It is interesting to note that traditional studies measuring attitudes toward race and gender may have missed this effect due to the fewer trials usually presented in those studies. A majority of those studies would have a single grouping of discrimination categories (i.e., male versus female). A possible alternative explanation might be that reported attitudes toward certain disability groups as traditionally measured using direct measures is not congruent with the more automatic and less controllable attitude outcomes being reported when using the IAT. This issue will be further addressed in Experiment 2.

The results of the IAT administrations did not reveal any effects due to race or gender. Although some researchers have suggested that differences in attitudes based on gender may be diminishing (e.g., Loo, 2001; Yuker & Block, 1986), previous research has generally found effects due to race (Satcher & Dooley-Dickey, 1992; Sciarra, Chang, & McLean, 2005) and gender (Nowicki & Sandieson, 2002; Wang, Thomas, Chan, & Cheing, 2003). In particular, Yuker and Block (1986) reported that women displayed more positive attitudes than did men towards individuals with disabilities (particularly in studies conducted in the United States). MacLean and Gannon (1995) also supported the gender differences with their finding that Australian men scored higher on a scale of social discomfort than did women using a direct measure, the IDP (Gething, 1994). However, these gender effects have typically been reduced in studies using more indirect measures such as projective or behavioral observation techniques (Antonak & Livneh, 2000; Cacciapaglia et al., 2004). It is possible therefore that the current results may be reflective of the implicit nature of the IAT.

A related explanation for these contradictory findings regarding the effects of race and gender is that the effects found when direct measures are being used may be due to socially desirable responding (Greenwald et al., 1998). For example, females may give more positive ratings toward individuals with disabilities not because they have more positive attitudes but because females may be more concerned about the impression that they convey to the world. If that is indeed the case, then the results of this study may indicate that the IAT seems to better capture true attitudes toward individuals with disabilities without the confound of social desirability. Of course, many other explanations are possible and Experiment 2 was designed to test the plausibility of these particular hypotheses. Nonetheless, the relationship between gender, social desirability, and IAT scores should be investigated further. Taken together, the results of the Experiment 1 indicated that the IAT may be a viable alternative to traditional paper-and-pencil measures of attitudes toward individuals with disabilities. These traditional, direct measures suffer from many shortcomings, including susceptibility to both socially desirable responding and faking. Indirect measures, such as the IAT, should not be as susceptible to these concerns. Experiment 1 found the IAT to be easy to implement and potentially valid at capturing attitudes toward different disability groups. Additionally, the finding that the IAT may be free from effects due to the gender and the race of the rater are also promising.

Experiment 2

Given the difficulties faced by individuals with disabilities in the workplace it is important to accurately measure and identify attitudes toward these individuals. Researchers have measured attitudes toward individuals with disabilities through various approaches (Altman, 1981). To date, the most widely used approach is the paper-and-pencil self report survey. As mentioned, the two most common paper-and-pencil surveys are the Attitude toward Disabled Persons Scale (ATDP; Yuker, Block, & Young, 1970) and the Interaction with Disabled Persons Scale (IDP; Gething, 1994). Each of these measures has been found to suffer from several limitations however. For example, among other things, the ATDP has been criticized for its unidimensional approach toward disabilities (Roush & Klockars, 1988; Thomas, Palmer, Coker-Juneau, & Williams, 2003) and for its susceptibility to faking (Speakman & Hoffman, 1979; Cannon & Szuhay, 1986; Vargo & Semple, 1984).

The IDP was developed in direct response to the two above mentioned criticisms of the ATDP. Gething (1994) designed the 20 item Interaction with Disabled Persons (IDP) scale to assess how an individual's level of personal discomfort might affect his/her attitude towards individuals with disabilities. An individual's feeling of discomfort is assumed to derive from the uncertainty of how to conduct oneself in the presence of an individual with a disability. Whereas the ATDP is designed as a unidimensional measure of affect, the IDP has been found to be a solid, multidimensional scale. Thomas et. al. (2003) found in an exploratory factor analysis and then confirmed in a confirmatory factor analysis three stable factors underlying the IDP. These were named social discomfort, empathy, and fear of having the disability. Social Discomfort tapped an individual's social uneasiness or anxiety level in interacting with an individual with a disability. Empathy captured the degree to which the rater could adopt the individual with disabilities perspective. Fear of having the disability captured a rater's uneasiness in interacting with the individual with a disability due to the potential of contracting the disability themselves. This factor also referred to the feeling of uneasiness associated with the realization that they could just as easily have been the individual with the disability. In addition, the three scales of the IDP were found to have adequate reliability and appropriate discriminant validity.

Gething (1994) also suggested that the IDP was less susceptible to faking than the ATDP. The IDP's approach is to measure attitudes on an individual, personal level, whereas the ATDP assesses attitudes at a societal level. Since the IDP involves asking participants to reflect on personal experiences with specific individuals with disabilities, Gething argued that participants are less likely to give socially desirable answers. Gething has reported that several studies conducted in Australia have found the IDP less susceptible to socially desirable responding than the ATDP, but those results have yet to be confirmed by other researchers. It remains to be seen if the IDP is truly free from socially desirable responding.

Traditional research has consistently, however, claimed direct measures of these attitudes (such as the IDP) are easily faked and are very susceptible to socially desirable responding. These researchers argue that whether the question is asked at the personal or societal level, any direct approach is susceptible to socially desirable responding. This research has consistently called for the development of more indirect measures of these attitudes (Antonak & Livneh, 1988, 2000; Yuker, 1986, 1988).

The computer based Implicit Association Test as developed and alpha tested in Experiment 1 may uncover attitudes and automatic evaluation associations that would not normally be expressed in everyday situations. As previously discussed, the IAT has been used to study attitudes toward race, gender, self-concept, and other constructs (Greenwald & Farnham, 2000; Greenwald et al., 1998; McConnell & Leibold, 2001; Rudman, Greenwald, Mellot, & Schwartz, 1999). In Experiment 1 the results indicated a significant difference in implicit attitudes expressed against each of the four disability exemplars used in that experiment. Further, the IAT, unlike traditional paper-and-pencil measures, was not related to race or gender of the rater. Thus, experiment 1 was taken as evidence that the IAT might be worthy of continued study.

The goal of the second experiment was therefore to further examine the feasibility of the IAT as a methodology for assessing attitudes toward individuals with disabilities. The relationship between responses on the IAT and responses for the best available paper-and-pencil measure of attitudes toward individuals with disabilities, the IDP, were investigated. Additionally, as the IAT is an indirect measure and the IDP is a direct measure, the second experiment also assessed the degree to which the IAT and the IDP were differentially susceptible to socially desirable responding.

Method

Participants. Participants were 74 students from a medium-sized Mid-Western university.

Materials

Interaction with Disabled Persons Scale. Participants completed the IDP scale. The IDP scale consists of 20 items that ask one to describe how he or she generally feels during an interaction with an individual with a disability. Responses range from 1, "I disagree very much" to 6, "I agree very much." Higher scores on the IDP indicate participants felt greater discomfort. Scores for each of the three factor scores (i.e., social discomfort, empathy, and fear of having the disability) found by Thomas et al. (2003) were also computed.

Marlowe-Crowne Social Desirability Scale. The Strahan and Gerbasi (1972) short form (10-items) of the Marlowe-Crowne Social Desirability Scale (Crowne & Marlowe, 1960) was used to measure socially desirable responding. Fischer and Fick (1993) found that the short form of the MCSDS was as equally reliable and valid as the long form. The MCSDS has a true/false response format and half of the items were reverse scored. A typical item is "I have never intensely disliked anyone."

Demographic Data. Participants were asked to provide gender, race, and age.

Implicit Association Test Administration. The computer-based IAT as developed in Experiment 1 was utilized.

Procedure. After entering the testing room, the participant was given an overview of the research being conducted. Then the participant read and signed the informed consent document, completed a brief demographics questionnaire, completed the IAT administration via computer, and finished with the paper-and-pencil versions of the IDP and the MCSDS. The administration of the IDP and the MCSDS took place in the same testing room as the IAT.

Results

As in experiment 1, the decision was made to delete those participants who had a congruent pair error rate greater than 27%. A total of five participants were deleted, leaving fifty-one females and eighteen males. Among the 69 participants, 63 indicated they were Caucasian, five were African American, and one participant identified himself/herself as other. The mean age of the participants was 20.9 (SD = 4.3).

Descriptives and the IAT Administration. The means and standard deviations for all three measures are provided in Table 3. It is important to note that the order of the IATs was different than in Experiment 1. In Experiment 2 the order was as follows: Alcoholic, Cancer, Mental Illness, and Paraplegic. [Please note the authors felt the use of less handicapping language such as individual with alcoholism, individual with paraplegia, etc. would be problematic given the IAT involves reaction times to stimuli.] Tukey's Post Hoc analysis found all disabilities to be significantly different from each other except for alcoholic and cancer. As in Experiment 1, there appears to be a practice and/or learning effect across the IAT administration. This practice effect in fact results in the "hierarchy of preference" for disabilities being somewhat unlike previous research and unlike Experiment 1. For example, Mental Illness would generally be expected to be less preferred than Cancer but that was not the case in this experiment. Based on this second occurrence of an apparent practice effect, an average IAT score was computed for each participant by adding the scores for each IAT test and dividing by four. This average IAT score was then used for the rest of the analyses.

Analyses of Relationship between IAT, IDP, and MCSDS. Pearson Product Moment Correlations were conducted to assess the relationships between the IAT Total Score, IDP Total Score, the IDP scale scores (i.e., Social Discomfort, Empathy, and Fear of Having the Disability), and the MCSDS Total Score. The results can be seen in Table 4. The IAT Total Score was significantly correlated with the IDP Total Score and the IDP subscale of Social Discomfort. Socially desirable responding, as measured by the MCSDS, was significantly correlated with the IAT Total Score. Supporting previous claims by Gething (1994), the IDP Total Score was not related to the MCSDS indicating that the IDP may not be susceptible to socially desirable responding even though it is a direct measure.

Analyses for Gender. A MANOVA was conducted with gender as the independent variable. The race variable was dropped from the analysis as there was only a single African American male. The dependent variables included the IAT Total Score and the IDP Total Score. There was no significant result due to gender for the IAT Total Score, F(1,67) = .00, p > .05 or the IDP Total Score, F(1,67) = .04, p > .05.

Discussion

The results of the current studies present mixed evidence for the potential validity and utility of the IAT as a measure of attitudes toward individuals with disabilities. In both experiments, it took participants longer to categorize stimulus words into incompatible categories than it did to categorize them into compatible categories. Also, as one would expect, the measure was related to another commonly accepted measure of the construct. The IAT was significantly related to the IDP, indicating that the IAT was possibly measuring attitude toward individuals with disabilities. The IAT was also significantly related to the Social Discomfort subscale of the IDP as well. This may suggest that the IAT is best tapping attitudes of discomfort that participants have toward individuals with disabilities or the disability itself. This might well be thought of as the "affect" associated with being a member of each particular disability group. Nonetheless, the IAT is tapping something that is related to other measures of attitudes toward individuals with disabilities.

Recall that one of the main proposed advantages of the IAT was the general belief by those in the field that it would not be susceptible to faking or socially desirable responding (Greenwald et al., 1998). In fact, that very belief was proposed as one explanation for the lack of a relationship between race and gender in attitudes as measured by the IAT in Experiment 1. It was anticipated therefore that there would not be a relationship between socially desirable responding and the IAT. However, a significant correlation was observed between the IAT and the measure of socially desirable responding. This unexpected relationship calls the validity of the IAT into question. Although the IAT is intended to remove social desirability, this response pattern appeared rather stubborn for the current sample and overcame the seemingly indirect methodology of the IAT.

One plausible explanation might be that participants learned the nature of the experiment and the methodology over the course of the experiment (recall that 4 disability exemplars were used) and those inclined to engage in socially desirable responding were able to engage in more socially desirable responding as the experiment continued. Although not reported herein, the correlation between responses on the IAT and socially desirable responding was larger for each successive disability exemplar. The implication of this finding is that someone high in social desirability may be able to fake responses on the IAT with enough exposure to the test (i.e., with enough disability exemplars or experience with IATs that measured other constructs such as race or gender attitudes). The finding that the IAT effect goes down over successive administrations in both experiments merits further research and may indicate that the IAT is not truly measuring "implicit" attitudes. Whether this decrease in IAT effect was actually caused by socially desirable responding remains to be seen, however. Another of many plausible explanations for this decrease in IAT effect is that those higher in general intelligence are quicker to figure out the IAT across administrations. Indeed, McFarland (2003) has found cognitive ability to be a confounding variable in the IAT measure of attitudes toward race. Nonetheless, the current study found the IAT to be susceptible to socially desirable responding.

Gething (1994) has found that the IDP was not related to socially desirable responding even though it is a rather transparent, direct paper-and-pencil measure of attitudes toward individuals with disabilities. To date, no other researchers had validated this claim. The current study found, somewhat surprisingly, that the IDP was indeed not related to socially desirable responding. Combined with previous research (Gething, 1994; Thomas et al., 2003), the current study provides further evidence that the IDP may be the best currently available measure of attitudes toward individuals with disabilities. Even though the ATDP is more widely used, several studies have indicated that the ATDP is quite susceptible to socially desirable responding and is not as psychometrically sound as the IDP.

Finally, previous research has found gender differences in social distance and acceptance of individuals with disabilities (Jones & Stone, 1995; Lyons & Hayes, 1993; Nowicki & Sandieson, 2002; Wang, Thomas, Chan,& Cheing, 2003) with females expressing greater acceptance than males. Therefore, it was believed that females would demonstrate more positive attitudes toward individuals with disabilities as measured by the IDP but not necessarily for the IAT as it was a more indirect measure. The current study found no gender differences for either instrument. These results for the IAT may be due to the implicit nature of the attitudes being measured by the IAT. Most of the traditional studies have used direct measures and it may be the case that such gender differences do not exist when implicit measures are used. It is also possible given no gender differences were found on the IDP that the current sample of male and female raters were unusual in some way. Alternatively, as several researchers have proposed (e.g., Loo, 2001; Yuker & Block, 1986), the fact that gender and race were not significantly different may not have anything to do with the measures used but might reflect a real shift in society.

Conclusions

The current results have important implications for practitioners. The IAT has become an accepted methodology for measuring attitudes toward race and gender. One implication of the current results, however, is that the methodology may not yet provide a valid measure of attitudes toward individuals with disabilities. Although the IAT was related to a direct measure of attitudes toward individuals with disabilities (i.e., the IDP) indicating some construct validity, the IAT was found susceptible to socially desirable responding. Given the indirect nature of the IAT this result is both surprising and quite troubling. If the IAT was truly capturing only involuntary implicit associations then it should not be related to social desirability. Additionally, if it is only capturing involuntary implicit associations then scores should not "get better" across successive administrations of the test. Practitioners should be cautious before implementing the IAT for general use in organizations as some individuals may be able to fake their responses. Until these issues are resolved, the IDP continues to provide a reliable, valid, measure of attitudes toward individuals with disabilities that does not appear to be susceptible to socially desirable responding.

Editor's Note 

Portions of this paper were presented at the meeting of the Society for Industrial/Organizational Psychology in Orlando, FL. Thomas, A, & Doyle, A. L. (2005, April). Development of an Implicit Association Test Measuring Attitudes towards Individuals with Disabilities. In N. D. Ainspan's (moderator), Measuring and Changing Attitudinal Barriers toward Hiring People with Disabilities.

References

Aidman, E. V. & Carroll, S. M. (2003). Implicit individual differences: Relationships between implicit self-esteem, gender identity and gender attitudes. European Journal of Personality, 17, 19-37.

Altman, B. (1981). Studies of attitudes toward the handicapped: The need for a new direction. Social Problems, 28, 321-337.

Antonak, R. F. (1982). Development and psychometric analysis of the Scale of Attitudes Toward Disabled Persons. The Journal of Applied Rehabilitation Counseling, 13, 22-29.

Antonak, R. F., & Livneh, H. (1988). The measurement of attitudes toward people with disabilities: Methods, psychometrics, and scales. Springfield, IL: C. C. Thomas.

Antonak, R. F., & Livneh, H. (1995). Randomized response technique: A review and proposed extension to disability attitude research. Genetic, Social and General Psychology Monographs, 121, 97-145.

Antonak, R. F., & Livneh, H. (2000). Measurement of attitudes towards persons with disabilities. Disability and Rehabilitation, 22, 211-224.

Asendorpf, J. B., Banse, R., & Muecke, D. (2002). Double dissociation between implicit and explicit personality self-concept: The case of shy behavior. Journal of Personality and Social Psychology, 83, 380-393.

Bordieri, J. E., Drehmer, D. E., & Taylor, D. W. (1997). Work life for employees with disabilities: Recommendations for promotion. Rehabilitation Counseling Bulletin, 40, 181-192.

Cacciapaglia, H. M., Beauchamp, K. L., & Howells, G. N. (2004). Visibility of disability: Effect on willingness to interact. Rehabilitation Psychology, 49, 180-182.

Cannon, B. J., & Szuhay, J. A. (1986). Faking can elevate scores on the Attitudes Toward Disabled Persons Scale. Rehabilitation Counseling Bulletin, 30, 120-123.

Castelli, L., Zecchini, A., & Deamicis, L. (2005). The impact of implicit prejudice about the elderly on the reaction to stereotype confirmation and disconfirmation. Current Psychology: Developmental Learning, Personality, Social, 24, 134-146.

Colella, A. (1994). Organizational socialization of employees with disabilities: Critical issues and implications for workplace interventions. Journal of Occupational Rehabilitations, 4, 87-106.

Colella, A., DeNisi, A. S., & Varma, A. (1997). Appraising the performance of employees with disabilities: A review and model. Human Resource Management Review, 7, 27-53.

Comer, R. C., & Piliavin, J. A. (1975). As others see us: Attitudes of physically handicapped and normals toward own and other groups. Rehabilitation Literature, 36, 206-221.

Crowne, D. P. & Marlowe, D. (1960). A new scale of social desirability independent of psychopathy. Journal of Consulting Psychology, 24, 349-354.

Cunningham, W. A., Preacher, K. J., & Mahzarin, R. B. (2001). Implicit attitude measures: Consistency, stability, and convergent validity. Psychological Science, 12, 163-170.

Deal, M. (2003). Disabled people's attitudes toward other impairment groups: A hierarchy of impairments. Disability and Society, 18, 897-910.

Drehmer, D. E., & Bordieri, J. E. (1985). Hiring decisions for disabled workers: The hidden bias. Rehabilitation Psychology, 30, 157-164.

Feldman, D. (2004). The role of physical disabilities in early career: Vocational choice, the school-to-work transition, and becoming established. Human Resource Management Review, 14, 247-274.

Fischer, D. G., & Fick, C. (1993). Measuring social desirability: Short forms of the marlowe-crowne social desirability scale. Educational and Psychological Measurement, 53, 417-424.

Ford, A. B., Liske, R. E., & Ort, R. S. (1962). Reactions of physicians and medical students to chronic illness. Journal of Chronic Diseases, 15, 785-794.

Gawronski, B. (2002). What does the Implicit Association Test measure? A test of the covergent and discriminant validity of prejudice-related IATs. Experimental Psychology, 49, 171-180.

Gething, L. (1994). The Interaction with Disabled Persons Scale. Journal of Social Behavior & Personality, 9, 23-42.

Greenwald, A. G. (2004). Revised top ten list of things wrong with the IAT. Presentation at the Attitudes Preconference of the fifth annual meeting of the Society for Personality and Social Psychology, Austin, TX.

Greenwald, A. G., & Banaji, M. R. (1995). Implicit social cognition: Attitudes, self-esteem, and stereotypes. Psychological Review, 102, 4-27.

Greenwald, A. G., & Farnham, S. D. (2000). Using the Implicit Association Test to measure self-esteem and self-concept. Journal of Personality and Social Psychology, 79, 1022-1038.

Greenwald, A. G., McGhee, D. E., & Schwartz, J. L. K. (1998). Measuring individual differences in implicit cognition: The Implicit Association Test. Journal of Personality and Social Psychology, 74, 1464-1480.

Hagler, P., Vargo, J., & Semple, J. (1987). The potential for faking on the Attitudes Toward Disabled Persons Scale. Rehabilitation Counseling Bulletin, 31, 72-76.

Harasymiw, S. J., Home, M. D., & Lewis, S. C. (1976). A longitudinal study of disability group acceptance. Rehabilitation Literature, 37, 98-102.

Hernandez, B. (2000). Employer attitudes toward workers with disabilities and their ADA employment rights: A literature review. Journal of Rehabilitation, 66, 4-17.

Jajodia, A., & Earleywine, M. (2003). Measuring alcohol expectancies with the Implicit Association Test. Psychology of Addictive Behaviors, 17, 126-133.

Janicki, M. P. (1970). Attitudes of health professionals toward twelve disabilities. Perceptual and Motor Skills, 30, 77-78.

Jones, G. E. (1997). Advancement opportunity issues for persons with disabilities. Human Resource Management Review, 7, 55-76.

Jones, G. E., & Stone, D. L. (1995). Perceived discomfort associated with working with persons with varying disabilities. Perceptual and Motor Skills, 81, 911-919.

Lerner, D., Adler, D. A., Chang, H., Lapitski, L., Hood, M. Y., Perissinotto, C., Reed, J., McLaughlin, T. J., Berndt, E. R., & Rogers, W. H. (2004). Unemployment, job retention, and productivity loss among employees with depression. Psychiatric Services, 55, 1371-1378.

Livneh, H., Lott, S. M., & Antonak, R. F. (2004). Patterns of psychosocial adaptation to chronic illness and disability: A cluster analytic approach. Psychology, Health & Medicine, 9, 411-430.

Loo, R. (2001). A psychometric re-analysis of the Interaction with Disabled Persons Scale. Canadian Journal of Behavioral Science, 33, 245-250.

Lyons, M., & Hayes, R. (1993). Student perceptions of persons with psychiatric and other disorders. American Journal of Occupational Therapy, 47, 541-547.

MacLean, D., & Gannon, P. (1995). Measuring attitudes towards disability: The Interaction with Disabled Persons Scale revisited. Journal of Social Behaviour and Personality, 10, 791-806.

McConnell, A. R., & Leibold, J. M. (2001). Relations among the Implicit Association Test, discriminatory behavior, and explicit measures of racial attitudes. Journal of Experimental Social Psychology, 37, 435-442.

McFarland, S. (2003). A cognitive skill confound on the Implicit Association Test. Social Cognition, 20, 483-510.

Nietupski, J. A., & Hamre-Nietupski, S. (2000). A systematic process for carving supported employment positions for people with severe disabilities. Journal of Developmental and Physical Disabilities, 12, 103-119.

Nowicki, E., & Sandieson, R. (2002). A meta-analysis of school-aged children's attitudes towards persons with physical and intellectual disabilities. International Journal of Disability, Development, and Education, 49, 243-265.

Pruett, S. R., & Chan, F. (2006). The development and psychometric validation of the Disability Attitude Implicit Association Test. Rehabilitation Psychology, 51, 202-213.

Reyna, M., & Sims, R. R. (1995). ADA and its implications for job training. In J. G. Veres III & R. R. Sims (Eds.), Human resource management and the Americans with Disabilities Act (pp. 127-148). Westport, CT: Quorum Books.

Roush, S. E., & Klockars, A. J. (1988). Construct validation of two scales measuring attitudes toward persons with disabilities. Journal of Rehabilitation, 54, 25-30.

Rowatt, W. C., Franklin, L. M., & Cotton, M. (2005). Patterns and personality correlates of implicit and explicit attitudes toward

Christians and Muslims. Journal for the Scientific Study of Religion, 44, 29-43.

Rudman, L. A., Greenwald, A. G., & Mellott, D. S. (1999). Measuring the automatic components of prejudice: Flexibility and generality of the Implicit Association Test. Social Cognition, 17, 437-465.

Satcher, J., & Dooley-Dickey, K. (1992). Attitudes of human-resource management students toward persons with disabilities. Rehabilitation Counseling Bulletin, 35, 248-252.

Schmelkin, L. P. (1984). Hierarchy of preferences toward disabled groups: A reanalysis. Perceptual and Motor Skills, 59, 151-157.

Sciarra, D., Chang, T., & McLean, R. (2005). White racial identity and attitudes toward people with disabilities. Journal of Multicultural Counseling and Development, 33, 232-242.

Siller, J., Ferguson, L. T., Vann, D. H., & Holland, B. (1968). Structure of attitudes toward the physically disabled: The Disability Factor Scales--amputation, blindness, cosmetic conditions. Proceedings of the Annual Convention of the American Psychology Association, USA, 3, 651-652.

Speakman, H. G., & Hoffman, C. M. (1979). The "fakeability" of the attitudes toward disabled persons scale: Form B. Physical Therapy, 59, 866-869.

Strahan, R. U., & Gerbasi, K. C. (1972). Short, homogeneous version of marlowe-crowne social desirability scale. Journal of Clinical Psychology, 28, 191-193.

Stone, E. F., Stone, D. L., & Dipboye, R. L. (1992). Stigmas in organizations: Race, handicaps, and physical unattractiveness. In K. Kelley (Ed.), Issues, theory, and research in industrial/organizational psychology (pp. 385-457). New York: Elsevier Science Publishers.

Swanson, J. E., Rudman, L. A., & Greenwald, A. G. (2001). Using the Implicit Association Test to investigate attitude-behaviour consistency for stigmatised behaviour. Cognition and Emotion, 13, 207-230.

Teachman, B. A., & Woody, S. R. (2003). Automatic processing in spider phobia: Implicit fear associations over the course of treatment. Journal of Abnormal Psychology, 112, 100-109.

Thomas, A. (2000). Stability of Tringo's hierarchy of preference toward disability groups: 30 years later. Psychological Reports, 86, 1155-1156.

Thomas, A. (2001). The multidimensional character of biased perceptions of individuals with disabilities. Journal of Rehabilitation, 67, 3-10.

Thomas, A., Palmer, J. K., Coker-Juneau, C. J., & Williams, D. J. (2003). Factor structure and construct validity of the Interaction with Disabled Persons scale. Educational and Psychological Measurement, 63, 465-483.

Tringo, J. L. (1970). The hierarchy of preference toward disability groups. The Journal of Special Education, 4, 295-306.

United States Census Bureau. (1995). Americans with Disabilities: 1994-1995. Retrieved May 13, 2002 from http:// ww.census.gov/hhes/www/disable/sipp/disab9495/asc9495 .html

United States Census Bureau. (2000a). Labor Force Status--Work disability status of civilians 16 to 74 years old, by sex: 2000. Retrieved January 24, 2002, from http://www.census.gov/ hhes/www/disable/cps/cps200.html

United States Census Bureau. (2000b). Work experience and mean earnings in 1999--Work disability status of civilians 16 to 74 years old, by sex: 2000. Retrieved January 24, 2002, from http://www.census.gov/hhes/www/disable/cps/cps300 .html

Vargo, J. W., & Semple, J. E. (1984). Honest versus fake scores on the Attitudes Toward Disabled Persons Scale-Form A. Rehabilitation Counseling Bulletin, 27, 182-185.

Wang, M., Thomas, K. R., Chan, F., & Cheing, G. (2003). A conjoint analysis of factors influencing American and Taiwanese college students preferences for people with disabilities. Rehabilitation Psychology, 48, 195-201.

Yuker, H. E. (1986). The attitudes toward disabled persons scale: Susceptibility to faking. Rehabilitation Counseling Bulletin, 29, 200-204.

Yuker, H. E. (1988). Attitudes toward persons with disabilities. New York: Springer Publishing Company.

Yuker, H., & Block, J. (1986). Research with the Attitude Toward Disabled People Scales: 1960-1985. Hampstead, NY: Holstra University Center for the Study of Attitudes Toward People with Disabilities.

Yuker, H. E., Block, J. R., & Campbell, W. J. (1960). A scale to measure attitudes toward disabled persons. (Human Resources Study No. 5). Albertson, NY: Human Resources Center.

Yuker, H. E., Block, J. R., & Young, J. H. (1970). The measurement of attitudes toward disabled persons. Rehabilitation Series Number 3. Albertson, NY: Human Resources Center.

Zych, K., & Bolton, B. (1972). Galvanic skin responses and cognitive attitudes toward disabled persons. Rehabilitation Psychology, 19, 172-173.

Adrian Thomas

Auburn University 

Andrea Doyle

Tennessee Department of Personnel 

Daly Vaughn

Auburn University

Adrian Thomas, 226 Thach Hall, Department of Psychology, Auburn University, AL 36849 Email:

Table 1
Means, Standard Deviations, and Intercorrelations for IAT Reaction
Time Scores (in milliseconds) and Overall Average IAT Score
(in milliseconds)

Variable               M       SD        1        2        3

1. "Paraplegic"        368.1   212.9   --
2. "Alcoholic"         301.8   209.0   .28 **   --
3. "Cancer"            327.2   240.0   .29 **   .32 **   --
4. "Mental Illness"    249.6   208.8   .29 **   .41 **   .41 **
5. Overall IAT Score   311.7   153.0   .65 **   .70 **   .74 **

Variable                 4      5

1. "Paraplegic"
2. "Alcoholic"
3. "Cancer"
4. "Mental Illness"    --
5. Overall IAT Score   .74 **   --

Note. N = 122. Overall IAT score for each individual was the average
of their scores for the four disability conditions.

* p <.05 ** p <.01.

Table 2
Results of Repeated Measures ANOVA

Source                df        F            [eta]   P

                      Between Subjects

Race                  1         2.2          .02     .14
Gender                1         0.0          .00     .90
Race x Gender         1         0.1          .00     .74
Error                 109       (90652.1)

                      Within Subjects

Disability (a)        3         6.2 **       .05     .00
Disability x Race     3         1.7          .02     .17
Disability x Gender   3         1.7          .02     .17
Disability x Race x   3         1.2          .01     .30
Gender

Error                 327       (29,843.7)

Note. Value enclosed in parentheses represents mean square error.

(a) Disability refers to disability condition (i.e. "Paraplegic,"
"Alcoholic," "Cancer," and "Mental Illness.")

* p< .05. ** p < .01.

Table 3
Means and Standard Deviation of Age, IDP, MCSDS, and for
IAT Reaction Time Scores (in milliseconds) and Overall
Average IAT Score (in milliseconds)

Continuous Variable           M       SD

Age                           20.9    4.3
Overall IAT Score             357.6   149.1
  "Alcoholic"                 472.7   212.2
  "Cancer"                    460.6   211.7
  "Mental Illness"            296.9   205.1
  "Paraplegic"                200.3   183.9
IDP                           72.6    11.6
  Social Discomfort           12.2    5.0
  Empathy                     25.8    3.1
  Fear of Having Disability   20.6    4.6
MCSDS                         4.6     2.2

Table 4
Intercorrelations for IAT Total Score, IDP Total Score, IDP Factors,
and MCSDS Total Score

Variable                         1        2        3        4

1. IAT Total                   --
2. IDP Total                   .25 *    --
3. Social Discomfort           .24 *    .83 **   --
4. Empathy                     -.08     .18      -.16     --
5. Fear of Having the Disability        .14      .78 **   .48 **
6. MCSDS Total                 -.30 *   -.06     -.18     .44 **

Variable                            5     6

1. IAT Total
2. IDP Total
3. Social Discomfort
4. Empathy
5. Fear of Having the Disability   .15    --
6. MCSDS Total                     -.06   --

Note. * p <. 05. ** p < .01.