The impact of Patient’s Implicit Racial Biases on Perceived Consultation Satisfaction
University of Amsterdam Graduate School of Communication Research Master of Communication Science
Thesis supervisor: Dr. Stephanie Welten Student: Beniam Gebeyehu Student number: 10583629 Date of Completion: February 5th, 2016
Abstract
Implicit racial biases have been shown to significantly interact with physician-patient communication. The focus of these studies have observed how physician’s implicit racial biases influence their communication, but have overlooked how implicit racial biases of the patient can also influence communication, which can lead to less effective healthcare service. This study aims to investigate the interaction between physician’s race and patient’s implicit racial biases and how this influences patient’s consultation satisfaction. Additionally, this study seeks to investigate if this potential moderated relationship is mediated by two common health communication factors, perceived interpersonal skills and competence of the doctor, both of which were measured implicitly and explicitly. A two condition (African American doctor vs. European American doctor) between subjects design was used to investigate this topic. Results indicate that implicit racial biases against the race of the doctor significantly moderate the relationship between the doctor’s race and ratings of consolation satisfaction. Additionally, implicit perceived interpersonal skills significantly mediated this relationship, but only in the African American condition. No effect was found for the mediation of perceived competence (both explicit and implicit) as well as explicit perceived interpersonal skills in both conditions.
Introduction
The communication between patients and health care providers is a vital component of treating patients effectively (Stewart, 1984; Roter, Hall & Katz, 1987; Kaplan, Greenfield & Ware, 1989; Ha & Longnecker, 2010). Adequate patient-provider communication has been linked with several positive treatment outcomes such as patient compliance (Roter, 1989, Wilson & McNamara, 1982) and general consultation satisfaction (Mead, Bower & Hann, 2002, Leckie, 1998). According to a report from the Joint Commission on
Accreditation of Healthcare Organizations, a lack of effective communication can have very detrimental effects. Between 2004 and 2014, communication issues caused 57% of
unexpected occurrences of death and serious physical or psychological injury (Joint commission sentinel events report, 2015).
A number of different factors can influence the communication between patient and health care provider, including incongruent ideas of patient information needs (Leydon, et. al., 2000) and the social distance between patient and provider (Mathews, 1983). One factor that is less controllable, yet still influential is the provider’s perception of the patient
(Sheehan et. al. 1985, Roter et. al. 1988, Hall et. al. 1993, Kaplan et. al. 1995). Studies have found that perception of patient communication skills, patients’ satisfaction with care and patients’ likeliness to adhere correlate with effective communication behavior from the doctor (Street, Gordon & Haidet, 2007, Rye & Burke, 2000)
Although perceptions can be created through the actual behavior of patients, research has shown that certain patient traits (ex. economic status) are significantly associated with these perceptions (Willems, et. al., 2005). For example, studies have found that doctors’ perceptions of their patients are significantly associated with the patients’ race (Rye & Burke, 2000, Street, Gordon & Haidet, 2007; Cooper, et. al., 2012; Sabin & Greenwald, 2012).
Doctors perceive African Americans as less effective communicators, less intelligent, less likely to adhere to medical advice and more likely to engage in risky behavior when
compared to European Americans (Rye & Burke, 2000; Street, Gordon & Haidet, 2007). This is particularly worrisome as some of these perceptions have already been associated with lower levels of effective communication, which would mean that certain patients are inadvertently receiving lower quality of health care. These perceptions have also been associated with behavior beyond communication such as diagnosis and treatment decisions (Burgess, et. al. 2008). Findings such as this present how the issue of inadvertent racial bias in healthcare can impact the healthcare received by people of different races.
The impact of racial biases on interactions between patient and healthcare provider has primarily focused on the doctor’s perceptions of race. Although racial bias research in healthcare has added significant insight into potential complications between patient and healthcare provider, it has neglected the fact that the racial perceptions of the patient may also have an influence. To fully understand the impact of racial biases on patient-provider communication, we should explore both sides as patients’ perceptions of doctors also contribute to this process (Hall, Roter & Rand, 1981; Boon & Stewart, 1998; Penner, et. al. 2009). Therefore, it would be of interest to assess if patients’ racial biases also play a role in forming adequate communication.
Although not much research has focused on patients’ racial biases, the influence of patients’ perceptions of doctors has been well documented. For instance, patient’s who perceive that their physicians like them, care for them and are interested in them as a person, are more active in the encounter, more satisfied and compliant with medical regimens (van Ryn & Burke, 2000). Moreover, patients who perceive that their doctor has communicated with them courteously and displayed competence seem to be more willing to comply and show higher consultation satisfaction (Willson & McNamara, 1982; Hall, Roter & Katz,
1988). Perceptions of the patient should not be underestimated, as they can be even better predictors of satisfaction than the actual behavior of the doctor (Blanchard, et. al., 1990). Therefore, examining how racial bias influences perceptions of communication can allow for a better understanding of how patient satisfaction is evaluated. Accordingly, this study will investigate how unfavorable racial perceptions or biases against the race of the healthcare provider interact with their ratings of consultation satisfaction. Additionally, this study will examine how this relationship is mediated by two common effective communication components in healthcare: perceived interpersonal skills and perceived competence of the healthcare provider.
Many studies that examine the influence of racial biases on behavior use explicit measures, which ask individuals directly about how they think about race. Although this may be the most practical form of measurement, it may not accurately predict the effect of racial bias on behavior. However, implicit measures of racial biases have shown great promise in measuring exactly this. Dovido, Kawakami and Gaertner (2002) conducted an experiment in which the explicit and implicit racial attitudes of participants were measured and used to predict how they would act in a one on one interaction with a confederate who was either white or black. They found that a respondent’s explicit attitudes were more accurate in predicting how one perceives their own behavior during the interaction, while implicit attitudes were more accurate in predicting how that behavior is perceived. Given that this study is focusing on how communication is perceived by another person, implicit measures will be utilized.
Observing how implicit or unconscious biases influence communication during medical examinations from the patient’s perspective can uncover one of the many factors that contribute to a patient’s evaluation of their healthcare provider’s communication.
Understanding this, can allow for adjustments that can lead to more effective communication and more positive patient outcomes.
Theoretical Framework
Implicit attitudes and stereotypes
The attitudes we have towards the social objects in our life can have a significant impact on our behavior towards these objects (Greenwald & Banaji, 1995). Social
psychological research has found that attitudes can be formed implicitly, outside of conscious cognition (Greenwald & Banaji, 1995). Implicit attitudes can be defined as “introspectively unidentified (or inaccurately identified) traces of past experience that mediate favorable or unfavorable feeling, thought or action toward social objects, such as people, places and policies” (Greenwald & Banaji, 1995, p.8).
Numerous research studies on the so-called “halo effect” have demonstrated the powerful impact of such attitudes. The halo effect was first coined by Edward Thorndike (1920) and describes how the existing attitudes people hold towards one object or
characteristic can have a significant influence on how they perceive other attributes
(Greenwald & Banaji, 1995). For example, Dion, Berscheid and Walster (1972) found that subjects rated attractive men and women as kinder, more interesting, more sociable, happier, stronger, of better character and more likely to hold prestigious jobs. Here, subjects’ attitudes towards someone’s level of attractiveness can be said to have operated implicitly, because the subjects did not notice how ratings of attractiveness influenced their judgment of other
unrelated attributes, such as kindness.
However, perceptions and judgments of other people are not only affected by attitudes, but can also be established through stereotypes. Stereotypes are socially shared
beliefs about certain traits that are typical for members of a social group (Grenwald & Banaji, 1995). Like implicit attitudes, implicit stereotypes guide our judgments and actions towards others, without our conscious awareness and control. These conceptions contrast with prejudice, which is used to describe people’s explicit negative attitudes toward out-groups (Fazio & Olson, 2003).
Implicit attitudes and stereotypes can be measured in a number of different ways. Some examples are the Affect Misattribution Procedure (AMP) (Payne, Cheng, Govorun & Stewart, 2005), the Extrinsic Affective Simon Task (EAST) (Houwer, 2003), and the Go/No go Association test (GNAT) (Nosek & Banaji, 2001). One of the most often used measures of implicit attitudes is the Implicit Association test (IAT), which assesses the strength between automatic associations. (Greenwald, McGhee & Schwartz, 1998). Scores on the IAT are based on the performance speeds of two classification tasks in which the strength of
association influences the performance of that task. The test has been used to uncover biases towards personal characteristics such as body weight (Schwartz, Vartanian, Nosek &
Brownell, 2006; Schwartz, et. al. 2003), gender (Levinson & Young, 2010), and race
(McConnell & Leibold, 2000). Furthermore, it has been able to uncover how certain implicit attitudes and stereotypes lead to biases that correlate with decisions made towards certain groups in a number of domains. Examples can be found in the criminal justice system (Rachlinski, Johnson, Wistrich & Guthrie, 2009), the employment process (Ziegert &
Hanges, 2005) and in the educational system (van den Bergh, Denessen, Hornstra, Voeten & Holland, 2010).
Health care providers’ implicit racial bias of patients
One particularly worrying domain in which implicit attitudes have been found to adversely affect individuals is in the healthcare system. Racial and ethnic disparities in
healthcare are well documented (Pletcher, Kertesz, Kohn & Goncales, 2008). These
disparities have been found across a number of health conditions, in diagnostic and treatment modalities (ex. African American patients less likely to receive pain relief medication than European American patients (Green et al., 2003)) as well as dimensions of technical quality (ex. African American patients receiving less patient talk time and shorter consultations than European American patients (Cooper et al., 2012)). Research on the treatment of different races in a healthcare setting introduces implicit attitudes as a possible explanation for this disparity (Blair et al., 2013; Green et al., 2003a; Green et al., 2003b; Sabin & Greenwald, 2012).
Green, et al.’s (2007) research on the explicit and implicit biases of doctors and how they influence their treatment decisions is a prime example of this. According to their research, doctors exhibited no explicit preference for European American versus African American patients and there was no difference in either races’ level of cooperativeness. However, implicitly, doctors favored European American patients over African American patients, and African American patients were associated with lower levels of cooperativeness generally and with medical procedures. Additionally, they found that as doctors’
pro-European American implicit bias increased, so did their probability of treating pro-European American patients and not African American patients efficiently (with thrombolysis). Other treatment decisions such as prescriptions for pain medication and attention deficit
hyperactivity disorder (ADHD), as well as referrals of asthma patients to specialists (pulmonary clinics), have also shown to be predicted by implicit bias measures (Sabin & Greenwald, 2012).
These perceptions of patients can also influence how doctors view their patients’ communication skills. Street, Gordon and Haidet (2007) examined audio recordings of twenty-nine doctors and 207 patients. In the study, doctors perceived African American
patients to be less effective communicators and less satisfied. However, an analysis of these patients’ actual communication revealed that it did not significantly differ from the
communication of other races. This is particularly troubling, as patients who were perceived as less effective communicators and less satisfied also received poorer physician
communication in terms of information giving and higher levels of contentiousness from their doctor.
Patients’ implicit racial bias of health care providers
Research results on implicit attitudes in healthcare show how a doctor’s implicit attitudes impact the way they perceive patient communication and most importantly, how they treat the patient. However, the patient’s perceptions of doctors’ communication are also pivotal in evaluating important health care outcomes, including general satisfaction and patient compliance (Smith, Polis & Hadac, 1981; Buller & Buller, 1987; Roter, Hall & Katz, 1987; Like & Zyzanski, 1987). Although patient perceptions of healthcare providers and healthcare providers perceptions of patient’s race have been found to influence the quality of healthcare, patients’ perceptions of the race of the healthcare provider has gone vastly overlooked.
Therefore, it would be interesting to investigate whether patients also carry these implicit biases and if this can influence how they perceive their healthcare provider’s communication. Considering how the patient’s implicit racial attitudes influence their evaluation of a healthcare worker’s communication can allow for an even better understanding of how communication functions between these two groups.
Patient satisfaction
The factors that lead to patients being content with healthcare quality and following the advice given by healthcare providers has been heavily researched. Several findings regarding this matter have identified patient satisfaction with the doctor and medical treatment serves as an important determinant of both of these goals (Smith, Polis & Hadac, 1981; Roter, Hall & Katz, 1987; Buller & Buller, 1987). Patient satisfaction is also crucial for doctors and hospitals as a negative association between satisfaction and the filing of
malpractice claims has also been found (Ha & Longnecker, 2010). Thus understanding how satisfaction is reached and what a doctor can do to help patients reach satisfaction is crucial to effective healthcare service.
Patient satisfaction can be defined as the fulfillment of expectations, needs or desires (Sitzia & Wood, 1997). This may vary from individual to individual as the feeling of
satisfaction comes from individual perceptions of service from healthcare workers (Crow et al., 2002). The patient’s perceptions of healthcare workers play a pivotal role in how satisfied they are with the care they receive. In fact, patients’ perceptions have been found to be stronger predictors of satisfaction than the actual care they received (Blachard, Labreque, Ruckdeschel & Blanchard 1990).
As previously mentioned, our perceptions of certain traits of individuals (ex. intelligence) can be influenced unconsciously by other unrelated personal traits (ex.
attractiveness.) As the race of a patient has been shown to unconsciously influence doctor’s ratings of their intelligence (Ryn & Burke, 2000) and communication skills (Street jr, Gordon & Haidet, 2007) during doctor-patient interactions, it would also be logical to assume that implicit attitudes towards the race of the doctor would inadvertently impact patients
evaluation of the interaction, including consultation satisfaction. Therefore, the following hypothesis was formulated:
H1: Patients with implicit racial biases against the doctor’s race will rate their consultation satisfaction lower than if patients’ implicit racial biases are not in favor of the doctor’s race.
Although implicit racial attitudes can inadvertently impact the consultation between patient and healthcare provider (Cooper et al., 2012, Penner et al., 2010, Green et al., 2007), they are clearly not the only contributors to a successful consultation in terms of patient satisfaction. In the consultation itself, the communication of the doctor is also a key
contributor to a patient’s perception of satisfaction (Kim, 2004). According to Donabedian’s (1980) classic subdivisions of healthcare satisfaction, one key subdivision to a patient’s satisfaction is based on the process of services received. Within this component, important communication elements, such as satisfaction with interpersonal communication and ratings of doctor’s responsiveness, are the main determinants. Research on certain communication traits that doctors exhibit show just how critical effective communication is in determining satisfaction (Kim, 2004; Buller and Buller, 1987; Zachariae et al., 2003). Components of effective interpersonal communication skills such as kindness, politeness, and sensitivity have been linked with higher ratings of healthcare satisfaction (Willson and McNamara, 1982; Bartlett, et al., 1984, Bertakis, Roter & Putnam, 1991). Similar results in patient satisfaction studies have also found that physician’s perceived competence also is influential (Needle, 1976; Greene, Weinberger & Mamlin, 1980; Gillette, Bryne & Cranston, 1982; Willson and McNamara, 1982;).
This study will examine how perceived interpersonal skills and physician competence mediate the relationship between implicit perceptions of a healthcare worker’s race and consultation satisfaction.
Perceptions of interpersonal skills
The interpersonal skills displayed by healthcare workers have been identified as one of the many ways quality healthcare has been defined by patients (Sofaer & Firminger, 2005). Some attributes associated with the concept of healthcare worker interpersonal skills include empathy, friendliness, kindness, and emotional support (Hall, Roter & Rand, 1981; Blanchard, Labrecque & Ruckdeschel & Blanchard, 1990). With that being said, it is intuitive that interpersonal skills have often been found to be associated with critical healthcare
outcomes such as patient satisfaction and compliance.
The impact of healthcare workers’ interpersonal skills have repeatedly been found to influence interactions between healthcare workers and patients in a positive manner. A possible explanation for this is how perceptions of interpersonal skills encourage closer and more sociable relationships between healthcare workers and patients (Mathews, 1983, Malat, 2001). Demonstrations of courtesy, kindness, concern and sensitivity allow patients to feel cared for and about which develops a more comfortable bond (Sofaer & Firminger, 2005). This increase in comfort allows for higher levels of trust in caregivers and a reduction in the feelings of vulnerability and anxiety (Sofaer & Firminger, 2005).
The consequences of courteous communication previously mentioned, make the connection between perceived interpersonal skills and patient satisfaction understandable. In a meta-analysis of physician communication and its impact on satisfaction found that
dimensions of interpersonal skills (ex. courtesy and emotional support) were significantly associated with not just more positive consultation satisfaction, but overall healthcare satisfaction (Sofaer & Firminger, 2005).
Similarly to patient satisfaction, a patient’s implicit attitudes can influence perceptions of a healthcare worker’s interpersonal skills. Implicit racial bias has been shown to predict
judgment of character that is not in accordance with behavior. In a study observing the evaluation of facial expressions it was found that images of African Americans were rated as more angry based on facial expressions that were the same for European Americans
(Hughenberg & Bodenhausen, 2003). Additionally, as mentioned previously, it has already been shown that during an interaction between a physician and patient, the physician’s implicit racial bias predicts their evaluation of the patient’s communication skills. If implicit racial bias can allow for stronger perceptions of anger and lower communication skills, it could also lead to differences in patient’s perseptions of a healthcare worker’s interpersonal skills based on their race. Therefore, the following hypothesis has been formulated:
H2: Perceived interpersonal skills will mediate the relationship between implicit racial bias and consultation satisfaction, in that implicit racial bias against the race of the healthcare worker will result in lower perceived interpersonal skills, which will lower their consultation satisfaction.
Perceptions of physician competence
Although significant resources have been used to determine the impact of physician’s competence, researchers are still in disagreement on how patients perceive the competence level of their physicians (Ware et. al, 1975; Roghmann, Hengst & Zastrowny, 1979; Willson & McNamara, 1982). Certain research would indicate that there are two dimensions to physician competence: technical competence and interpersonal competence. Technical
competence can be defined as the thoroughness a physician exhibits during a consultation and the ability to provide appropriate and effective treatment (Thom & Campbell, 1997).
Interpersonal competence is described as the quality of psychosocial care, interpersonal skills and interviewing skills.
According to Slack (1977), technical competence is what matters during the
consultation, and that measuring perceived competence is irrelevant because patients do not poses the knowledge to accurately evaluate that component. When patients’ are asked about the physician’s competence, they then actually use the doctor’s interpersonal skills to
determine how technically competent the doctor was. Additionally, the technical component of competence has been shown to lack variation as receiving a degree in medicine serves as an indication of the physician’s technical competence (Buller & Buller, 1987).
However, there is empirical evidence that counters this approach. Wilson and McNamara (1982) conducted an experiment to determine if patients were able to accurately evaluate a doctor’s interpersonal skills and competence separately. In the experiment they manipulated a simulated doctor-patient interaction where respondents viewed one of four conditions: doctor with high courtesy and low competence, doctor with low courtesy and low competence, a doctor with high courtesy and high competence, and a doctor with low
courtesy and high competence. They found that patients were able to separate their evaluations of competence from their evaluation from courtesy. Additionally, this study found that the construct competence significantly influenced consultation satisfaction.
In a meta-analysis conducted by Hall, Roter and Katz (1988) both technical and interpersonal competence significantly predicted patient satisfaction. Justification for this relationship is that competence improves patient trust in them as well as reduces anxiety, which makes the consultation less stressful (Hall, Roter and Katz, 1988).
Perceptions of competence however are not solely decided on the communication between patient and doctor. The race of the communicator also has been found to alter perceptions of competence, particularly for African Americans. In a study conducted by Rye and Burke (2000), doctor’s perceived African American patients as less component, although
actual observations of the encounters did not come to the same conclusion. A potential reason for this could be the implicit racial biases that doctors possess, as this has already been shown to predict other perceptions of patients, such being less cooperative (Green, et. al., 2007). Therefore, it is within reason to expect that patient’s implicit racial biases can also alter perceptions of the doctor, including their competence, which can affect a patient’s consultation satisfaction. Accordingly, the following hypothesis was postulated:
H3: Perceived competence will mediate the relationship between implicit racial biases and consultation satisfaction, in that patients whose racial biases are against the healthcare worker’s race will rate perceived competency lower, leading to lower levels of consultation satisfaction, compared to patients whose implicit racial biases are in favor of the doctor’s race.
The conceptual model that will be examined can be found in figure 1. Figure 1: Conceptual Model
Methods
In order to select the appropriate stimulus material for the main study a pretest was conducted. The main goal of the pretest was to select images of two doctors, which did not differ in terms of any perceivable characteristics other than their race.
Pretest
Design and Participants. This pretest had a two-condition mixed between (image sequence) and within (images of doctors) subject design. Participants for the pretest were recruited through Amazon’s survey panel, Mechanical Turk (Mturk). As this platform was used for the actual experiment as well, it was imperative that the pretest sample resembled the actual sample used for the experiment. Respondents living in the United States were selected as the target group for this pretest. Respondents were compensated $.70 for their participation. A total of 32 respondents participated in the pretest (39% females, Mage = 37.15, SD = 12.54).
Stimuli. Each condition consisted of six images, three European American doctors and three African American doctors. All images were found through Google image searches. The requirements for the images were that they had to be a profile shot where only the doctor was present and had to be wearing a lab coat in order to increase the believability of the image. The two conditions only differed in the order in which the images were shown. In the first condition, the European American doctors were shown first, followed by the African American doctors. In the second condition, the African American doctors were shown first, followed by the European American doctors. This was done to make sure that the
presentation order of the images did not influence the overall rating of each doctor. Images of the doctors can be seen in Appendix 1.
Procedure. Once participants agreed to the terms and conditions, they were told that the purpose of the survey is simply to rate six images of doctors. After each image,
participants rated the characteristics of the imaged doctor. Participants rated the doctor’s perceived competence, empathy, friendliness, happiness, sadness, attractiveness, seriousness and age. These items were selected at face value to measure characteristics about doctor that could inadvertently influence the respondent.
To insure the validity of this self-constructed scale called “Doctor Image evaluation”, an exploratory principle component factor analysis was conducted. A total of nine items were added to this analysis with orthogonal rotation. A Kaiser-Meyer-Olkin was used to make sure the sample was adequate and returned a KMO score of .81, which is well above the minimum criteria of .5 (Field, 2009). All KMO values for individual items were above .58. Results from Bartlett’s Test of Sphericity indicate that all correlations between items are large enough for a PCA (χ2(36) = 932.08, p < .001). Eigenvalues for each component were obtained and produced three components with an eigenvalue above 1 and explained 76.09% of the variance. Based on the results of the factor analysis, three components were kept to form the Doctor image evaluation scale. Table 1 shows how each item loaded on one of three components. After examining the items loaded on each component, it was deemed that the first component measures the professionalism of the doctor (Professional), the second
component measures the emotional rating of the doctor (Emotional), and the third component measures the attractiveness of the doctor (Attractiveness.
Two of the three scales were tested for their reliability and were reliable: the
professional component (Cronbach’s α=.83); the emotional component (Cronbach’s α= .81), both of which are very good. The attractiveness component was not tested for reliability as it
only consisted of one item. Rating on all three scales were calculated for each image. The ratings of each doctor can be seen in Table 2.
Results from the mixed ANOVA, with doctor image ratings as the within factor and image presentation order as the between factor, revealed no main effect on the ratings on the professional dimension (F(5, 155)= .74, p = .274), emotional dimension (F(5, 155)= .90, p = .484), attractiveness dimension (F(5, 155)= .12, p = .274) and age (F(5, 155)= .36, p = .172). The within-subjects factor, images of doctors, was found to be significant for the professional dimension (F(5, 155) = 6.09, p < .001), emotional dimension (F(5, 155) = 4.22, p = .001), attractiveness dimension (F(5, 155) = 7.04, p < .001) and age (F(5, 155) = 12.37, p < .001) as expected. Pairwise comparisons were used to determine which doctors were not significantly different on the three dimensions (professional, emotional, attractiveness) and age. After careful examination of the doctors it was found that doctor 3 did not significantly differ from doctor 6 on the professional dimension (Mdifference = 1.45, p = 1.00, 95% CI [-2.076, 4.98]), emotional dimension (Mdifference = -.84, p = 1.00, 95% CI [-3.33, 1.66]), attractiveness
dimension (Mdifference = .63, p = 1.00, 95% CI [-.645, 1.91]) and age (Mdifference = -.05, p =
1.000, 95% CI [-.57, .66]). Therefore these two images were used for the manipulation of this study.
The study
Design and Participants. To test the purposed model an experiment with a two condition between subject design was conducted. Respondents for the main experiment in this study were collected from the survey panel Mechanical turk (Mturk). Respondents were told that they would take part in a two-part survey that was unrelated to each other. The first part was a categorization task of words and faces (Implicit association test) and the second part was to rate a consultation with a doctor. Respondents were recruited through mturk as
the entire survey required a significant amount of time, and it was deemed that respondents deserve compensation for their participation in this experiment, which this platform allowed. Mturk was also chosen as it allowed for the researcher to collect a sample of people living in the United States, while living abroad. Respondents were paid $3.15 for their participation. As we were interested in a diverse sample, the only requirement for participation was that the respondent had to be above the age of 18. In total, 221 respondents participated. 29
respondents were excluded for having not completed the IAT. An additional eleven respondents were excluded for completing only the IAT without filling out any further information. Therefore, the final sample consisted of 181 respondents (Mage = 34.88, SD = 9.73, 49.4% female).
Stimulus Material. The stimulus material mainly consisted of a peer-reviewed narrative that allows for the respondent to imagine that he or she is a patient of a doctor they have never met. The narrative begins by describing how the respondent woke up feeling a bit ill. As the day progresses, while the participant is at work, the symptoms become stronger. After the work day is complete the respondent imagines going home and progressively feeling worse, to the point where they decide to cancel work for the next day and to see a doctor. When they arrive at the doctor’s office respondents are told to imagine that there is a substitute doctor there. This is where the manipulation for this experiment takes place. Each respondent will be shown a picture of the substitute doctor, which will be either the African American or European American doctor from the pretest. After examining the picture, participants will read a conversation between them and the pictured doctor, including a formal greeting, several questions and ends with a diagnosis of the illness. The narrative can be found in Appendix 2.
Procedure. Upon entering the online survey, respondents were told that they would be participating in two unrelated studies. The first study was described as a categorization
task, but was in fact an IAT, which assessed respondent’s implicit racial biases. The second study was described as a hypothetical situation in which respondents were to imagine being sick and having to go to the doctor. This served as the stimulus for this experiment.
Respondents were told that this was a two-part study to conceal the true purpose of the experiment.
Implicit Association Test. An IAT was used to measure the level of implicit racial
bias each respondent possesses. Respondents began the task by reading instructions on how to complete the “categorization task”. The stimuli used for this task included eight schematic faces (two black and white men and black and white women) and positive (ex. joy) and negative (ex. evil) words. Pictures used for this part of the study were created and validated through the non-profit organization “Project implicit”, whose main goal is to make research on the implications of implicit attitudes more accessible (Projectimplicit.net, 2011). The IAT was split into five blocks where images, words and images and words together were
categorized. The task assigned to each block can be seen in Table 3 in Appendix 3. Each block contained a different categorization task, where either respondents had to categorize faces by race (African American or European American), words by positive or negative meaning, or by both faces and words (ex. African American/positive or European
American/negative). Respondents were instructed to categorize these images and words as quickly as possible. The reason they were told this was to make sure that their more
unconscious associations become present.
Once each respondent completed the IAT, they were instructed to begin the second study. Respondents began the second study by filling out a list of three questions about their experience with the IAT and six questions about their own personal characteristics. This was done in order to create a separation between the IAT and the second part of this study, so that respondents didn’t link the two surveys together and figure out the purpose of the survey.
Respondents then were instructed to read a hypothetical situation, where they gradually experienced the symptoms of the flu that increasingly got worse, and due to these symptoms they decide to go to the doctor. Respondents were also advised to play close attention to the images in the story because they help with the imagination process. In the narrative a total of four images were used, one of which was the actual manipulation (image of the doctor). Afterwards, respondents answered a number of questions used to measure the dependent variables for this particular study. Dependent variables were measured using implicit and explicit methods. Tasks and questions from the entire survey can be found in Appendix 2.
Measures
Implicit racial bias
Respondent’s IAT scores were calculated through a multistep process, in accordance to Greenwald’s criteria (Greenwald, Nosak & Banaji, 2003). First, all response latencies, or the amount of time it took a respondent to make an association, were dropped out of analysis except for the two test blocks (Block 3 and Block 5). The second step was to drop the first two associations from the test blocks as they are seen as practice for the association task. The third step was to recode any extreme latency times. An extreme response latency was
considered to be longer than 3000 milliseconds or less than 30 milliseconds. Scores that were higher than 3000 milliseconds were recoded to 3000 milliseconds, and scores below 30 milliseconds were recoded to 30 milliseconds (Greenwald, Nosak & Banaji, 2003). Following this step, all response latency scores were transformed to natural logs. The reasoning behind using the log-transformed latency times is to curb the usually extended upper tail of latency distributions (Greenwald, Nosak & Banaji, 2003).
Log-transformed latency scores were then averaged to give each respondent two means, one for associations between good words/European Americans - bad words/ African
Americans (Block 3) and good words/African Americans – bad words/European Americans (Block 5). The last step was to create a differential mean score where block three was
subtracted by block five. Therefore if associations took longer in block three than block five, respondents would have a positive score and if associations in block five took longer than in block three they would have a negative score. Positive scores mean that the respondent has an implicit racial bias in favor of African Americans and negative scores mean that respondents have an implicit racial bias in favor of European Americans (M = -.0788, SD = .17). Once scores were calculated, a median split was used to separate the groups into two equal groups (Median = -.0785). The median was chosen instead of the actual neutral point of zero, due to the fact that most of the respondents were racially biased in favor of European Americans and the use of this point would result in a significant difference in the amount of respondents in each group.
Explicit Interpersonal Skills & Competence. Perceived interpersonal skills and competence were both measured explicitly and implicitly. These constructs were
operationalized explicitly based on a previous study conducted by Willson and McNamara (1982). To measure interpersonal skills, nine items were used (friendly, courteous, polite, kind, pleasant, likeable, considerate, sensitive and sympathetic) (M = 8.61, SD = 1.14, Cronbach’s α = .96), and seven items were used (skillful, experienced, efficient, accurate, competent, educated and thorough) to measure competence (M = 8.83, SD = 1.08,
Cronbach’s α = .96).
Implicit Interpersonal Skills & Competence. As previous research would suggest, pre-meditated decision-making, may not truly represent the opinion one has about an object (Dovidio, Kawakami & Gaertner, 2002, Hagiwara et al., 2013, Perry, Murphy & Dovidio, 2015). Therefore, these constructs were also measured implicitly. To operationalize these constructs, respondents were shown a word on their screen and were asked whether they felt
that the doctor in the narrative could be described by that word. The words used for this test were taken from the words used for the explicit scales. Additionally, an antonym of each word in the scale was included. Respondents were required to answer as quickly as possible. Respondents could answer whether they thought the doctor could be described by the
adjective by typing the “e” key and if they thought the adjective did not describe the doctor they answered by typing the “i” key. In total, 32 words were used for these measures. Scores for both constructs were calculated by adding a point each time a respondent described the doctor positively, and subtracted a point each time the doctor was described negatively (Minterpersonal skills = 8.31, SDinterpersonal skills = 1.23; Mcompetence = 6.31, SDcompetence = 1.25). A screenshot of this test can be seen in Appendix 2, which contains this survey.
Patient Satisfaction. To measure patient satisfaction, the Consultation Satisfaction Questionnaire was used (Poulton, 1996). This scale was created and validated specifically to measure a patient’s satisfaction with the consultation directly and not their overall
satisfaction with healthcare providers. This scale measures three components of consultation satisfaction: professional care, depth of relationship and perceived time given. The
components depth of relationship and perceived time given were not used in this analysis. Depth of relationship was not used as the narrative included a substitute doctor. Therefore no depth of relationship could possibly be established. Perceived time given was not used because no indication of time was included in the narrative. Additionally, the focus of this study is on the communication between patient and doctor, and the professional care component adequately measured this construct. Therefore, questions only from the professional care component were used. This construct was operationalized through five-point likert scale items, in which respondents gave their level of agreement with five statements (M = 4.29, SD = .49, α = .83). An example item from this scale is “Dr. Williams was very careful to check everything during the examination.”
Control variables. To control for extraneous influence outside of the actual
manipulation, data on potential confounding variables were also collected. Respondents were asked about how believable the narrative was, how often they have visited their doctor, how often they have similar symptoms, their opinion on antiviral drugs, and their general
satisfaction with healthcare. Additionally, several questions concerning the respondents’ characteristics or demographics were also collected. Traits include, respondent’s age, gender, education, race, native language, and birth country. Descriptive statistics of demographic characteristics of the sample can be found in Table 4.
Results
A manipulation check was used to assess if respondents read and understood the narrative. In order to check for this, eleven true or false questions about the symptoms of the illness and advice given by the doctor were asked, along with one multiple-choice question about the recovery time from the illness. For each correct response, respondents got a point and could receive a maximum of 12 points (M = 11.29, SD = .93). Respondents who could not answer seven out of twelve questions correctly were excluded from the analysis (n = 3). The reason for this cut off is that most of the questions have only two answer options and if answers were chosen at random without reading the text, a respondent has a chance of answering six answers correctly. This resulted in a final sample of N = 178.
Confound checks were conducted to make sure that extraneous influences were not included in the analysis. This was done through a two-step process. The first step was to make sure that all demographic characteristics and values from control variables were evenly distributed between the two conditions. If any of the variables were not evenly distributed between the two conditions, these variables were tested for significant correlations with the
final outcome variable (consultation satisfaction) and the two mediators (Perceived competence and perceived interpersonal skills).
Results from the first step indicated that all of the demographic and control variable scores were evenly distributed between groups. Values can be seen in Table 5. Due to the fact that none of the potential extraneous influencers passed through the first step in the confound check process, we continued directly to analysis without adding any covariates.
Implicit racial bias and Doctor’s race.
A multivariate analysis of variance was conducted in order to analyze the main effect of the interaction between independent variable (African American vs. European American doctor) and the moderator (implicit racial bias: pro European American bias vs. pro African American bias) on the dependent variable (Consultation satisfaction). Descriptive statistics and results of the ANOVA can be seen in Tables 6 and 7 respectively.
The first hypothesis (H1) postulated that those with an implicit racial bias against the race of the doctor in the narrative will have a significantly lower rating of consultation
satisfaction compared to when implicit racial bias is in favor of the race of the doctor. Results from the ANOVA found that there is a significant interaction between the implicit racial biases of respondents with their exposure to doctors of different races in terms of consultation satisfaction, (F(1, 174) = 5.06, p = .026, η2 = .03).
European American Condition. Those with an implicit racial bias towards European Americans (M = 4.26, SD = .47) rated their satisfaction with the consultation lower than those who had an implicit racial bias towards African Americans (M = 4.29, SD = .43) A separate t-test analysis found that the difference in consultation satisfaction (.03) between those who
have implicit racial bias toward African Americans and those who have implicit racial bias towards European Americans is not statistically significant (t (96) = -.34, p = .738).
African American Condition. Those who had an implicit racial bias in favor of African Americans (M = 4.45, SD = .45) rated their satisfaction with the consultation higher than those with an implicit racial bias in favor of European Americans (M = 4.15, SD = .59). A separate t-test analysis found that this difference (.16) is indeed statistically significant (t (80) = -2.58, p = .012). Therefore, H1 can only be partially confirmed as there is a significant interaction between respondent implicit racial bias and race of the doctor on consultation satisfaction, but only a significant difference in mean was found in the condition with an African American doctor. A line graph of consultation satisfaction comparing the two implicit racial bias groups by condition can be seen in Figure 2.
Figure 2: Line graph of Consultation satisfaction by Condition
Note: AA = African American (n = 82); EA = European American (n = 96)
4 4.05 4.1 4.15 4.2 4.25 4.3 4.35 4.4 4.45 4.5 AA condition EA condition AA bias EA Bias
Mediation of the Moderated Relationship
A regression model testing the mediation of the main effect of implicit racial bias on consultation satisfaction by both implicit and explicit perceived interpersonal skills as well as competence were run using the Preacher and Hayes’s Process regression tool in SPSS (model 4). For this analysis, each doctor condition was analyzed separately because the interaction between doctor and implicit racial bias was significant. This was done in order to see if the model holds in both conditions. Output from the process analysis would suggest that the overall model significantly predicts consultation satisfaction for the African American doctor (R2= .076, F(1,80) = 6.32, p = .014), but not for the European American doctor (R2 = .001, F(1, 94) = .111, p = .740).
The Mediating Effect of Perceived Interpersonal Skills. H2 states that perceived interpersonal skills mediate the relationship between implicit racial bias and consultation satisfaction in that bias against the doctor will result in lower perceived interpersonal skills, which will lower a patient’s consultation satisfaction.
European American condition. Results from the regression analysis indicate that no
significant mediation was found for the implicit racial bias and consultation satisfaction relationship (b = .001, BCa CI [-.017, .035]). Additionally, for the explicit measure of
perceived interpersonal skills there was no significant mediation found (b = .001, BCa CI = -.57, .048). Therefore for this condition, the mediation of implicit and explicit perceived interpersonal skills cannot be confirmed.
African American condition. Results from the regression analysis indicate that a
significant mediation was found for the implicit measure of implicit perceived interpersonal skills (b = .112, BCa CI [.013, .329]). As for explicit interpersonal skills, no mediation effect was found (b = .056, BCa CI [-.006, .245]). Therefore for this condition, the mediation of the
relationship between implicit racial bias and consultation satisfaction was moderated by implicit perceived interpersonal skills, but not explicit perceived interpersonal skills. Accordingly, H2 can only be partially supported.
The Mediating Effect of Perceived Competence. H3 states that perceived competence mediates the relationship between implicit racial bias and consultation satisfaction, in that bias against the race of the doctor will result in lower ratings of competence, which in turn will lower consultation satisfaction.
European American Condition. Results from the regression analysis indicate that
perceived competence does not mediate the relationship between implicit racial bias and consultation satisfaction for both implicit perceived competence (b = .001, BCa CI[-.010, .050]) and explicit perceived competence (b = -.001, BCa CI[-.101, .092]). Therefore in this condition, the notion that perceived competence mediates the relationship between implicit racial bias and consultation satisfaction on both the implicit and explicit level was not supported.
African American condition. Results from the regression analysis indicate that
implicit perceived competence significantly mediates the relationship between implicit racial bias and consultation satisfaction (b = -.088, BCa CI[-.297, -.003]). However, the regression coefficient from implicit racial bias to perceived competence was not significant (b = .48, t(94) = 1.53, p = .131). This mix in results may have resulted in the bootstrapping procedure used in process. This could have been caused by the lack of variance in the condition. Therefore, the mediation of implicit competence cannot be confirmed. As for the mediation of explicit perceived competence on the relationship between implicit racial bias and
consultation satisfaction, no significant mediation was found (b = .068, BCa CI[-.014, .249]). Therefore, H3 cannot be supported, as no mediation of implicit and explicit perceived
competence on the relationship between implicit racial bias and consultation satisfaction was found in either condition.
Discussion
The main aim of this study was to observe the relationship between doctor’s race and perceptions of consultation satisfaction moderated by the patient’s implicit racial bias. Secondly, this study examined the role of two mediating variables (interpersonal skills and competence) on this relationship. The results from this study are a step in the right direction to further understanding the implications of these biases in healthcare settings.
The findings from this study display interesting insight into the influence of our implicit racial biases and their interaction with communication in a healthcare setting. The main interaction between the race of the doctor and the implicit racial bias of the patient on how patients perceive their satisfaction from the consultation was only present when the doctor was African American and not when the doctor was European American. The partial support of this finding depicts a reality that is hard to ignore. The fact that implicit racial biases were found to have an impact on perceptions of the consultation with the African American doctor only, show that our implicit racial biases may disproportionately impact African Americans compared to European Americans. Although the images of the doctors were pre-tested to be as similar as possible, they resulted in altering the way the interaction was evaluated, potentially because of how these images were processed implicitly. A
possible explanation to this is that the image of a doctor for most Americans would be one of a European American male, and this incongruence of their expectation with what is being presented could have strengthened the more critical aspects of their judgment.
The findings from the mediation of perceived interpersonal skills were also partially supported in that mediation was only found in the African American condition and only
implicit perceived interpersonal skills was found to be a significant mediator. This also supports the previous explanation, as well as add backs the use of implicit measures. Firstly, the fact that mediation was found only in the African American condition and not in the European American condition shows that again the evaluation of communication can be perceived differently depending on the race of the person communicating, even in a situation where that person is of a high authority, such as a doctor. The fact that mediation of
perceived interpersonal skills was found only for the implicit measure adds an additional layer to these findings. This adds support to other research (Dovido, Kawakami & Gaertner, 2002; Penner, et. al. 2010) that show the predictive validity and at times superiority of implicit measures when compared to explicit measures. It also shows that even though these two types of measures can observe the same construct, they can obtain different aspects of that construct.
The results concerning the mediation of competence on the relationship between implicit racial bias and consultation satisfaction could not support the postulated hypothesis (H3) for both conditions (African American & European American) and for both types of measures (Implicit & Explicit). A possible explanation for this lack of effect could be due to the stereotypes associated with doctors. As previously mentioned, the degree obtained by a doctor can also act as a validation of competence, regardless of their communication or action during the consultation. Being that this was a narrative that depicted a standard consultation, the fact that respondents were assessing the competence of someone who may be more educated than they are could have reduced the impact of this construct. This was also shown through the lack of variance and that the mean was so close to the maximum possible score for both measures.
Limitations and Suggestions for further research
Although this study has provided valuable insight into how race interactions with health communication implicitly, there were a few limitations to this research project. The first limitation is that typically, the use of implicit racial bias measures in healthcare research tends to use real behavior and not simulated behavior. The lack of ecological validity in this study creates issues with how generalizable these results actually are. Additionally, non-verbal communication was omitted as it was too difficult to integrate in a narrative format. Non-verbal behavior has been used to great effect in measuring how implicit racial attitudes predict interaction behavior between European and African Americans (Dovido, Kawakami & Gaertner, 2002) and it would have improved the findings of this research if this could have been integrated. A suggestion for future research would be to take real patients, measure their satisfaction with their general practitioner and at a later point measure their implicit racial bias. Additionally, the influence of verbal and non-verbal behavior could be integrated to create a more realistic picture of communication between patient and physician or healthcare worker.
Secondly, the construct consultation satisfaction was measured only explicitly. At this current time, there isn’t a measure that has been validated that measures patient satisfaction implicitly. Using an implicit measure to predict an explicit outcome measure has been shown to be unsuccessful in other studies (Dovido, Kawakami & Gaertner, 2002). A suggestion for future research would be the conceptualization and validation of implicit measures that can evaluate different health communication constructs. This could lead to the discovery and improvement of how we operationalize concepts that could more accurately depict patient’s healthcare state.
Lastly, the narrative only considered the difference between how communication is perceived from a European and African American doctor. In the healthcare industry there are several different personal characteristics that could be considered. For example, an
investigation on how gender interacts with race in terms of perceived communication for a healthcare worker, or how implicit associations interact with perceived communication of doctors from other nationalities. The addition of other characteristics can build on these current findings and provide a clearer picture of how our implicit biases interact with perceived communication in different situations.
Conclusion
The inequalities that have been found in healthcare concerning race must be addressed. In order to do this, it is vital to study the different components of healthcare consultations that can result in these inequalities, including communication. As
communication between patient and healthcare worker is a critical component of quality healthcare, research that seeks to understand how these inequalities can come about can allow for more appropriate action to take place, to deal with this situation. Although this study did not address how this issue can be resolved, it did provide more information as to how the communication between patients and physicians is perceived by patients. A practical implication for this finding would be for doctors and healthcare workers to consider this when speaking to individuals of different races. To be conscious of the fact that patient’s implicit attitudes of race may play a role in how their communication is being processed will allow for doctors to adjust their communication accordingly, countering this influence.
References
Bartlett, E. E., Grayson, M., Barker, R., Levine, D. M., Golden, A., & Libber, S. (1984). The effects of physician communications skills on patient satisfaction; recall, and
adherence. Journal of chronic diseases, 37(9), 755-764. doi:10.1016/0021-9681(84)90044-4
Bertakis, K. D., Roter, D., & Putnam, S. M. (1991). The relationship of physician medical interview style to patient satisfaction. Journal of Family Practice, 32(2), 175-182. Blair, I., Steiner, J., Fairclough, D., Hanratty, R., Price, D., & Hirsh, H. et al. (2013).
Clinicians' Implicit Ethnic/Racial Bias and Perceptions of Care Among Black and Latino Patients. The Annals Of Family Medicine, 11(1), 43-52. doi:
10.1370/afm.1442.Division
Blanchard, C., Labrecque, M., Ruckdeschel, J., & Blanchard, E. (1990). Physician
behaviors, patient perceptions, and patient characteristics as predictors of satisfaction of hospitalized adult cancer patients. Cancer, 65(1), 186-192. doi: 10.1002/1097-0142(19900101)65:1<186::aid-cncr2820650136>3.0.co;2-4
Boon, H., & Stewart, M. (1998). Patient-physician communication assessment instruments: Patient Education And Counseling, 35(3), 161-176. doi:
10.1016/s0738-3991(98)00063-9
Buller, M., & Buller, D. (1987). Physicians' Communication Style and Patient Satisfaction. Journal Of Health And Social Behavior, 28(4), 375. doi: 10.2307/2136791
Burgess, D., Crowley-Matoka, M., Phelan, S., Dovidio, J., Kerns, R., & Roth, C. et al. (2008). Patient race and physicians' decisions to prescribe opioids for chronic low
back pain. Social Science & Medicine, 67(11), 1852-1860. doi:10.1016/j.socscimed.2008.09.009
Cooper-Patrick, L. (1999). Race, Gender, and Partnership in the Patient-Physician Relationship. JAMA, 282(6), 583. doi: 10.1001/jama.282.6.583
De Houwer, J. (2003). The Extrinsic Affective Simon Task. Experimental Psychology, 50(2), 77-85. doi: 10.1026//1618-3169.50.2.77
Dion, K., Berscheid, E., & Walster, E. (1972). What is beautiful is good. Journal Of Personality And Social Psychology, 24(3), 285-290. doi: 10.1037/h0033731 Dovidio, J., Kawakami, K., & Gaertner, S. (2002). Implicit and explicit prejudice and
interracial interaction. Journal Of Personality And Social Psychology, 82(1), 62-68. doi: 10.1037/0022-3514.82.1.62
Fazio, R., & Olson, M. (2003). Implicit Measures in Social Cognition Research: Their Meaning and Use. Annual Review Of Psychology, 54(1), 297-327. doi:
10.1146/annurev.psych.54.101601.145225
Gillette, J. L., Byrne, T. J., & Cranston, J. W. (1982). Variables affecting patient satisfaction with health care services in the college health setting. Journal of American College Health Association, 30(4), 167-170. doi: 10.1080/01644300.1981.10393067
Green, A., Carney, D., Pallin, D., Ngo, L., Raymond, K., Iezzoni, L., & Banaji, M. (2007). Implicit Bias among Physicians and its Prediction of Thrombolysis Decisions for Black and White Patients. J GEN INTERN MED, 22(9), 1231-1238. doi:
Green, C., Anderson, K., Baker, T., Campbell, L., Decker, S., & Fillingim, R. et al. (2003). The Unequal Burden of Pain: Confronting Racial and Ethnic Disparities in Pain. Pain Med, 4(3), 277-294. doi: 10.1046/j.1526-4637.2003.03034.x
Greene, J., Weinberger, M., & Mamlin, J. (1980). Patient attitudes toward health care: Expectations of primary care in a clinic setting. Social Science & Medicine. Part A: Medical Psychology & Medical Sociology, 14(2), 133-138. doi: 10.1016/s0271-7123(80)80032-0
Greenwald, A., & Banaji, M. (1995). Implicit social cognition: Attitudes, self-esteem, and stereotypes. Psychological Review, 102(1), 4-27. doi: 10.1037/0033-295x.102.1.4 Greenwald, A., McGhee, D., & Schwartz, J. (1998). Measuring individual differences in implicit cognition: The implicit association test. Journal Of Personality And Social Psychology, 74(6), 1464-1480. doi: 10.1037/0022-3514.74.6.1464
Greenwald, A., Nosek, B., & Banaji, M. (2003). Understanding and using the Implicit Association Test: I. An improved scoring algorithm. Journal Of Personality And Social Psychology, 85(2), 197-216. doi: 10.1037/0022-3514.85.2.197
Ha, J. F., & Longnecker, N. (2010). Doctor-patient communication: a review. The Ochsner Journal, 10(1), 38-43.
Hagiwara, N., Penner, L., Gonzalez, R., Eggly, S., Dovidio, J., & Gaertner, S. et al. (2013). Racial attitudes, physician–patient talk time ratio, and adherence in racially discordant medical interactions. Social Science & Medicine, 87, 123-131. doi:
Hall, J., Epstein, A., DeCiantis, M., & McNeil, B. (1993). Physicians' liking for their patients: More evidence for the role of affect in medical care. Health Psychology, 12(2), 140-146. doi: 10.1037/0278-6133.12.2.140
Hall, J., Roter, D., & Katz, N. (1987). Task Versus Socioemotional Behaviors in Physicians. Medical Care, 25(5), 399-412. doi: 10.1097/00005650-198705000-00004
Hall, J., Roter, D., & Katz, N. (1988). Meta-analysis of Correlates of Provider Behavior in Medical Encounters. Medical Care, 26(7), 657-675. doi:
10.1097/00005650-198807000-00002
Hall, J., Roter, D., & Rand, C. (1981). Communication of Affect between Patient and Physician. Journal Of Health And Social Behavior, 22(1), 18. doi: 10.2307/2136365
Hugenberg, K., & Bodenhausen, G. V. (2003). Facing prejudice implicit prejudice and the perception of facial threat. Psychological Science, 14(6), 640-643. doi:
10.1046/j.0956-7976.2003.psci_1478.x
Joint Commission of Accreditation of Hospitals and Healthcare organizations (JCAHO). Sentinel Event Data: Root causes by event type 2004-2014
(http://www.jointcommission.org/assets/1/18/root_causes_by_event_type_200 4-2014.pdf)
Kaplan, S., Gandek, B., Greenfield, S., Rogers, W., & Ware, J. (1995). Patient and Visit Characteristics Related to Physicians: Participatory Decision-Making Style. Medical Care, 33(12), 1176-1187. doi: 10.1097/00005650-199512000-00002
Kaplan, S., Greenfield, S., & Ware, J. (1989). Assessing the Effects of Physician-Patient Interactions on the Outcomes of Chronic Disease. Medical Care, 27(Supplement), S110-S127. doi: 10.1097/00005650-198903001-00010
Leckie, J. (1998). Satisfaction with communication in patient/physician encounters. The patient perspective. Patient Education And Counseling, 34, S23-S24. doi:
10.1016/s0738-3991(98)90049-0
Levinson, J. D., & Young, D. (2010). Implicit Gender Bias in the Legal Profession: An Empirical Study. Duke Journal of Gender Law & Policy, 18(1).
Leydon, G. M., Boulton, M., Moynihan, C., Jones, A., Mossman, J., Boudioni, M., & McPherson, K.. (2000). Cancer Patients' Information Needs and Information Seeking Behaviour: In Depth Interview Study. British Medical Journal, 320(7239), 909–913 Like, R., & Zyzanski, S. J. (1987). Patient satisfaction with the clinical encounter: social
psychological determinants. Social science & medicine, 24(4), 351-357. doi: 10.1016/0277-9536(87)90153-5
Malat, J. (2001). Social distance and patients' rating of healthcare providers. Journal of Health and Social Behavior, 360-372. doi: 10.2307/3090184
Mathews, J. (1983). The communication process in clinical settings. Social Science & Medicine, 17(18), 1371-1378. doi: 10.1016/0277-9536(83)90197-1
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(5), 435-442. doi: 10.1006/jesp.2000.1470
Mead, N., Bower, P., & Hann, M. (2002). The impact of general practitioners’ patient-centredness on patients’ post-consultation satisfaction and enablement. Social Science & Medicine, 55(2), 283-299. doi: 10.1016/s0277-9536(01)00171-x
Needle, R. H. (1976). Patterns of Utilization of Health Services by College Women. Journal of the American College Health Association, 24(5), 307-12.
Nosek, B., & Banaji, M. (2001). The Go/No-Go Association Task. Social Cognition, 19(6), 625-666. doi:10.1521/soco.19.6.625.20886
Payne, B., Cheng, C., Govorun, O., & Stewart, B. (2005). An inkblot for attitudes: Affect misattribution as implicit measurement. Journal Of Personality And Social
Psychology, 89(3), 277-293. doi: 10.1037/0022-3514.89.3.277
Pearson, A., Dovidio, J., & Gaertner, S. (2009). The Nature of Contemporary Prejudice: Insights from Aversive Racism. Social And Personality Psychology Compass, 3(3), 314-338. doi: 10.1111/j.1751-9004.2009.00183.x
Pletcher, M., Kertesz, S., Kohn, M., & Gonzales, R. (2008). Trends in Opioid Prescribing by Race/Ethnicity for Patients Seeking Care in US Emergency Departments. JAMA, 299(1). doi:10.1001/jama.2007.64
Projectimplicit.net,. (2011). ProjectImplicit. Retrieved 30 January 2016, from https://www.projectimplicit.net/index.html
Rachlinski, J., Johnson, S., Wistrich, A., & Guthrie, C. Does Unconscious Bias Affect Trial Judges? SSRN Electronic Journal. doi: 10.2139/ssrn.999490
Roter, D. (1989). Communication and medical practice: Social relations in the clinic. Patient Education And Counseling, 13(3), 325-326. doi: 10.1016/0738-3991(89)90025-6 Roter, D. L., Hall, J. A., & Katz, N. R.. (1987). Relations between Physicians' Behaviors and
Analogue Patients' Satisfaction, Recall, and Impressions. Medical Care, 25(5), 437– 451. doi: 10.1097/00005650-198705000-00007
Sabin, J., & Greenwald, A. (2012). The Influence of Implicit Bias on Treatment
Recommendations for 4 Common Pediatric Conditions: Pain, Urinary Tract Infection, Attention Deficit Hyperactivity Disorder, and Asthma. Am J Public Health, 102(5), 988-995. doi: 10.2105/ajph.2011.300621
Schwartz, M., Chambliss, H., Brownell, K., Blair, S., & Billington, C. (2003). Weight Bias among Health Professionals Specializing in Obesity. Obesity Research, 11(9), 1033-1039. doi:10.1038/oby.2003.142
Schwartz, M., Vartanian, L., Nosek, B., & Brownell, K. (2006). The Influence of One’s Own Body Weight on Implicit and Explicit Anti-fat Bias. Obesity, 14(3), 440-447. doi: 10.1038/oby.2006.58
Sheehan, T., Candee, D., Willms, J., Donnelly, J., & Husted, S. (1985). Structural Equation Models of Moral Reasoning and Physician Performance. Evaluation & The Health Professions, 8(4), 379-400. doi: 10.1177/016327878500800401
Smith, C. K., Polis, E., & Hadac, R. R. (1981). Characteristics of the initial medical interview associated with patient satisfaction and understanding. The Journal of family practice, 12(2), 283-288.
Stewart, M. (1984). What is a successful doctor-patient interview? a study of interactions and outcomes. Social Science & Medicine, 19(2), 167-175. doi: 10.1016/0277-9536(84)90284-3
Street, R., Gordon, H., & Haidet, P. (2007). Physicians’ communication and perceptions of patients: Is it how they look, how they talk, or is it just the doctor?. Social Science & Medicine, 65(3), 586-598. doi:10.1016/j.socscimed.2007.03.036
van den Bergh, L., Denessen, E., Hornstra, L., Voeten, M., & Holland, R. (2010). The Implicit Prejudiced Attitudes of Teachers: Relations to Teacher Expectations and the Ethnic Achievement Gap. American Educational Research Journal, 47(2), 497-527. doi: 10.3102/0002831209353594
van Ryn, M., & Burke, J. (2000). The effect of patient race and socio-economic status on physicians' perceptions of patients. Social Science & Medicine, 50(6), 813-828. doi: 10.1016/s0277-9536(99)00338-x
Willems, S., De Maesschalck, S., Deveugele, M., Derese, A., & De Maeseneer, J. (2005). Socio-economic status of the patient and doctor–patient communication: does it make a difference? Patient Education And Counseling, 56(2), 139-146. doi:
10.1016/j.pec.2004.02.011
Willson, P., & McNamara, J. (1982). How perceptions of a simulated physician-patient interaction influence intended satisfaction and compliance. Social Science & Medicine, 16(19), 1699-1704. doi: 10.1016/0277-9536(82)90095-8
Ziegert, J., & Hanges, P. (2005). Employment Discrimination: The Role of Implicit
Attitudes, Motivation, and a Climate for Racial Bias. Journal Of Applied Psychology, 90(3), 553-562. doi: 10.1037/0021-9010.90.3.553
