Reflection paper

Organizational Behavior and Human Decision Processes 101 (2006) 152–167

www.elsevier.com/locate/obhdp

Implicit sources of bias in employment interview judgments and decisions �

Sharon L. Segrest Purkiss a, Pamela L. Perrewé b,¤, Treena L. Gillespie a, Bronston T. Mayes a, Gerald R. Ferris b

a Department of Management, College of Business and Economics, California State University, Fullerton, Fullerton, CA 92834-6850, USA b Department of Management, College of Business, Florida State University, Tallahassee, FL 32306-1110, USA

Received 17 March 2003 Available online 7 September 2006

Abstract

This study empirically examined implicit sources of bias in employment interview judgments and decisions. We examined two eth- nic cues, accent and name, as sources of bias that may trigger prejudicial attitudes and decisions. As predicted, there was an interac- tion between the applicant name and accent that aVected participants’ favorable judgments of applicant characteristics. The applicant with the ethnic name, speaking with an accent, was viewed less positively by interviewers than the ethnic named applicant without an accent and non-ethnic named applicants with and without an accent. Furthermore, modern ethnicity bias had a negative association with the favorable judgments of the applicants, which, in turn, aVected hiring decisions. Implications of the results, limi- tations of the study, and directions for future research are discussed. © 2006 Elsevier Inc. All rights reserved.

Keywords: Employment interview; Modern ethnicity bias; Implicit sources of bias; Decision to hire

Introduction

The employment interview is an important source for information and remains, by far, the most frequently used employment selection and decision-making device in orga- nizations (e.g., Eder & Harris, 1999; Posthuma, Morgeson, & Campion, 2002). Unfortunately, this reality stands in stark contrast to the continued questions about interview validity and the persistence of biases in the interview pro- cess, suggesting that more research in this area is needed (Pingitore, Dugoni, Tindale, & Spring, 1994; Roehling, Campion, & Arvey, 1999). In particular, the eVects of vari-

� David Harrison provided unusual and outstanding guidance and support throughout the review and revision process. The authors feel that his insight made an important contribution to the quality of this paper and we would like to express our gratitude to him.

* Corresponding author. Fax: +1 850 644 7843. E-mail address: [email protected] (P.L. Perrewé).

0749-5978/$ - see front matter © 2006 Elsevier Inc. All rights reserved. doi:10.1016/j.obhdp.2006.06.005

ous interviewer and applicant characteristics on the inter- view process and outcomes deserve additional exploration (Dipboye, 1992). Although some research has addressed these issues, most of the studies of interviewer biases and stereotypes have focused on non-subtle, demographic eVects on interviewers’ judgments and decisions.

In a review of the interview literature, Posthuma et al. (2002) suggested that researchers redirect attention from examining simple demographic eVects and consider these as potential cues for other causal factors, particu- larly attitudes and values. The present study addresses this appeal, and it extends previous work on applicant characteristics by focusing on the eVects of implicit or subtle cues on interview outcomes within the framework of modern racism or modern ethnicity bias. SpeciWcally, the purpose of the present study is to investigate the extent to which ethnic name and accent serve as cues that trigger biased interviewer judgments and decisions in the employment interview process.

S.L. Segrest Purkiss et al. / Organizational Behavior and Human Decision Processes 101 (2006) 152–167 153

Employment interview

As a traditional component of most organizations’ human resource management selection systems, research has been conducted on the employment interview for nearly a century (e.g., Eder & Harris, 1999). Interview scholars have been interested in a broad range of topics over the years, including psychometric properties of the interview as a measurement device, interview format and type (e.g., structured, unstructured and situational), notions of Wt (e.g., person–job and person–organization), and interviewer cognition and decision-making pro- cesses. However, we need more employment interview research examining the eVects of applicant demographic characteristics as cues aVecting interviewer judgments and decisions.

Research needs to probe beyond simple demographic category eVects to investigate potential underlying rea- sons for what are observed as judgment and decision biases. With increased interest in person–organization Wt in the interview, there has come a realization that the homogenization eVects from such assessments, which drive employment decisions, potentially could account for discrimination eVects (e.g., Judge & Ferris, 1992). However, we still need to know much more about the perceptual cues associated with applicants of diVerent races and ethnicities that might be driving these assess- ments, judgments, and employment decisions.

The psychological processes of prejudice and stereotyping

Prejudicial attitudes, as well as other interviewer char- acteristics, such as race and personality, aVect inter- viewer perceptions of applicants (Dipboye, 1992). Prejudice and ethnicity stereotypes tend to be positively related to each other in both the historical and current views (Dovidio, Brigham, Johnson, & Gaertner, 1996), and some researchers have agreed that the positive rela- tionship is due to stereotypes being the cognitive compo- nent of racial attitudes or prejudice (Jones, 1986). Stereotypes are particular types of knowledge structures or cognitive schema that link group membership to cer- tain traits (Ford & Stangor, 1992; Nesdale & Rooney, 1990), and which have been found to inXuence the inter- pretation of others’ behavior (Duncan, 1976), the mem- ory of others (Stangor & McMillan, 1992), and behavior toward others (Snyder, Tanke, & Berscheid, 1977).

Research has suggested that prejudice tends to evoke negative stereotypes. Participants with high levels of prejudice are more likely to use cultural stereotypes, and high levels of prejudice correspond with more negative stereotypes (Kawakami, Dion, & Dovidio, 1998). These stereotypes could have been elicited through the use of a cue, and in the Kawakami et al. (1998) study, the cate- gory label, Black, was purported to activate the stereo- type. This type of cue likely activates judgments of a

speciWc group without the awareness of the perceiver, consistent with an implicit form of racism, called mod- ern racism (McConahay, 1986).

We would expect that prejudice against a speciWc eth- nic group (e.g., Hispanics) would aVect judgments about that group, but would not necessarily inXuence judg- ments about a diVerent group (e.g., non-Hispanics). It is important to note that prejudicial attitudes and stereo- types about race and ethnicity may be generated by mul- tiple cues. We argue that examining multiple cues, such as ethnic accent and name, is key to understanding how prejudicial attitudes and stereotypes are triggered. Work in these areas is examined next.

Ethnic speech accent and name

Subtle cues may play a role in triggering implicit dis- criminatory responses. One possible cue may be appli- cant accent. Whereas other Welds, such as linguistics and communication, have recognized the important role of accent in the perception of individual characteristics, organizational research has neglected this area. Accent can initiate perceptions regarding intelligence and kind- ness, as well as status, solidarity, economic class, national origin, or ethnicity (Lippi-Green, 1994; Nesdale & Rooney, 1990).

For example, in the U.S., French accents often are associated with sophistication, Asian accents tend to be linked with high economic and educational attainments (Cargile, 2000; Lippi-Green, 1997), and in England, the Liverpool accent is considered less cultured than accents associated with Oxford and Cambridge (Lippi-Green, 1997). Due to the verbal nature of the employment inter- view process, and the potential for triggering biased judgments, accent may prove to be a particularly impor- tant factor aVecting interview decisions.

Although it may be subtle, accent has been demon- strated to be easily perceptible. Research has demon- strated that even linguistically naïve individuals can make basic distinctions among diVering accents (Cargile, 2000; Giles, Williams, Mackie, & Rosselli, 1995; Podb- eresky, Deluty, & Feldstein, 1990). However, this recog- nition of accent distinctiveness seems to apply only to a certain degree. SpeciWcally, when presented with four varieties of Spanish-accented English (i.e., Cuban, Costa Rican, Argentinean, and Puerto Rican), and four varie- ties of Asian-accented English, most American listeners could not distinguish between the diVerent varieties of Spanish- and Asian-accented English speech (Podbere- sky et al., 1990). It appears that a general Spanish accent is recognized by most listeners, and often evokes similar reactions, regardless of the speciWc variety of Spanish spoken.

Accents associated with countries of lower socio-eco- nomic status and darker skin colors frequently are deni- grated (Lippi-Green, 1997). However, some regional

154 S.L. Segrest Purkiss et al. / Organizational Behavior and Human Decision Processes 101 (2006) 152–167

accents are looked upon less favorably, even when skin color is not an issue. For example, in the U.S., “Appala- chian English” is downgraded (Atkins, 1993). In general, the accent of the dominant or majority group in a soci- ety is evaluated most positively (Nesdale & Rooney, 1990). Interestingly, the dominant accent often is judged more positively not only by the dominant group, which is Anglo Americans in the U.S., but also by minority groups, such as African Americans and Hispanics (Bren- nan & Brennan, 1981; DeShields, Kara, & Kaynak, 1996). Therefore, one would expect that interviewers would evaluate applicants more positively if their accent matched that of the majority group, regardless of whether the interviewers were minority group members.

Another problem occurs when the limited selection research examining minorities fails to distinguish between race and ethnicity. Although African Ameri- cans and Hispanics both share the distinction of being minorities in the United States, one diVerence needs to be clariWed. The terms race and ethnicity often elicit con- fusion. Race has been deWned as a social grouping based on visible physical characteristics, such as skin color, and on supposed common ancestral origins, whereas ethnic- ity has been deWned as a group’s cultural and social heri- tage that has been transferred through generations of group members (BolaY, Braham, Gindro, Tentori, & Bracalenti, 2003; Singer & Eder, 1989; Slavin, Rainer, McCreary, & Gowda, 1991). For example, one study selected research participants on the basis of their appearance, speech, and name being indicative of His- panic descent (Kenney & Wissoker, 1994). The study, designed to diVerentiate between the success of an Anglo and a Hispanic job candidate, revealed that an Anglo candidate was more likely to be successful than the His- panic counterpart at Wling an application, obtaining an interview, and receiving a job oVer. However, these results may have been confounded by the failure to con- trol for race and/or accent, exemplifying the diYculties of research in this area.

Similarly, a Weld study found signiWcant same-race bias between the interviewer and applicants for custodial jobs. It was acknowledged that, whereas their Black/ White and Black/Hispanic comparisons examined racial diVerences, the White/Hispanic comparisons examined diVerences in ethnic background (Lin, Dobbins, & Farh, 1992). Even though it was stated that race similarity eVects were examined, it appeared that no data were gathered regarding the actual race of the Hispanic appli- cants, or other potentially confounding factors, such as the degree of accent.

Singer and Eder (1989) separated the eVects of ethnic- ity and accent cues in the selection process, and found negligible eVects for accent, but signiWcant eVects for eth- nicity. In contrast to the statistical results, participants in the role of interviewer perceived and rated applicant accent as moderate in importance and applicant ethnic-

ity as low in importance in their selection decision (Singer & Eder, 1989). This suggests that interviewers may rely on applicant accent as a more concrete, legiti- mate justiWcation for ethnicity discrimination. Consider- ing applicant accent also could reXect some implicit theory on the part of the interviewer that an applicant should not have an accent, because having an accent might aVect job performance negatively. Of course, it also might be that accent is, in fact, job-related, and reXects an important requirement in job applicants.

An ethnic cue (e.g., accent) that is paired with another minority ethnic group cue (e.g., name) may evoke a con- sistent stereotype, resulting in a negative evaluation of an applicant. Consistent with the premise of modern rac- ism, these negative judgments are likely made automati- cally, not consciously. However, when accent alone or ethnicity alone is perceived, a single cue may not be enough to trigger modern racism. Because research inconsistently has demonstrated lower evaluations of minority applicants (Lewis & Sherman, 2003; Mullins, 1982; Vrij & Winkel, 1994), it may be a combination of cues that elicit modern racism. In other words, when two minority ethnic cues are paired together, the ethnicity of the target person is clearer, evoking automatic negative stereotypes. However, if only one cue is present, the eth- nicity is less clear, which may trigger a more conscious process of evaluation.

Modern racism and the complex eVects of ethnic cues

Racial and ethnic demographic classiWcations may solicit categorical reactions and decisions that really mask underlying subtle cue eVects related to individuals who are members of those categories. Modern racism could be a potential explanation for such eVects. Racial prejudice is deWned as “an unfair negative attitude toward a social group or a person perceived to be a member of that group” (Dovidio, 2001, p. 329). Prior to the Civil Rights era, prejudice was viewed as a psychopa- thology, with those perpetuating prejudice as individuals in need of reform (Dovidio, 2001). Racial prejudice was deWned as blatant and overt. However, this “old fash- ioned” racism soon melted into a more implicit form of racism, or “modern racism” (McConahay, 1986). This more subtle racial prejudice was recognized as a normal process that emerged from, and was perpetuated by, socialization and social norms (Dovidio, 2001).

As unintentional and subtle, individuals who are high in modern racism may denounce racism, but still act in ways that discriminate against others without con- sciously doing so (Dovidio, 2001). Modern racists espouse egalitarianism, so they do not openly discrimi- nate. However, their underlying feelings may lead them to engage in unintentional discrimination when another factor exists to sway their decision (Dovidio, 2001). For example, Dovidio and Gaertner (2000) discovered that

S.L. Segrest Purkiss et al. / Organizational Behavior and Human Decision Processes 101 (2006) 152–167 155

when applicants had marginally acceptable qualiWca- tions, Black candidates were less often selected than White candidates, although their qualiWcations were identical. However, when qualiWcations were low (i.e., a clear need to reject the candidate) or qualiWcations were high (i.e., a clear decision to select the candidate), bias was not evident.

Modern racism, hereafter, referred to as modern eth- nicity bias, oVers researchers the context for less detect- able discrimination in the workplace. Whereas blatant or “old-fashioned” racism is unacceptable and illegal as a means for making selection decisions, subtle cues may be triggering unconscious or implicit forms of ethnicity bias in judgments and decisions.

A combination of ethnic minority cues (i.e., as opposed to a single cue) may be more likely to trigger an unconscious and automatic negative reaction because of the salience of the cues and the ease in which one is more conWdent about placing someone in a class or category; essentially, stereotyping. Further, “when one’s attention is diVerentially directed to one portion of the environ- ment rather than to others, the information contained in that portion will receive disproportionate weighting in subsequent judgments” (Taylor & Thompson, 1982, p. 175). Thus, observing a combination of two or more eth- nic cues might lead to an unconscious, automatic, nega- tive labeling of an individual. However, a single ethnic cue is less likely to trigger an automatic stereotype. In this case, a single cue might trigger a more conscious process of labeling an individual. When an individual is conscious of placing another into a class or category, ste- reotyping due to modern racism is less likely to occur.

Hypotheses

Based on the previous discussion, we argue that the interaction of ethnic cues (i.e., ethnic name and accent) is more likely to elicit ethnic stereotypes and negative appraisals than a single cue. Thus, the following hypoth- esis is proposed:

Hypothesis 1. Ethnic name and ethnic accent will inter- act to predict unfavorable judgments about the appli- cant and a reduced likelihood of deciding to hire the applicant. The synergistic combination of two ethnic cues (i.e., when both ethnic name and ethnic accent are present), will lead to the most negative judgments of applicants.

Within the context of modern ethnicity bias, individu- als may unconsciously attend to the ethnicity without recognizing its impact on their decisions. We expect that those who have an ethnicity bias will react more nega- tively to ethnic cues and will be more likely to hold unfa- vorable judgments about ethnic minorities than those who do not have an ethnicity bias. Further, these judg- ments will likely aVect interviewers’ decisions to hire eth-

nic minorities. In accordance with the previous discussion of ethnicity bias, it would be expected that ethnic applicants also would be judged less favorably in an interview context by both minority and non-minority interviewers. In light of the relationship between prejudi- cial attitudes, stereotypes, and perceptions of ethnic group members, the following relationship is hypothe- sized:

Hypothesis 2. The ethnicity of the applicant will interact with modern ethnicity bias such that the negative rela- tionship between the ethnicity of the applicant and judg- ments and decisions to hire the applicant will be exacerbated when modern ethnicity bias is high.

Method

Participants

Two hundred and twelve students enrolled in basic management courses at a large southeastern university voluntarily participated in this study in exchange for extra course credit. The mean age of the participants was 22 with a range from 18 to 47 years. The average total work experience was 2.7 years. The ethnicity composition of the sample was as follows: 66% Caucasian (not of His- panic origin); 18% African American; 11% Hispanic; 4% Asian/PaciWc Islander; and 1% Other. For data analysis purposes, the following three classiWcations were used for participants’ ethnicity: 0DCaucasian/White (not of His- panic origin); 1DOther Minorities; 2DHispanic. Fifty- six percent of the participants were male.

Procedure

Although previous employment interview research has examined ethnicity cues on interviewer decisions, the study of Hispanic ethnicity has been neglected relative to other minority applicants (Lin et al., 1992). This dearth in research persists even though the Hispanic segment of the population is growing rapidly in the United States (Grow, 2004; Mosisa, 2002; Sanchez & Brock, 1996), and despite evidence that biases against Hispanic employees exist (Kenney & Wissoker, 1994; Sanchez & Brock, 1996). Thus, we chose to examine sources of bias toward Hispanics.

In order to reduce the potential of experimenter bias due to diVerences in sex or ethnicity, White (non-His- panic), male doctoral students were selected and trained to administer the surveys. The administrators were per- sonally trained by one of the researchers, and given spe- ciWc written instructions to follow.

Two large entry-level management classes (ND115, ND 150) were selected for the study. One week prior to data collection, the instructor informed these students of

156 S.L. Segrest Purkiss et al. / Organizational Behavior and Human Decision Processes 101 (2006) 152–167

an extra credit opportunity that would take place the following week during the scheduled class time. The par- ticipants reported to their regular classroom where they were randomly assigned to one of four separate, prear- ranged rooms. All participants were told they would serve as employment interviewers, and they were asked to watch a video of a job applicant participating in an interview. However, the video in each of the rooms diVered on the basis of applicant name and applicant accent. Participants were exposed to one of four condi- tions: a Hispanic accent with a Hispanic name, a His- panic accent with a non-Hispanic name, a standard American-English accent with a Hispanic name, or a standard American-English accent with a non-Hispanic name.

After the participants reported to their assigned room, they were seated so that they could clearly view and hear the videotaped interview. They signed an informed consent form. The participants were instructed to imagine that they were hiring for the Human Resources Manager position. The general procedure was then explained.

The participants were given the job description and the resume with the appropriate name manipulation (Michael Fredrickson/Miguel Fernandez) to review. The job description for the Human Resource Manager was adapted from the Dictionary of Occupational Titles (U.S. Department of Labor, 1991). The resume informed the participants that the applicant had the following qualiW- cations: an MBA with a concentration in Human Resource Management from a large state university (3.7 Grade Point Average – GPA); a B.S. in Business Admin- istration from a large state university (3.5 GPA); and internship experience with two major corporations, per- forming duties such as designing training programs, updating job position descriptions, and working with salary surveys.

Finally, the participants watched the 10-min video- taped job interview that included the accent manipula- tion and name manipulation (i.e., ethnicity cue), and then answered a two-part anonymous survey related to the interview. When participants Wnished with the Wrst part of the survey, they turned it in to the administrator and received the second matched part of the survey. The Wrst part of the survey contained questions related to the following: the applicant’s perceived characteristics, the interviewer’s attitude toward hiring the applicant, decision to hire, hire decision, participant demographics, and perceptions of the videotaped job applicant’s demo- graphics.

Precautions were taken to conceal the true nature of the study. Items tapping individuals’ perceptions of accent and ethnicity were embedded among many other demographic-type items. The second part of the survey included the ethnicity bias scale questions. This section of the survey was given separately in order to prevent the

participants’ answers on the Wrst part of the survey from being primed by the modern ethnicity bias scale ques- tions.

Experimental manipulations

Early linguistics researchers often used a matched- guise technique in an experimental situation in order to control for extraneous factors. The present study utilized this approach to examine the inXuence of ethnicity cue (i.e., name) and accent in the interview process by having the same actor perform identical interview scripts while the accent and ethnicity cues of the actor were manipu- lated. With this technique, factors such as appearance and voice tone were held constant in order to focus on the variables of interest (i.e., accent and ethnicity cue).

When creating the matched-guise videotapes for the accent and ethnicity cue manipulations, details were thoroughly considered to insure quality manipulations. An experienced videographer donated her time and equipment to the project, including a professional video camera, lights, and microphones. An interview script, adapted from research by Howard and Ferris (1996), was used to ensure the same information was communi- cated in all conditions. The interview script combined with the applicant’s resume showed that the applicant was articulate, enthusiastic and motivated, as well as highly qualiWed for the position with the relevant univer- sity degrees and work experience.

Accent Auditions were held in order to Wnd an actor for the

applicant role who had the ability to speak with a stan- dard American-English accent and a Hispanic accent. The chosen actor was a White male who had experi- ence with theater and with national commercials in both Spanish and English. Three linguistic experts independently veriWed the realism and the understand- ability of the Hispanic accent. The actor was instructed to keep body movements, facial expression, and pos- ture as similar as possible in both the accented and non-accented conditions. Because the applicant was the same person in both videos, factors such as applicant attractiveness, tone of voice, and other mannerisms were virtually identical. Finally, the actor wore the same conservative business suit and tie in both condi- tions.

The second actor, a male with a standard American- English accent, assumed the role of the interviewer and was not shown in the video to prevent interference with the manipulations. Both actors had microphones, with the applicant’s microphone hidden under his clothing to prevent interference with realism. Participants were either exposed to the Hispanic-accented applicant (coded 1) or to the standard American-English accent (coded 0).

S.L. Segrest Purkiss et al. / Organizational Behavior and Human Decision Processes 101 (2006) 152–167 157

Name For the name manipulation, two identical resumes

were constructed, with the only diVerence being the name. Miguel Fernandez was used for the Hispanic eth- nic cue, and Michael Fredrickson was used for the non- Hispanic ethnic cue. Thus, participants encountered one of the following four conditions: Miguel with a standard American-English accent; Miguel with a Hispanic accent; Michael with a standard American-English accent; or Michael with a Hispanic accent. Additionally, the videotapes were professionally edited to reinforce the name manipulation by inserting the title “Human Resource Manager Applicant: Michael Fredrickson (Miguel Fernandez)” into the introduction. At the begin- ning of the videotaped interviews, the applicant name was displayed for approximately 7 s. Editing also was used to insert the beginning segment of dialogue in which both the interviewer and the applicant use the appropriate applicant name for added emphasis. The participants were asked to write the applicant name on the survey to check that they were aware of the name manipulation. All of the participants correctly recorded the applicant’s name.

The main actor (Miguel/Michael) was a White male. Care was taken to choose an individual with physical characteristics such as white skin, brown eyes, and brown hair that could typically be considered either Anglo American or Hispanic American. This race/gen- der mix was chosen in an eVort to control for the poten- tially negative eVect of other races and sex, because white males are still predominant in high-status posi- tions in U.S. organizations (Ely, 1995). Controlling the race of the individual was imperative; as previously men- tioned, past studies examining biases against Hispanics have failed to control for this potentially important factor. The participants were exposed to either Michael (non-Hispanic, coded 0) or to Miguel (Hispanic, coded 1).

Model variables

Interviewer perceptions of applicant accent Perceptions of accent were measured with an item

used in previous linguistics research (Ryan, Carranza, & MoYe, 1977). This item was embedded with other items that measured the applicant’s perceived characteristics in an eVort to conceal the fact that accent was a main variable of interest in the study. Participants rated the applicant on a seven-point scale ranging from ethnic accented speech to Standard American accented speech (e.g., television/radio accent). Higher scores indicate per- ceptions of ethnic accented speech.

Interviewer perceptions of applicant ethnicity The participants indicated which of the following cat-

egories they believed applied to the video applicant:

Caucasian (0); African American (1); Hispanic (2); Native American (3); Asian/PaciWc Islander (4); and Other (5). These categories for race/ethnicity were based on EEOC guidelines. Perceived ethnicity was coded “0” for Caucasian, “1” for Other Minority, and “2” for His- panic.

Modern ethnicity bias Currently, no published modern ethnicity bias scale

exists in the research literature that focuses speciWcally on Hispanics. Therefore, a scale was derived from McC- onahay’s (1986) Modern Racism Scale in order to spe- ciWcally assess the degree of the participants’ biases against Hispanics.

The Modern Racism Scale was originally designed to inconspicuously measure prejudice against African Americans (McConahay, 1986). In our scale, all occur- rences of the word “African American/s” were changed to “Hispanic/s”. In addition to the word adaptations, Wve items were added to the scale based on research examining controversial issues related to Hispanics, such as Spanish language usage in the United States, border crossing issues, aYrmative action, and treatment of migrant farm workers. One item that related to segrega- tion issues appeared to be irrelevant to Hispanics, so it was adapted to reXect issues related to Hispanics and school language issues.

All of the items were signiWcantly correlated, and the Cronbach � reliability estimate was .79 in this study, and .85 in an earlier pilot study. Evidence of the construct validity (Nunnally, 1978) of our measure is demon- strated in the present study by virtue of its signiWcant negative correlation with perceptions of Hispanics (rD¡.42, p < .001). Participants who have higher scores on the modern ethnicity bias scale tend to describe His- panics in more negative terms than those with lower scores on the scale. Please see the Appendix A for all of the items included in the modern ethnicity bias scale. Participants responded to a 7-point scale (7D strongly agree and 1D strongly disagree) with higher numbers indicating greater levels of prejudice.

This scale was designed to be less vulnerable to social desirability eVects due to the type of questions used. The items used for this scale dealt with issues that are politi- cal in nature (e.g., Discrimination against Hispanics is no longer a problem in the United States), instead of directly asking the respondent about their prejudice (e.g., Do you believe that Hispanics are less industrious than non-Hispanics). Therefore, it assessed prejudice in a less conspicuous manner than previous scales.

Interviewer favorable judgments of applicant characteristics

Participants’ judgments of the applicant’s characteris- tics were assessed by asking the participants to rate the applicant on 26 bipolar pairs of adjectives using a

158 S.L. Segrest Purkiss et al. / Organizational Behavior and Human Decision Processes 101 (2006) 152–167

seven-point scale, with 7 as the anchor for favorable traits and 1 as the anchor for unfavorable traits (�D .87). The adjective pairs were adapted from previous research focusing on characteristics of the ideal employee, eVec- tive top managers, and motivated workers (Larkin & Pines, 1979), and from research concentrating on His- panics and accent discrimination by employment recruiters (Brennan & Brennan, 1981). The following are examples of the adjective pairs used: unsuccessful–suc- cessful, lazy–industrious, unstable–stable, and tardy– prompt.

Interviewer decision to hire Three statements, coded 1–7 (1D strongly disagree,

7D strongly agree; �D .94), measured interviewers’ deci- sion to hire the applicant. The scale items were: “I will probably NOT hire the applicant for the Human Resource Manager position” (reverse-coded); “It is likely that I WILL hire the applicant for the Human Resource Manager position”; and “I plan to hire the applicant for the Human Resource Manager position.” Higher scores indicate a stronger decision to hire the applicant.

Control variables

Social desirability Because past research has indicated that social desir-

ability among raters may aVect the results of ethnicity- oriented studies (Mullins, 1982), an abbreviated 10-item form of the social desirability scale was used (Strahan &

Gerbasi, 1972) (coded 1–7, with 1D strongly disagree, 7D strongly agree; �D .70) as a control variable. Higher scores indicate a tendency to give socially desirable responses.

Accent understandability Due to concerns that negative evaluations might be

due to the applicant not being understandable, and not due to the ethnic cues of accent and name, we controlled for the understandability of the applicant. We asked par- ticipants to indicate their ability to understand the appli- cant’s speech on a one-item, seven-point scale that ranged from “not understandable accent” to “under- standable accent”. Higher scores indicate a higher degree of understandability.

Participants’ demographic features Self-reported demographic information on partici-

pants’ race/ethnicity, gender, GPA, and work experience were collected and used as control variables, based on previous research suggesting these variables may bias the results (Kenney & Wissoker, 1994; Vrij & Winkel, 1994).

Results

Table 1 presents the means, standard deviations, and the zero-order correlations among study variables. After list-wise deletion of cases with missing data, 200 partici- pants were included in the analyses. Diagonal entries indicate the internal consistency reliability estimates (i.e.,

Table 1 Means, standard deviations, and correlations among study variables (diagonal values are reliabilities) (N D 200)

One-tail signiWcance: *p 6 .05; **p 6 .01; ***p 6 .001. a 0 D Anglo name cue; 1D Hispanic name cue. b 0 D North American English accent; 1 D Hispanic accent. c 0 D Caucasian; 1 D Other Minority; 2D Hispanic. d 0 D Caucasian; 1 D Other Minority; 2D Hispanic. e 0 D Male; 1 D Female.

M SD 1 2 3 4 5 6 7 8 9 10 11 12 13

1. Applicant Namea 0.52 0.50 — 2. Applicant Accentb 0.46 0.50 ¡.118* — 3. Perceived Accent 3.70 1.97 ¡.122* .792*** — 4. Perceived Accent

Understandability 5.95 1.47 .114 ¡.198** ¡.155* —

5. Perceived Applicant Ethnicityc

1.53 0.80 .312*** .380*** .376*** ¡.014 —

6. Modern Ethnicity Bias 3.18 0.95 ¡.064 ¡.017 .060 ¡.086 ¡.065 .84 7. Interviewer Favorable

Judgment of Applicant 5.09 0.54 .032 ¡.094 ¡.146* .292*** .172** ¡.223*** .87

8. Interviewer Decision to Hire

5.10 1.60 .035 ¡.002 ¡.039 .172** .116* ¡.286*** .447*** .94

9. Social Desirability 4.36 0.87 ¡.028 .036 .020 .071 ¡.053 ¡.192** .151* .160* .70 10. Work Experience

(Months) 5.05 3.65 .061 ¡.123* ¡.032 .019 ¡.075 ¡.145* ¡.076 ¡.071 .160* —

11. Grade Point Average 3.00 0.43 .013 .034 ¡.039 .001 .025 .009 ¡.148* ¡.058 ¡.042 ¡.052 — 12. Participant Ethnicityd 0.47 0.70 ¡.027 .096 .156* .005 .037 ¡.294*** .035 .083 ¡.001 ¡.053 ¡.131* — 13. Participant gendere 0.46 0.50 ¡.078 .093 ¡.014 ¡.089 .040 ¡.105 .118* .061 ¡.028 ¡.140* .186** .082 —

S.L. Segrest Purkiss et al. / Organizational Behavior and Human Decision Processes 101 (2006) 152–167 159

CoeYcients �). As indicated in the table, understandabil- ity and social desirability among participants had a greater number of signiWcant correlations with the study variables than the other control variables (i.e., work experience, gender, GPA, and participant ethnicity). Fol- lowing are more details regarding the manipulation checks, as well as the analytical results relating to the hypotheses.

Manipulation checks

For the manipulation checks, there were two primary considerations: the name and the accent manipulations. Among the four experimental conditions, the strongest ethnicity manipulation was when accent and Hispanic name were combined. In this case, 100% of the partici- pants identiWed the applicant as Hispanic or Other Minority. When Miguel had no accent, 85% of the par- ticipants identiWed the applicant as Minority, and when Michael had an accent, 96% of the participants identiWed him as Minority. In the condition where there were no Hispanic ethnicity cues, 32% of the participants identi- Wed the applicant as Hispanic.

We conclude that the combination of Hispanic name and accent is a strong cue to the ethnicity of the appli- cant. However, it is also clear that only one ethnic cue is needed to trigger the identiWcation of the applicant as Hispanic or Other Minority. Interestingly, 32% of the participants identiWed the applicant with no Hispanic

ethnicity cues as Hispanic or Other Minority. Because all of the participants were from management classes, per- haps topics such as diversity in organizations were salient to them, which may have aVected their percep- tion. Unfortunately, because the surveys were anony- mous, no follow-up interviews were possible.

The correlation between perceived accent and the manipulated applicant accent was .79 (p 6 .001), which demonstrates that the accent manipulation was eVective. We also asked the participants to indicate their ability to understand the applicant’s speech. A mean score of 5.95 (SDD 1.47), on a seven-point scale, was obtained for this measure, which indicates that the applicant was gener- ally well understood. The mean score for understand- ability in the Ethnic Accent condition was 5.64, and for the Standard Accent condition the mean score was 6.19 (tD 2.63, dfD188; p 6 .01). These results suggest that somewhat lower understandability ratings were pro- vided when a Hispanic accent was present, but in both conditions understandability was close to 6 on a 7-point scale.

Tests of hypotheses

We tested Hypotheses 1 and 2 using hierarchical regression, and Tables 2 and 3 provide the results of the data analyses for both hypotheses. To test Hypothesis 1, we entered the control variables, social desirability, mod- ern ethnicity bias, work experience, GPA, participant

Table 2 Hierarchical regression of interviewer favorable judgment of applicant on name cue, accent cue, and their interaction (N D 200)

Two-tail signiWcance: *p 6 .05; **p 6 .01; ***p 6 .001. a 0 D Caucasian; 1D Other Minority; 2D Hispanic. b 0 D Male; 1 D Female. c 0 D Anglo name cue; 1D Hispanic name cue. d 0 D North American English accent; 1 D Hispanic accent.

Independent variables Step 1 Step 2 Step 3

B � t B � t B � t

Control variables Accent understandability .104 .283 4.331*** .099 .270 4.017*** .089 .242 3.584***

Social desirability .068 .109 1.632 .114 .071 1.694 .071 .115 1.723 Work experience ¡.018 ¡.120 ¡1.790 ¡.019 ¡.128 ¡1.889 ¡.018 ¡.120 ¡1.787 GPA ¡.234 ¡.185 ¡2.768** ¡.231 ¡.183 ¡2.726** ¡.225 ¡.179 ¡2.682**

Participant ethnicitya ¡.052 ¡.067 ¡0.971 ¡.047 ¡.061 ¡0.878 ¡.036 ¡.047 ¡0.683 Participant genderb .161 .149 2.204* .165 .152 2.240* .161 .149 2.213*

Modern ethnicity bias ¡.112 ¡.197 ¡2.790** ¡.112 ¡.197 ¡2.776** ¡.109 ¡.192 ¡2.739**

Predictor IV’s Name cue manipulationc .005 .005 0.074 .157 .145 1.613 Accent cue manipulationd ¡.071 ¡.065 ¡0.965 .088 .081 0.872 Name-accent product ¡.325 ¡.243 ¡2.271*

Intercept 5.275 13.349*** 5.314 13.262*** 5.254 13.223***

Regression statistics R .441 .446 .469 F (df) Regression 6.639 (7,192)*** 5.242 (9,190)*** 5.337 (10,189)***

Adj. R2 .166 .161 .179 �R2 .195 .004 .021 F (df) of �R2 6.639 (7,192)*** .478 (2,190) 5.157 (1,189)*

160 S.L. Segrest Purkiss et al. / Organizational Behavior and Human Decision Processes 101 (2006) 152–167

ethnicity, perceived accent understandability, and partic- ipant gender in the Wrst step of the regression analysis predicting the favorability of judgments of the applicant. At step two, we entered the name cue manipulation and accent cue manipulation variables, and in step three, we entered the interaction of the name cue and accent cue manipulations. Step three produced a signiWcant interac- tion eVect (�R2D .021, F1,189D5.157, p < .05) for the pre- diction of interviewers’ judgments of the applicant. These results are presented in Table 2, indicating that Hypothesis 1 was partially supported. Namely, applicant name and accent interacted to predict interviewers’ favorable judgments of the applicant. The name and accent interaction was not signiWcantly related to deci- sion to hire.

Procedures outlined by Cohen and Cohen (1983) were used to compute regression equations showing the relationship between accent and favorable judgments of the applicant for the Hispanic and non-Hispanic name conditions. A graphic representation of these equations is shown in Fig. 1.

Fig. 1 demonstrates that perceptions of the Hispanic named applicant became more negative when the His- panic named applicant also had an accent. However, this eVect did not occur for the applicant with an Anglo name, which corresponds to research that suggests accents associated with countries of lower socio-eco- nomic status or darker skinned people are often viewed negatively (Ryan & Carranza, 1975; Callan, Gallois, & Forbes, 1983). Factors such as the applicant qualiWca-

tions, clothing, physical attractiveness, and age were held constant by utilizing the same person and identical resume content. Only accent and name varied, yet per- ceptions of the applicant changed. Moreover, in this study, the accented applicant was Xuent in English, used correct grammar, and had an understandable accent. Therefore, any accent discrimination against this indi- vidual could not be justiWed as a legitimate communica- tion issue.

As previously mentioned, it appears that only one ethnic cue is needed to trigger interviewers to identify applicants as Hispanic. However, as expected, one ethnic cue did not result in as negative of a judgment as the syn-

Fig. 1. Graphic representation of name cue £ accent cue interaction on interviewer judgment of applicant. For Hispanic Name: Interviewer Judgment of Applicant D¡.237£ Accent + 5.411. For Anglo Name: Interviewer Judgment of Applicant D .088 £ Accent + 5.254.

5.05

5.1

5.15

5.2

5.25

5.3

5.35

5.4

5.45

No Accent Accent

Anglo Name Cue

Hispanic Name Cue

Table 3 Hierarchical Regression of Interviewer Favorable Judgment of Applicant on Modern Ethnicity Bias, Perceived Applicant Ethnicity, and Their Inter- action (N D 200)

Two-tail SigniWcance: *p 6 .05; **p 6 .01; ***p 6 .001. a 0 D Caucasian; 1 D Other Minority; 2D Hispanic. b 0 D Male; 1 D Female. c 0 D Caucasian; 1 D Other Minority; 2D Hispanic.

Independent variables Step 1 Step 2 Step 3

B � t B � t B � t

Control variables Accent understandability .110 .299 4.508*** .105 .286 4.428*** .105 .285 4.390 Social desirability .088 .142 2.119* .073 .118 1.787 .073 .117 1.774 Work experience ¡.013 ¡.091 ¡1.345 ¡.016 ¡.108 ¡1.627 ¡.016 ¡.108 ¡1.625 GPA ¡.228 ¡.180 ¡2.652** ¡.238 ¡.188 ¡2.851** ¡.237 ¡.187 ¡2.826 Participant Ethnicitya ¡.007 ¡.009 ¡0.130 ¡.053 ¡.069 ¡1.010 ¡.052 ¡.067 ¡0.978 Participant genderb .185 .170 2.496* .159 .147 2.200* .159 .147 2.199

Predictor IV’s Perceived ethnicityc .110 .163 2.539* .083 .123 0.506 Modern ethnicity bias ¡.104 ¡.183 ¡2.627** ¡.117 ¡.206 ¡1.378 Product .008 .046 0.17 Intercept 4.724 13.561*** 5.057 12.674*** 5.099 10.879

Regression statistics R .403 .470 .470 F (df) Regression 6.229 (6,193)*** 6.780(8,191)*** 5.999 (9,190)***

Adj. R2 .136 .189 .184 �R2 .162 .059 .000 F (df) of �R2 6.229 (6,193)*** 7.226 (2,191)*** .029 (1,190)

S.L. Segrest Purkiss et al. / Organizational Behavior and Human Decision Processes 101 (2006) 152–167 161

ergistic eVect of two cues. Interestingly, the most favor- able judgment was triggered by applicants with a Hispanic name and with no accent.

Hypothesis 2 stated that participants’ modern ethnic- ity bias will interact with the ethnicity of the applicant such that the negative relationship between ethnicity and judgments and decisions to hire will be stronger when modern ethnicity bias is high. Hierarchical regression was used to test this hypothesis, with modern ethnicity bias, perceived applicant ethnicity, and their interaction as predictors of favorable judgments of applicant char- acteristics and decision to hire. The results of this analy- sis for favorable judgments are presented in Table 3.

The interaction term introduced at step 3 was not sig- niWcant, which indicates that participant modern ethnic- ity bias was not selectively associated with judgments about only the Hispanic applicant. However, step 2 was signiWcant (�R2D .059; F2,191D7.226, p < .001). The stan- dardized beta weight for perceived accent understand- ability (�D .286, tD4.428, p 6 .001) shows that the applicant was judged more favorably when his accent was perceived higher in understandability. The � weight for participant GPA (�D¡.188, tD¡2.851, p 6 .01) and gender (�D .147, tD 2.200, p 6 .05) were also signiWcant indicating that male participants and those with higher GPA’s tended to judge the applicant less favorably than female participants and those with lower GPA’s. The � for modern ethnicity bias (�D¡.183, tD¡2.627, p 6 .01) shows that this variable was negatively related to favor- able judgments of applicant characteristics.

In regard to decision to hire, the interaction term introduced at step 3 was not signiWcant, which indicates

that participant modern ethnicity bias was not selec- tively associated with the decision to (or not to) hire the Hispanic applicant. However, modern ethnicity bias was also negatively related to decision to hire (�D¡.261, tD¡3.564, p 6 .01).

Path analysis results

Although the hypothesized name and accent cue interaction was related to favorable judgments of the applicant, but not signiWcantly related to the decision to hire, modern ethnicity bias was signiWcantly related to these variables. To further understand the nature of the relationship among ethnic cues, modern ethnicity bias, judgments of the applicant, and decision to hire, we per- formed a path analysis using these variables. Fig. 2 shows the results of the path analysis.

Fig. 2 demonstrates that these causal paths explain signiWcant amounts of variance in decision to hire (R2D .23; F(5,206)D 12.456; p 6 .001). As one would expect, a signiWcant positive path was demonstrated between favorable judgments of the applicant and deci- sion to hire (�D .42; tD 6.562; p 6 .001). A signiWcant negative path to decision to hire was also obtained for modern ethnicity bias (�D¡.16; tD¡2.582; p 6 .01). SigniWcant negative paths to favorable judgments of the applicant were obtained for modern ethnicity bias (�D¡.24; tD¡3.703; p 6 .001), and the name/accent cue interaction (�D¡.31; tD¡2.934; p 6 .01).

Table 4 shows the decomposition of causal eVects through the path model (Alwin & Hauser, 1975). The path analysis demonstrates that modern ethnicity bias

Fig. 2. Path analysis of the eVects of modern ethnicity bias, name cue and accent cue on interviewer judgment of applicant and interviewer decision to hire.

Numbers on paths are r’s (i.e., double arrows) and b’s (i.e., single arrows). * p < .05 ** p < .01 *** p < .001

-.24***-.07

-.13*

.54***

Favorable Judgment of Applicant

R2 =.10***

Modern Ethnicity Bias

Decision to Hire R2=.23***

-.31**

.42***

.08

.10

Name Cue

Accent Cue

Name x Accent Interaction

.14

-.16**

-.04

.01

.50***

162 S.L. Segrest Purkiss et al. / Organizational Behavior and Human Decision Processes 101 (2006) 152–167

indirectly aVects the decision to hire through the inter- vening variable, favorable judgments of the applicant, but it also exerts a signiWcant negative direct eVect as well. The ethnicity cue interaction seems to aVect deci- sion to hire through its negative indirect eVect on favor- able judgments about the applicant, but its direct eVect is not signiWcant.

Discussion

Inappropriate, inaccurate, and even illegal decisions can occur regardless of the type of human resource man- agement selection device utilized. However, the employ- ment interview should be a prime target for research in this area, because it is the most frequently used tool for making employment decisions, and because more than other selection devices, the interview presents consider- able opportunity for the inXuence of subtle cues and per- ceptual and judgmental biases to aVect decisions. The present study examined the eVects of ethnic cues on interviewers’ favorable judgments and their decision to hire applicants. We hypothesized that the synergistic eVect of ethnic cues (i.e., ethnic name and ethnic accent), were more likely to trigger negative interviewer reactions toward an applicant than one ethnic cue or no ethnic cues. Even after controlling for participant modern eth- nicity bias, support was found for the eVect of an accent£ ethnicity cue interaction on the favorable judg- ments of applicants’ characteristics.

As hypothesized, the most unfavorable judgments of the applicant were triggered by the combination of ethnic name and accent. Interestingly, the most favor- able judgments were triggered when the applicant did not have an accent, but had an ethnic name. These

Wndings can be partially explained by the expectancy- violation theory (Jussim, Coleman, & Lerch, 1987), which suggests that there often are lower expectations for minorities, and when these expectations are vio- lated in a positive direction (i.e., no accent), evalua- tions will be in the direction of the violation. In other words, the Hispanic named applicant might have been viewed positively because he spoke without an accent. This theory is similar to the accommodation hypothe- sis reported in the linguistics Weld by Giles and Bour- his (1976), which states that eVorts by ethnic minorities to increase similarities between themselves and the majority group are associated with more favorable evaluations. To the extent that speech style contains prejudicial triggers, it can be altered to a style that is deemed more socially acceptable. Accommo- dating their speech allows minority members to poten- tially reduce social costs and increases the likelihood of social approval. Consistent with the accommoda- tion hypothesis, participants may have perceived the candidates to be more similar to the majority group, thus rating them more positively.

Modern ethnicity bias toward Hispanics was pre- dicted to relate negatively to favorable judgments of the Hispanic applicant, and not relate to judgments of the non-Hispanic applicant. We examined modern ethnicity bias, perceived applicant ethnicity, and their interaction as predictors of favorable judgments of applicant char- acteristics, and found that the interaction was not signiW- cant. This indicates that modern ethnicity bias was not selectively associated with judgments about only the Hispanic applicant. Based on this Wnding, we then exam- ined the main eVects of modern ethnicity bias on judg- ments of the applicant, attitudes about hiring, and decision to hire the applicant.

Table 4 Decomposition of direct and indirect eVects for modern ethnicity bias and name/accent cue interaction on decision to hire coeYcients (�)

Independent variables Dependent variables

Favorable judgment of applicant Decision to hire step 1 Decision to hire step 2

Modern ethnicity bias ¡.244*** ¡.267*** ¡.163*

Name cue .144 .018 ¡.043 Accent cue .101 .039 ¡.004 Name£ accent ¡.314** ¡.028 .105 Favorable judgment of applicant .423***

Dependent variables Independent variables Total eVect Indirect eVect via favorable judgment of applicant

Direct eVect

Favorable judgment of applicant Modern ethnicity bias ¡.244 — ¡.244 Name cue .144 .144 Accent cue .101 .101 Name £ accent ¡.314 ¡.314

Decision to hire Modern ethnicity bias ¡.265 ¡.103 ¡.162 Name cue .061 .061 .000 Accent cue .042 .042 .000 Name £ accent ¡.051 ¡.132 .081 Favorable judgment of applicant .421 .421

S.L. Segrest Purkiss et al. / Organizational Behavior and Human Decision Processes 101 (2006) 152–167 163

Modern ethnicity bias demonstrated a negative rela- tionship with favorable judgments of the applicant, and these judgments of the applicant showed a positive asso- ciation with the decision to hire. Further, modern ethnic- ity bias had a direct negative relationship with the decision to hire. Together, these results indicate that modern ethnicity bias seems to have a negative associa- tion with the favorable judgments of, and decisions to hire all applicants, not just Hispanic applicants. Perhaps interviewer ethnic biases trigger a skeptical and guarded view of others, which is translated into more negative perceptions of applicants in general. Additional research on the eVects of modern ethnicity biases is needed. Finally, having favorable judgments of applicants is associated with the decision to hire the applicant.

Limitations of the study

Some researchers may consider generalizability a problem when using students as interviewers. However, when examining ethnicity issues, the deviations between students and actual employees or managers may be min- imal (Barr & Hitt, 1986). This may be due to the fact that students, like managers, have been exposed to similar ethnic stereotypes through the media and society in gen- eral. If stereotypes are less prevalent among students, due to more progressive ideas among the youth of soci- ety, then the evidence of ethnicity bias among students found in this study may be a conservative estimate of the ethnicity bias of practicing managers. Alternatively, it may be that practicing managers are more experienced with, and aware of, discrimination issues, and therefore would be less likely to respond in a biased manner. In order to investigate these potential diVerences, examin- ing practicing professionals should help to extend this research.

A related concern is the potential lack of realism in the situation. As with most laboratory experimental situ- ations, some realism is lost, but control is gained, by allowing for more precise manipulation of the variables of interest. Posthuma et al. (2002), in a review of the research pertaining to interviews, suggested that having participants view an interview without actively partici- pating could lead to lack of involvement for partici- pants, thus alleviating the responsibility that organizational members may feel in a real interview situ- ation. Presumably, this lack of accountability could lessen participants’ attention to the task at hand.

In the present study, the procedure was designed in order to increase participant involvement. Interviewers were instructed to examine the applicant’s resume, to watch the interview carefully and imagine that they were actually interviewing the applicant, to rate the applicant on various aspects, and to make a hiring decision. Addi- tionally, the use of a matched-guise video provided an opportunity to control verbal and nonverbal cues (Post-

huma et al., 2002), which allowed accent to be teased apart and isolated from other cues in the environment.

In this study, the job description used for the appli- cant was for the Human Resource Manager position. The choice of this particular job description may have aVected the results if participants perceived that this job was associated with a certain degree of status. Previous research by Kalin and Rayko (1978) documented vary- ing eVects due to diVering degrees of job status. SpeciW- cally, they found that foreign accented applicants were given lower evaluations for high status jobs and higher evaluations for low status jobs. Therefore, diVerent job descriptions of varying degrees of status should be examined in future research.

Further, the qualiWcations of the applicant for the Human Resource Manager position in this study were high, indicating the applicant was clearly qualiWed for the position. Dovidio and Gaertner (2000) found that ethnic bias was most likely to occur when applicants had only reasonable qualiWcations and least likely to occur when applicants had qualiWcations that were either low (i.e., clear decision not to hire) or high (i.e., clear decision to hire). This study found ethnic biasing eVects even when the applicant was clearly qualiWed. In the work- place, it is likely that applicants will have some good qualities as well as some less attractive qualities and that interviewers are normally dealing with individuals who do not have such high qualiWcations that the decision to hire is clear. Thus, the results of this study may actually underestimate the degree to which ethnic biases aVect the judgments and decisions of interviewers.

The results of this research also may vary depending on where the study is conducted, and on the composition of the sample. In other parts of the country, Hispanic ethnicity cues may be perceived more or less readily, and Hispanics may face more or less discrimination. For example, in the Miami, Florida, area where there is a large Cuban-American population, is there more or less accent and general ethnicity discrimination against His- panics than in an area of the country where there is very little exposure to Hispanics? Research has found some support for the contention that ethnic minorities, like members of the majority ethnic group, tend to have neg- ative perceptions of ethnic accents and positive percep- tions of standard accents (Ryan & Carranza, 1975; Callan et al., 1983).

Other research has suggested that individuals are more likely to gravitate toward others who are similar. For example, in a study of workgroup preferences, indi- viduals demonstrated a clear desire for working with others who were racially similar (e.g., Hinds, Carley, Krackhardt, & Wholey, 2000). More research is needed to validate these results, and to examine various popula- tions of Hispanics (e.g., Mexican Americans, Puerto Ricans, and Cuban Americans) to observe whether there are any subculture diVerences in responses.

164 S.L. Segrest Purkiss et al. / Organizational Behavior and Human Decision Processes 101 (2006) 152–167

Implications and future research directions

A key contribution of the present study is that it allowed a closer look at the potential underlying triggers of bias in the employment interview process by examin- ing cues associated with ethnicity. Building on this study, there are some important directions for future research. One area that needs attention is to investigate the inXu- ence of interview structure on ethnic/racial cue eVects on interviewer decisions. The present study used a standard stimulus (i.e., videotaped interview) in presentation of applicant cues to an interviewer. This most closely resembles a structured interview format, where it would be argued that biases might be less observable because attention and focus is maintained on job-related content issues. Indeed, the unstructured interview tends to be where job-irrelevant information tends to emerge to inXuence interviewer decisions (e.g., Dipboye, 1994). It would be interesting to compare structured to unstruc- tured interviews to see if the observed eVects from this study regarding ethnic cues diVer by interview format.

There has been growing research interest in recent years in the use of applicant impression management tactics, and their eVects on interviewer ratings (e.g., Gil- more, Stevens, Harrell-Cook, & Ferris, 1999). Indeed, Gilmore et al. proposed an adaptation of the Ferris and Judge (1991) framework, which shows applicant impres- sion management tactics aVect interviewer decisions and actions through the potential mediating variables of lik- ing, perceived similarity, or perceived competence. It would be interesting to investigate whether applicants’ impression management tactics overshadowed, and thus neutralized, their race or ethnicity in aVecting interview- ers’ judgments and decisions. It might be the case that minority job applicant self-promotion tactics are suc- cessful in elevating their competence in the eyes of the interviewer to a level that eliminates any eVects of eth- nicity bias. This would be a new area of research because virtually no work has been done relating ethnicity to social inXuence (Ferris et al., 2002).

Besides interview format and impression manage- ment, future research should examine other factors that might constrain or magnify the eVects of applicant eth- nic/racial cues on interviewer judgments and decisions. It would be interesting to investigate the ethnicity of both applicant and interviewer in employment inter- views to see if there are rating eVects for ethnicity simi- larity. EVects of ethnicity or racial similarity on interview ratings have been reported for both African Americans (McFarland, Sacco, Ryan, & Kriska, 2000; Prewett-Livingston, Feild, Veres, & Lewis, 1996), and for Hispanics (Lin et al., 1992). However, all three of these studies used panel interviews as opposed to the more conventional one-on-one interviews. Because one-on-one interviews would seem even more likely to produce ethnic/racial similarity eVects (Sacco, Scheu,

Ryan, & Schmitt, 2003), interview scholars should pro- ceed in this direction.

Additionally, we would suggest that future research investigate the area of person–organization Wt as it relates to interviewer decision making regarding ethnic/ racial cues. Some recent work has proposed social-cogni- tive theoretical underpinnings for an integrative theory of multidimensional Wt that focuses on a prototype- matching approach (Wheeler, Buckley, Halbesleben, Brouer, & Ferris, 2005). Most Wt research in the employ- ment interview has investigated supplementary Wt, which considers how applicants seek to match particular char- acteristics they possess to the employing organization’s environment. The investigation of ethnic/racial cues in the area of person–organization Wt would be applicable to the concept of complimentary Wt, whereby applicants’ personal attributes and characteristics are viewed as adding something new that is not presently found in the organizational environment.

Future research is needed to determine if the results found in this study replicate to other scenarios or sam- ples. One area for further exploration is investigate whether name and accent cues trigger interviewer per- ceptions based on country of origin as well as race. For instance, is a Caucasian applicant with a Russian accent and Russian name perceived more negatively or posi- tively than when the applicant has a Russian name and no accent, or a Russian accent and non-Russian name? Furthermore, future research may address the extent to which these eVects occur in decision-making processes with internal applicants (e.g., promotions, opportunities for training). Perhaps these eVorts will help delineate the relative strength of name and accent cues in diVerent samples, as well as identify situations in which the eVects generalize.

The results of the present study suggest the need for continued eVorts to increase the eVectiveness of inter- viewer judgment and decision making. As demonstrated in this study, interviewers are vulnerable to making biased judgments about applicants. Potential solutions to reducing interviewer biases include training interviewers, structuring the rating procedures, using multiple inter- viewers, using videotaped interviews, and selecting eVec- tive interviewers. In particular, future research focused on interviewer training is needed. Although there is evidence that trained interviewers may be able to make more objective hiring decisions, most interviewers still do not receive much training, if any at all, before being allowed to conduct employment interviews (Howard & Ferris, 1996; Kennedy, 1994). More research is needed that explores the eVectiveness of interviewer-training methods in reducing biases.

Interviewer characteristics, besides a bias toward ethnic minorities, also should be examined in future research. For example, international experience may be correlated with more positive perceptions of applicants with diverse

S.L. Segrest Purkiss et al. / Organizational Behavior and Human Decision Processes 101 (2006) 152–167 165

characteristics, such as non-standard accents. Personality diVerences among interviewers also may be important in this research. Perhaps interviewers that rate high on the “openness to experience” dimension of the Five-Factor Model of personality are less likely to be prejudiced toward ethnic minorities, or less likely to apply these prej- udices toward particular job applicants. Openness to expe- rience or extraversion of the trainees also may be important individual diVerence variables related to the eVectiveness of interviewer training.

Conclusion

This study oVers several notable contributions to the research literature. First, because it allows for excellent control in experimental conditions, the matched-guise

technique was employed in the present study. Although this technique has not been widely used in management research to date, perhaps the present study will encour- age researchers to consider using this type of methodo- logical approach in studying organizational phenomena. Second, this research separates the eVects of accent from the eVects of ethnicity cues. Previous ethnicity research generally has failed to separate the confounding factors of accent and ethnicity, factors that appear to have interactive eVects. The results of the present study indicate that, to a degree, interviewers are still allowing illegal and often irrelevant factors, such as the combined eVects of ethnicity and accent, to aVect judgments and decisions about job applicants, instead of focusing only on job-related qualiWcations. In essence, we are still judging the book by the cover rather than solely by the contents.

Appendix A

Please indicate the degree to which you disagree or agree with each of the following statements by circling the appropri- ate number

Modern Ethnicity Bias Scale (Adapted from McConahay’s Modern Racism Scale, 1986). Items 1–7 are adapted from the original scale and items 6–12 are additions. Questions 2, 10, and 12 are reverse-coded.

1. Over the past few years, the government and news media have shown more respect to Hispanics than they deserve. Strongly Disagree 1 2 3 4 5 6 7 Strongly Agree

2. It is easy to understand the frustration of Hispanics in America. Strongly Disagree 1 2 3 4 5 6 7 Strongly Agree

3. Discrimination against Hispanics is no longer a problem in the United States. Strongly Disagree 1 2 3 4 5 6 7 Strongly Agree

4. Over the past few years, Hispanics have gotten more economically than they deserve. Strongly Disagree 1 2 3 4 5 6 7 Strongly Agree

5. Hispanics have more inXuence upon school language issues than they ought to have. Strongly Disagree 1 2 3 4 5 6 7 Strongly Agree

6. Hispanics are getting too demanding in their push for the usage of the Spanish language. Strongly Disagree 1 2 3 4 5 6 7 Strongly Agree

7. Hispanics should not push themselves where they are not wanted. Strongly Disagree 1 2 3 4 5 6 7 Strongly Agree

8. Hispanics are taking advantage of their minority status. Strongly Disagree 1 2 3 4 5 6 7 Strongly Agree

9. Hispanics are taking too many jobs from non-minorities. Strongly Disagree 1 2 3 4 5 6 7 Strongly Agree

10. Migrant farm-workers have been treated poorly in many instances. Strongly Disagree 1 2 3 4 5 6 7 Strongly Agree

11. Hispanics often intentionally exclude non-Spanish speakers in their conversations. Strongly Disagree 1 2 3 4 5 6 7 Strongly Agree

12. Mexicans crossing the US border are often dealt with too harshly. Strongly Disagree 1 2 3 4 5 6 7 Strongly Agree

166 S.L. Segrest Purkiss et al. / Organizational Behavior and Human Decision Processes 101 (2006) 152–167

References

Alwin, D. F., & Hauser, R. M. (1975). The decomposition of eVects in path analysis. American Sociological Review, 40, 37–47.

Atkins, C. P. (1993). Do employment recruiters discriminate on the basis of nonstandard dialect? Journal of Employment Counseling, 30, 108–118.

Barr, S. H., & Hitt, M. A. (1986). A comparison of selection decision models in manager versus student samples. Personnel Psychology, 39, 599–617.

BolaY, G., Braham, P., Gindro, S., Tentori, T., & Bracalenti, R. (2003). Dictionary of Race, Ethnicity and Culture. Thousand Oaks, CA: Sage Publications.

Brennan, E. M., & Brennan, J. S. (1981). Accent scaling and language attitudes: reactions to Mexian American English speech. Language and Speech, 24, 207–221.

Callan, V. J., Gallois, C., & Forbes, P. A. (1983). Ethnic stereotypes: Australian and Southern European youth. Journal of Social Psy- chology, 119, 287–288.

Cargile, A. C. (2000). Evaluations of employment suitability: does accent always matter? Journal of Employment Counseling, 37, 165.

Cohen, J., & Cohen, P. (1983). Applied multiple regression/correlation analysis for the behavioral sciences (2nd ed.). Hillsdale, NJ: Law- rence Erlbaum.

DeShields, O. W., Kara, A., & Kaynak, E. (1996). Source eVects in purchase decisions: the impact of physical attractiveness and accent of salesperson. International Journal of Research in Marketing, 13, 89–101.

Dipboye, R. L. (1994). Structured and unstructured selection inter- views: beyond the job-Wt model. In G. R. Ferris (Ed.), Research in personnel and human resources management (Vol. 12, pp. 79–123). Greenwich, CT: JAI Press.

Dipboye, R. L. (1992). Selection interviews: Process perspectives. Cin- cinnati, OH: South-Western.

Dovidio, J. F. (2001). On the nature of contemporary prejudice: the third wave. Journal of Social Issues, 57, 829–849.

Dovidio, J. F., Brigham, J. C., Johnson, B. T., & Gaertner, S. L. (1996). Stereotyping, prejudice, and discrimination: another look. In C. N. Macrae, C. Stangor, & M. Hewstone (Eds.), Stereotypes and stereo- typing (pp. 276–319). New York: The Guilford Press.

Dovidio, J. F., & Gaertner, S. L. (2000). Aversive racism and selection decisions: 1989 and 1999. Psychological Science, 11, 319–323.

Duncan, B. L. (1976). DiVerential social perception and attribution of intergroup violence: testing the lower limits of stereotyping of blacks. Journal of Personality and Social Psychology, 34, 590–598.

Eder, R. W., & Harris, M. M. (1999). Employment interview research: historical update and introduction. In R. W. Eder & M. M. Harris (Eds.), The employment interview handbook (pp. 1–27). Thousand Oaks, CA: Sage Publications.

Ely, R. J. (1995). The power in demography: women’s social construc- tions of gender identity at work. Academy of Management Journal, 38, 589–703.

Ferris, G. R., & Judge, T. A. (1991). Personnel/human resources man- agement: a political inXuence perspective. Journal of Management, 17, 447–488.

Ferris, G. R., Hochwarter, W. A., Douglas, C., Blass, F. R., Kolodinsky, R., & Treadway, D. C. (2002). Social inXuence processes in organiza- tions and human resources systems. In G. R. Ferris & J. J. Martoc- chio (Eds.), Research in personnel and human resources management (Vol. 21, pp. 65–127). Oxford, UK: JAI Press/Elsevier Science.

Ford, T. E., & Stangor, C. (1992). The role of diagnosticity in stereo- type formation: perceiving group means and variances. Journal of Personality and Social Psychology, 63, 356–367.

Giles, H., & Bourhis, R. Y. (1976). Methodological issues in dialect per- ception: a social psychological perspective. Anthropological Lin- guistics, 19, 294–304.

Giles, H., Williams, A., Mackie, D. M., & Rosselli, F. (1995). Reactions to Anglo- and Hispanic-American-accented speakers: aVect, iden-

tity, persuasion, and the English-only controversy. Language and Communication, 15, 107.

Gilmore, D. C., Stevens, C. K., Harrell-Cook, G., & Ferris, G. R. (1999). Impression management tactics. In R. W. Eder & M. M. Harris (Eds.), The employment interview handbook (pp. 321–336). Thou- sand Oaks, CA: Sage Publications.

Grow, B. (2004). Hispanic nation: is America ready? Business Week, 59–70.

Hinds, P. J., Carley, K. M., Krackhardt, D., & Wholey, D. (2000). Choosing work group members: balancing similarity, competence, and familiarity. Organizational Behavior and Human Decision Pro- cesses, 81, 226–251.

Howard, J. L., & Ferris, G. R. (1996). The employment interview con- text: social and situational inXuences on interviewer decisions. Journal of Applied Social Psychology, 26, 112–136.

Jones, J. M. (1986). Racism: a cultural analysis of the problem. In J. F. Dovidio & S. L. Gaertner (Eds.), Prejudice, discrimination, and rac- ism (pp. 279–314). Orlando, FL: Academic Press.

Judge, T. A., & Ferris, G. R. (1992). The elusive criterion of Wt in human resources staYng decisions. Human Resource Planning, 15, 47–67.

Jussim, L., Coleman, L. M., & Lerch, L. (1987). The nature of stereo- types: a comparison and integration of three theories. Journal of Personality and Social Psychology, 52, 536–546.

Kalin, R., & Rayko, D. S. (1978). Discrimination in evaluative judg- ments against foreign-accented job candidates. Psychological Reports, 43, 1203–1209.

Kawakami, K., Dion, K. L., & Dovidio, J. F. (1998). Racial prejudice and stereotype activation. Personality and Social Psychology Bulle- tin, 24, 407–416.

Kennedy, R. B. (1994). The employment interview. Journal of Employ- ment Counseling, 31, 110–114.

Kenney, G. M., & Wissoker, D. A. (1994). An analysis of the correlates of discrimination facing young Hispanic job-seekers. American Economic Review, 84, 674–683.

Larkin, J. C., & Pines, H. A. (1979). No fat persons need apply: experi- mental studies of the overweight stereotype and hiring preference. Sociology of Work and Occupations, 60, 312–327.

Lewis, A. C., & Sherman, S. J. (2003). Hiring you makes me look bad: social-identity based reversals of the in-group favoritism eVect. Orga- nizational Behavior and Human Decision Processes, 90, 262–276.

Lin, T. R., Dobbins, G. H., & Farh, J. L. (1992). A Weld study of race and age similarity eVects on interview ratings in conventional and situational interviews. Journal of Applied Psychology, 77, 363–371.

Lippi-Green, R. (1994). Accent, standard language ideology, and dis- criminatory pretext in the courts. Language in Society, 23, 163–198.

Lippi-Green, R. (1997). English with an accent: Language, ideology, and discrimination in the United States. London: Routledge.

McConahay, J. B. (1986). Modern racism, ambivalence, and the Modern Racism Scale. In J. F. Dovidio & S. L. Gaertner (Eds.), Prejudice, dis- crimination, and racism (pp. 91–125). Orlando: Academic Press.

McFarland, L.A., Sacco, J.M., Ryan, A.M., & Kriska, S.D. (2000). Racial similarity and composition eVects on structured panel inter- view ratings. Paper presented at the 15th Annual Conference of the Society for Industrial and Organizational Psychology, New Orleans.

Mosisa, A. T. (2002). The role of foreign-born workers in the U.S. economy. Monthly Labor Review, 125, 3–14.

Mullins, T. W. (1982). Interviewer decisions as a function of applicant race, applicant quality and interviewer prejudice. Personnel Psy- chology, 35, 163–174.

Nesdale, A. R., & Rooney, R. (1990). Evaluations and stereotyping. Australian Journal of Psychology, 42, 309–319.

Nunnally, J. C. (1978). Psychometric theory (2nd ed.). New York: McGraw-Hill.

Pingitore, R., Dugoni, B. L., Tindale, R. S., & Spring, B. (1994). Bias against overweight job applicants in simulated employment inter- view. Journal of Applied Psychology, 79, 909–917.

S.L. Segrest Purkiss et al. / Organizational Behavior and Human Decision Processes 101 (2006) 152–167 167

Podberesky, R., Deluty, R. H., & Feldstein, S. (1990). Evaluations of Spanish- and Oriental-accented English speakers. Social Behavior and Personality, 18, 53–63.

Posthuma, R. A., Morgeson, F. P., & Campion, M. A. (2002). Beyond employment interview validity: a comprehensive narrative review of recent research and trends over time. Personnel Psychology, 55, 1–81.

Prewett-Livingston, A. J., Feild, H. S., Veres, J. G., & Lewis, P. M. (1996). EVects of race on interviewer ratings in a situational panel interview. Journal of Applied Psychology, 81, 178–186.

Roehling, M. V., Campion, J. E., & Arvey, R. D. (1999). Unfair discrim- ination issues. In R. W. Eder & M. M. Harris (Eds.), The employ- ment interview handbook (pp. 49–67). Thousand Oaks, CA: Sage Publications.

Ryan, E. B., & Carranza, M. A. (1975). Evaluative reactions of adoles- cents toward speakers of standard English and Mexican-American accented English. Journal of Personality and Social Psychology, 31, 855–863.

Ryan, E. B., Carranza, M. A., & MoYe, R. W. (1977). Reactions toward varying degrees of accentedness in the speech of Spanish–English bilinguals. Language and Speech, 20, 267–273.

Sacco, J. M., Scheu, C. R., Ryan, A. M., & Schmitt, N. (2003). An inves- tigation of race and sex similarity eVects in interviews: a multilevel approach to relational demography. Journal of Applied Psychology, 88, 852–865.

Sanchez, J. I., & Brock, P. (1996). Outcomes of perceived discrimina- tion among Hispanic employees: is diversity management a luxury or a necessity? Academy of Management Journal, 39, 704–719.

Singer, M., & Eder, G. S. (1989). EVects of ethnicity, accent, and job status on selection decisions. International Journal of Psychology, 24, 13–34.

Slavin, L. A., Rainer, K. L., McCreary, M. L., & Gowda, K. K. (1991). Toward a multicultural model of the stress process. Journal of Counseling and Development, 70, 156–163.

Snyder, M., Tanke, E. D., & Berscheid, E. (1977). Social perception and interpersonal behavior: on the self-fulWlling nature of social stereotypes. Journal of Personality and Social Psychology, 35, 656–666.

Stangor, C., & McMillan, D. (1992). Memory for expectancy-congru- ent and expectancy-incongruent information: a review of the social and social developmental literatures. Psychological Bulletin, 111, 42–61.

Strahan, R., & Gerbasi, K. (1972). Short, homogeneous versions of the Crowne–Marlowe Social Desirability Scale. Journal of Clinical Psy- chology, 28, 191–193.

Taylor, S. E., & Thompson, S. C. (1982). Stalking the elusive “vivid- ness” eVect. Psychological Review, 89, 155–181.

U.S. Department of Labor, Employment and Training Administration, U.S. Employment Service, (1991). Dictionary of occupational titles (4th edition revised). Washington, DC, U.S. Government Printing OYce.

Vrij, A., & Winkel, F. W. (1994). Perceptual distortions in cross-cul- tural interrogations: the impact of skin color, accent, speech style, and spoken Xuency on impression formation. Journal of Cross-Cul- tural Psychology, 25, 284–295.

Wheeler, A. R., Buckley, M. R., Halbesleben, J. R., Brouer, R. L., & Ferris, G. R. (2005). The elusive criterion of Wt” revisited: toward an integrative theory of multidimensional Wt. In J. J. Martocchio (Ed.), Research in personnel and human resources management (Vol. 24, pp. 265–304). Oxford, UK: JAI Press/Elsevier Science.