Utilization and validity of nonverbal cues in the structured interview

MISSING IMAGE

Material Information

Title:
Utilization and validity of nonverbal cues in the structured interview
Physical Description:
ix, 195 leaves : ill. ; 29 cm.
Language:
English
Creator:
Burnett, Jennifer R
Publication Date:

Subjects

Subjects / Keywords:
Interviewing -- Evaluation -- United States   ( lcsh )
Nonverbal communication   ( lcsh )
Genre:
bibliography   ( marcgt )
theses   ( marcgt )
non-fiction   ( marcgt )

Notes

Thesis:
Thesis (Ph. D.)--University of Florida, 1993.
Bibliography:
Includes bibliographical references (leaves 186-194).
Statement of Responsibility:
by Jennifer R. Burnett.
General Note:
Typescript.
General Note:
Vita.

Record Information

Source Institution:
University of Florida
Rights Management:
All applicable rights reserved by the source institution and holding location.
Resource Identifier:
oclc - 30608397
ocm30608397
System ID:
AA00011147:00001

Full Text












UTILIZATION AND VALIDITY OF NONVERBAL CUES
IN THE STRUCTURED INTERVIEW

















BY


JENNIFER R. BURNETT


A DISSERTATION PRESENTED TO THE GRADUATE SCHOOL
OF THE UNIVERSITY OF FLORIDA IN PARTIAL FULFILLMENT
OF THE REQUIREMENTS FOR THE DEGREE OF
DOCTOR OF PHILOSOPHY



UNIVERSITY OF FLORIDA

1993




~--



























Copyright 1993

by

Jennifer R. Burnett






























This work is dedicated to my supportive, patient, and ever

loving husband, Tom, who has helped me become all that I am,

and all that I will ever be; to my wonderful and caring

parents, who always believed in me and who instilled in me

the ambition and strength to reach for my goals; and to my

two brothers for their constant encouragement and

inspiration.














ACKNOWLEDGEMENTS


Special thanks are extended to my chairman, Dr. Stephan

Motowidlo, for his enduring guidance. His confidence in the

results and related implications of this line of research has

been very inspiring. As a committed mentor, he made the

dissertation process a true learning process. I am also

sincerely grateful for the support of Dr. Henry Tosi. His

belief in my abilities, constant encouragement, and

invaluable advice helped me to reach my goals. Dr. Norman

Markel was also a integral part of my learning experience.

He not only offered his expertise in the field of nonverbal

communication, but he also provided a much appreciated,

realistic perspective to this experience. Finally, I would

like to acknowledge the contribution of Dr. Wesley

Hutchinson. His suggestions and viewpoint were greatly

appreciated.

The financial and logistical support of Dr. Sanford Berg

and the Public Utility Research Center at the University of

Florida was greatly appreciated. His assistance made this

research possible.

A special acknowledgement is given to my parents and

brothers. Their faith in me was unwavering, and their

excitement in my accomplishments has made it all worthwhile.









Finally, none of this would have been possible without my

husband, Tom. The long hours, doubts, and frequent obstacles

could not have been overcome without his undying patience,

understanding and strong belief in me. My accomplishments

are a direct result of his love and encouragement.















TABLE OF CONTENTS



ACKNOWLEDGEMENTS ............................................ iv

ABSTRACT ................................................. viii

INTRODUCTION AND LITERATURE REVIEW ......................... 1

Introduction..............................................1
Structured Interviews..................................... 2
Theoretical Perspectives on the Effectiveness of the
Structured Interview .................................13
Attention to and Utilization of Nonverbal Cues in the
Interview ........................ ....................24
Issues of Concern and Future Directions...................46

PRELIMINARY STUDIES AND THE CURRENT STUDY RESEARCH
QUESTIONS................................................59

Study #1: The Development and Validation of the
Interview .............................................60
Study #2: Aural and Visual Sources of Validity.......... 66
Study #3: Identifying Important Nonverbal Behaviors..... 71
Implications of Preliminary Studies on the Current
Study ..................................................81
Current Research Study...................................84

HYPOTHESES, METHODS, AND PROCEDURES .......................88

Research Hypotheses ............................................88
Methods and Procedures ..................................92
Reliabilities .. .......................................105
Data Analysis........................................... 113

RESULTS ....................................................115

Total Sample Analyses.................................... 115
Analyses by Interviewee Sex.............................121

DISCUSSION AND CONCLUSIONS ................................130

Cue Utilization and the Ecological Validity of
Nonverbal Cues ........................................130
Incremental Validity.....................................133
Sex Differences. ........................................ 134









Contributions and Limitations.......................... 138
Summary and Conclusions................................ 140

APPENDIX A INTERVIEW PROTOCOL AND INTERVIEW DIMENSIONS .... 142

APPENDIX B SUPERVISORY RATING FORMS ......................156

APPENDIX C TRAITS ASSOCIATED WITH BEHAVIORAL DIMENSIONS ...164

APPENDIX D NONVERBAL CUE RATING FORMS .....................169

REFERENCES ...................................................186

BIOGRAPHICAL SKETCH ....................................... 195


vii














Abstract of Dissertation Presented to the Graduate School
of the University of Florida in Partial Fulfillment of the
Requirements for the Degree of Doctor of Philosophy

UTILIZATION AND VALIDITY OF NONVERBAL CUES
IN THE STRUCTURED INTERVIEW


By

Jennifer R. Burnett

August, 1993




Chairman: Stephan J. Motowidlo, Ph.D.
Major Department: Management


The purpose of this study was to determine if nonverbal

cues are utilized by interview raters, are related to

performance ratings, and contribute incrementally to the

validity of a structured interview. Simulated job interviews

with 60 managers from 4 utility companies were recorded on

videotape. Supervisors provided performance ratings.

Undergraduate raters (N=167) watched the videotaped

interviews and rated subjects on 4 dimensions of managerial

effectiveness. Eight nonverbal behaviors, including gaze,

smiling, hand movement, back/side lean, body orientation,

physical attractiveness, dress, and vocal attractiveness,

were scored by independent judges.

Nonverbal cues, in combination, were significantly

correlated with both interview (r=.33, p<.01) and performance


viii









(r=.21, p..05) ratings. Gaze was the only nonverbal cue

which was not only utilized by interview raters, but also was

a valid indicator of performance. Independent measures of

nonverbal cues did not contribute incrementally to the

validity of the interview, suggesting that interview raters

were utilizing cues sufficiently.

Results differed according to the sex of the

interviewee. Although nonverbal cues correlated

significantly with interview ratings for both men (r=.30,

p.05, N=32) and women (r=.39, p<.05, N=28), correlations

with performance substantially differed by sex (men: r=-.18,

n.s.; women: r=.67, p<.001). For women only, combined

dynamic cues were both utilized by interview raters (r=.31,

p.05), and significantly correlated with supervisory ratings

of performance (r=.56, p.001). Also, for women,

independently rated nonverbal cues added significantly to the

prediction of performance, beyond that of interview ratings

alone. This indicates that interview raters were not

sufficiently utilizing nonverbal cues when judging female

interviewees.

It was concluded that nonverbal cues influence

structured interview ratings, and are related to supervisory

ratings of performance. Gaze, in particular, seems to be a

valid nonverbal cue to utilize in the interview. Three

explanations are proposed regarding the compelling sex

differences found in this study.














INTRODUCTION AND LITERATURE REVIEW


Introduction


Despite years of criticism about the interview's low

reliability and validity, it has prevailed as one of the most

popular methods of selection. Recently developed structured

interview techniques have lifted some of the negativity

surrounding the use of the interview. Studies that have

tested structured interview formats have provided favorable

reports of their use to increase the validity and reliability

of interviewer's judgments (M. Campion, Pursell, and Brown,

1988; Janz, 1982; Latham, Saari, Pursell, and M. Campion,

1980; Motowidlo, Carter, Dunnette, Tippins, Werner, Burnett,

and Vaughan, 1992). Researchers must continue to investigate

how to improve the structured interview in order to obtain

accurate ratings that identify qualified job candidates.

This study will focus on a set of factors that, research

has shown, affects the favorability of interviewer's

judgments. These factors are nonverbal cues. The specific

focus here is on the effect of nonverbal cues (e.g. body

positioning, gestures, facial expressions, voice

characteristics, appearance and attractiveness) on the

validity with which interviewers make predictions of

applicants' future job performance.









In order to establish the direction and contribution of

the current study, related literature will be reviewed.

First, an overview of structured interviewing will be

presented, including an explanation of what is meant by

"structured," and a description of different types of

structured interview techniques. Second, since it has been

suggested that the structure of the interview may affect

interviewers' ability to process information presented during

the interview, a closer look will be taken of each step of

the information processing system and how the structured

interview format can affect those steps. The step of

greatest interest in the current study is the process of

attending to and utilizing cues, particularly nonverbal cues.

This focus will lead to a review of studies that have

investigated the effect of nonverbal cues on raters'

judgments. What types of cues are important, how do they

affect interviewers' judgments, and to what degree do they

influence judgments beyond that of verbal information?

Finally, important issues which need to be addressed by

further research in this area will be discussed.


Structured Interviews



The Meaning of "Structure"


Structuring an interview may mean different things to

different people. Structure, in the context of the selection









interview, means that the interviewer follows systematic and

predetermined rules to observe and evaluate applicants.

Structuring may also mean that these rules are applied in the

same way for all applicants of a position. These rules may

designate the type of questions asked by the interviewer, the

elements of an applicant's answer to which the interviewer

should attend, and/or the scoring system used by the

interviewers to make judgments based on the applicant's

responses. We can think of structure in varying degrees. At

the low extreme there is the unstructured interview where no,

or few, rules exist with regard to conducting the interview,

and the interviewer has full discretionary control of the

process. At the high extreme there is the very structured

interview, where the interviewer only asks predetermined

questions and nothing more, and therefore has no discretion.

The highly structured interview is, in effect, an orally

administered objective test, and could just as well be

administered without the interviewer. Although we do not

know if there is an optimal degree of structure, there is

evidence that some structure is better than none.

For example, in 1949 Wagner noted that an interview

should be conducted according to a standardized form. He

proposed that an interview can be valid if all information

that can be obtained on a candidate is taken into

consideration and weighed properly. Furthermore, the

interviewer must be skilled at obtaining full and complete

information from an applicant, observing significant









behavior, ignoring irrelevant behavior, and synthesizing all

information into a valid prediction of success (Wagner,

1949). In fact, Wagner noted that another researcher,

McMurray (1947), developed a "patterned" interview that

accomplished these objectives. Using McMurray's format,

interviewers (1) were trained in the techniques of

interviewing, (2) attended to definite job specifications,

(3) had a plan and knew before the interview what questions

they would ask of applicants, (4) had clinical concepts to

use when interpreting and evaluating applicant's responses,

(5) checked outside references prior to the interview, and

(6) were assessed for their own degree of emotional

adjustment and intelligence. McMurray found a significant

correlation (r=.68) between interviewers' ratings of 587

factory workers and supervisor's evaluations of their work.

In another study he found a correlation of .43 between

interview ratings and truck drivers' tenure on the job.

These suggestions by Wagner and the early findings by

McMurray (1947) sound very similar to the rules and formats

used, and results reported, of many structured interview

techniques of today.

Empirical evidence of the advantage of more structure

was presented more recently by Wiesner and Cronshaw (1988)

who reported an uncorrected validity coefficient of .35 for

structured interviews, but only .11 for unstructured

interviews in a meta-analysis of validity studies of

selection interviews. Accordingly, the interest and





5



popularity of the structured interview is based on both long

held beliefs (Mayfield, 1964; McMurray, 1927; Schmitt, 1976;

Ulrich and Trumbo, 1965; Wagner, 1949; Wright, 1969) and

recent evidence (Hakel, 1989; Janz, 1982; Latham, Saari,

Pursell, and M. Campion, 1980; Motowidlo, Carter, Dunnette,

Tippins, Werner, Burnett, and Vaughan, 1992; Wiesner and

Cronshaw, 1988) that structured interviews have an advantage

over unstructured interviews.


Types of Structured Interview Formats


Two well known structured interview formats are the

Situational Interview (SI), developed by Latham et al.

(1980), and the Patterned Behavioral Description Interview

(PBDI), developed by Janz (1982; Janz, Hellervik, and

Gilmore, 1986). Another, more recently developed, structured

interview format is the Structured Behavioral Interview,

developed by Motowidlo et al. (1992). Similarities and

differences in format of each of these interview techniques

and evidence of the validity and reliability of each type of

structured interview provide insight into the advantages of

structured interviews over traditional interviews.

First, in the situational interview (SI), questions are

presented as hypothetical situations, and ask applicants what

they would do if they were in that situation. The SI

questions are based on goal theory, or the belief that

behavioral intentions are strongly linked with subsequent









behavior. SI questions are derived from a critical incidents

job analysis of the job to be filled.

The validity of the situational interview has received

support. Latham et al. (1980) reported three studies with

concurrent validity coefficients of .46 (p<.05, N=49) and .30

(p<.05, N=63) with incumbents, and predictive validities of

.33 (p<.05, N=56) with applicants. The criterion in all

cases was supervisor's composite performance ratings of

workers. Two additional studies by Latham and Saari (1984)

also support the validity of the situational interview. In

one study, they reported correlations of interview ratings

with concurrent supervisory ratings at .39 (p<.05) and with

peer ratings at .42 (p<.05)(N=29). In another study, they

reported a predictive validity coefficient of .14 (p<.05,

N=157) between interview scores and composite supervisory

ratings. Weekley and Gier (1987) gave evidence of the

predictive validity of the situational format when they

reported a correlation of .45 (p<.05) between interview

ratings of 54 individuals applying for a retail sales

position and their sales productivity nine months later.

Across six studies, validity estimates for the situational

interview ranged from .14 to .46 (all significant at the

p<.05 level) with a total sample of 378 interviewees. This

results in a mean validity of .28 weighted by sample size.

Reliability estimates from these same seven studies are

adequate. Interobserver reliabilities ranged from .76 to .90,

and internal consistency estimates ranged from .61 to .78.









In the second well-known structured interview format,

the Patterned Behavior Description Interview (PBDI),

applicants are asked to recall specific past experiences, and

to report their actual behavior when they were in those

situations. This format is based on the theory of behavioral

consistency, or the idea that past behavior will predict

future behavior (Wernimont and Campbell, 1968). Janz (1989)

states that behavior descriptions provided by job applicants

reveal specific choices made by those individuals, and the

circumstances under which they made those choices. He argues

that this type of information reveals behavioral patterns,

rather than opinions or impressions, and the more long-

standing the applicant's behavior pattern, the more likely it

will predict future behavior. The behavioral dimensions

tapped by the PBDI questions are derived from the results of

a critical incidents job analysis.

There have been two published studies of the validity of

the PBDI. The first, by Janz (1982), involved fifteen

teaching assistants who were interviewed twice by

undergraduate students. Interviewers were divided into two

groups. One group received training for the PBDI, and the

other group received general instruction on establishing

rapport, active listening, etc. The traditionally trained

interviewers exhibited greater interrater agreement, but the

PBDI-trained interviewers' ratings were significantly

correlated, at .54 (p .001), with an independent rating of

the teaching assistants' effectiveness in class. The









traditionally trained interviewers' ratings only correlated

.07 (n.s.) with the criterion. It is important to note that

in Janz's study interviewers were not randomly assigned to a

training condition, and the sample size was small. These two

weaknesses cast speculation on the accuracy of these results.

In response, in a study by Orpen (1985), interviewers

were randomly assigned to two conditions of behavior

description training and traditional training. Nineteen real

applicants for an insurance sales position were each

interviewed by at least four of the sixteen total

interviewers. Supervisory ratings and the value of insurance

sales in one year were both used as criteria. Orpen (1985)

reported a correlation of .48 (p<.01) between supervisory

performance ratings and the ratings made by interviewers

trained in the behavioral description technique, versus a

nonsignificant validity coefficient of .08 (n.s.) for the

traditionally trained interviewers. For the criterion of the

value of insurance sold, the PBDI interviewers' ratings

validity was .61 (p<.01), and the traditional interviewers'

ratings validity was .10 (n.s.).

The mean validity coefficient for the Janz (1982) and

Orpen (1985) studies is .55, weighted by sample size.

However, the small total sample size (N=34), again, calls

into questions the precision of these results. The test-

retest reliability in Orpen's study was .72 and did not

differ significantly from that of the traditional

unstructured interview. In Janz's study the standard









interview had a greater interobserver reliability than that

of the patterned behavior interview, .71 and .46,

respectively.

Lastly, the Structured Behavioral Interview (SBI),

developed by Motowidlo et al. (1992), consolidates some

aspects of both the SI and PBDI techniques, while adding some

unique features. The SBI is similar to both the SI and PBDI

in that it derives its behavioral dimensions from the results

of critical incident job analysis. The questions in the SBI

elicit answers representing one or more of the interview

dimensions by asking how interviewees behaved in past

situations that are similar to situations encountered on the

job for which they are applying. Therefore, the SBI

questions are similar to, but more structured than the PBDI,

since this format requires interviewers to ask the same

questions. Additionally, the interviewer using the SBI not

only asks a standard set of questions, but also is trained to

probe for details of the situation, the interviewee's

behavior in that situation, and the outcome that resulted

from that behavior. Like the SI, after the interview is

completed, the interviewer using the SBI rates the

interviewee on behaviorally anchored scales for each

dimension.

Motowidlo and his colleagues (1992) reported the SBI's

criterion validity, construct validity, and reliability. In

three studies using eight telecommunications companies, with

a total validation sample of approximately 500 interviewees,









the uncorrected mean estimate of the criterion-related

validity of the SBI is .22. In one study they reported

concurrent validities ranging from .15 (p<.05) to .32

(p<.01), with a mean of .25 weighted by sample size varying

from 139-146, depending on missing data. In another study,

predictive validity coefficients were reported between .13

(p<.05) and .25 (p<.01), with a mean of .18, weighted by a

sample size of 192-195. In a third study, they reported a

predictive validity coefficient of .15 (n.s.) to .45(p<.01),

with a mean correlation of .33, weighted by a sample size of

30-33. These three studies involved an interview format that

measured dimensions important for management jobs. In the

third study, however, another format designed for marketing

jobs, was also tested. The concurrent validity of the

marketing position interview ranged from .05 (n.s.) to .32

(p<.01), with a mean of .20, weighted by a sample size of

124-159.

Support for the construct validity of the SBI was based

on three results. First, correlations between ratings by

interviewers and subjects who listened to tapes of the

interviews, and ratings by interviewers and supervisors, were

higher for same dimensions (i.e. convergent validity) than

for different dimensions (i.e. discriminant validity).

Second, SBI ratings were significantly correlated with

ratings made with another interview format designed to

measure the same, or similar, dimensions. And, finally, the

SBI interview ratings were not strongly, or significantly,









correlated with other measures, such as an aptitude test,

college grade point averages, or rank in college class, that

rely heavily on cognitive ability.

The reliability of the SBI was determined in the first

study conducted by Motowidlo et al. (1992) and was based on

(1) the correlation between ratings by different interviewers

who interviewed the same applicants on different occasions,

and (2) the correlation between ratings by pairs of listeners

of audiotapes of those interviews. Those reliability

estimates were .64 and .63, respectively.

In addition to evidence of validity and reliability of

these three particular interview techniques, there are two

additional criteria that should be taken into account--

namely, freedom from bias and practicality. For the

situational interview (Campion et al., 1988; Latham, 1989)

and the structured behavioral interview (Motowidlo, et al.,

1992) there is significant evidence that these techniques are

not biased in favor of race or sex of the applicant. There

is no reported evidence regarding the degree of bias in the

patterned behavior description interview (Janz, 1982; Janz,

1989).

The practicality of these three interview techniques is

evidenced by indicators such as their usefulness for making

selection decisions, interviewers' and applicants'

perceptions of fairness, and the feasibility of their use.

The most apparent test of an interview's practicality may be









its continued use and acceptance by recruiters and managers

who use the technique to make important hiring decisions.

In one study, by Latham and Finnegan (1987),

interviewers and applicants were asked to report their

perceptions of and preferences for the unstructured,

patterned, and situational interview formats. Two groups of

managers reported that they preferred the patterned to the

unstructured interview. They did, however, consider the

situational interview significantly better than the patterned

on aspects such as the opportunity to compare applicants

objectively, and to hire or reject applicants on job related

grounds. The SI, however, was perceived as low on ease of

preparation. Employees who were recently hired using the

three formats did not view one interview method as better, or

more preferable, than another. College students, however,

who were preparing for a job search, preferred the

unstructured interview over the other two formats, because

the unstructured format allowed them to sell themselves more

effectively, and gave them more freedom to say what they

wanted to say.

In summary, for users, in this case, managers, the

consensus seems to be in favor of some structure, rather than

none. But the jury is still out regarding what format

applicants prefer, or whether one particular structured

format is more preferable or practical than another to users.

Because the SBI was not included in Latham and

Finnegan's (1987) study since it had not yet been developed,









there is a need to conduct a more current survey of users'

and applicants' perceptions of the practicality of each

interview method for meeting their objectives. Although, if

the popularity and widespread use of the structured,

behaviorally-based interviews in corporate America is any

indication of the practicality of these techniques, then it

seems reasonable to argue that they are indeed perceived as

useful, feasible, and fair.

Convincing evidence has been presented of not only the

overall favorability of structured interviews, but also the

results of validity and reliability research of three

specific structured interview formats. Additionally, support

has been given for the SI and SBI's freedom from bias.

Finally, the structured techniques seem more practical than

the unstructured interview to interviewers, but evidence of

overall practicality is still officially undetermined. One

question not yet addressed, however, with regard to the

success of the structured interview, is "Why does it work?".

In order to try to answer this question theoretical

perspectives of interviewing itself must be considered.


Theoretical Perspectives on the Effectiveness of the
Structured Interview



The Validation Model


In order to discuss the possible reasons why structured

interviews are more valid and reliable than the traditional,









unstructured interview it must be explained why the latter

have had poorer results. According to a model developed by

Schwab (1980), when considering the criterion-related

validity (or empirical validity) of the interview, it is also

necessary to be concerned with the construct validity of the

predictor and the criterion used (See Figure 1.1).





Independent Dependent
Variable Variable


Construct
Validity

I' -- CI

Empirical Validity

Figure 1.1. An illustration of construct and empirical
validity in the employment interview.(Adapted from Schwab,
1980)



The validity correlations reported by researchers are between

interview ratings (I') and some type of job performance

criterion rating (C'). These are indicators of the real

constructs of I, performance in the interview, and C,

performance on the job. When a significant correlation is

found between I' and C' it suggests a significant

relationship between I and C. But, there must also be a high

correlation between I and I' and C and C' to make that

assumption. Therefore, both construct and empirical validity









are necessary to make inferences from the research data

(Cronshaw and Wiesner, 1989).

In the unstructured interview, the indicator I' (i.e.

interview ratings) may lack construct validity as a measure

of I in the structured interview because of the failure of

the traditional interviewer to obtain relevant, job-related

information. Janz (1989) suggested that, in the unstructured

interview, the questions interviewers actually ask, and the

information for which they probe, have little to do with the

constructs that are important for predicting job performance.

Necessary and relevant information may not be consistently

uncovered in an unstructured interview (Mayfield, 1964).

Also, I' may lack construct validity as a measure of I

because of the influence of interviewer biases (such as

similar-to-me and contrast effects). Interviewers who use

unstructured interviews have much self discretion, thereby

allowing individual differences, such as differential

attention and weighting of information, to come into play.

These biases and individual differences can interfere with

the collection of a relevant sample of the true domain of

information about the applicant (I). Likewise, interviewers

who are not guided by any specific rules or structure tend to

focus on how the applicant responds in the limited context of

the interview session, rather on long term patterns of

behavior. This allows the impression management strategies

of the interviewee to influence judgments to a greater degree

(Dipboye, 1992; Janz, 1989).









Therefore, the lack of construct validity of I' may

attenuate the criterion-related validity coefficient between

I' and C'. In response to the low reliabilities and low

validities of unstructured interview techniques arose the

development of structured interview formats.

There is not one widely accepted theory regarding why

structured interviews work. The advantage of the structured

interview developed by Janz (1982), Latham et al. (1980) and

Motowidlo et al. (1992) may be due to the fact that the

questions are job-related because they were systematically

derived using critical incidents job analysis. If questions

are more job-related, then applicants can provide more

relevant information in their answers for interviewers to use

for predicting effective job performance. Therefore,

structured interview ratings (I') that are based on those

relevant answers would have greater construct validity,

because they would more accurately reflect the true predictor

of I (see Figure 1.1). Similarly, the structured interview

format may have greater validity results because

interviewers' judgments are more reliable. The extensive

training typically involved in the use of structured formats,

the standardization of questions across interviewers, and the

provision of behaviorally anchored rating scales may all

contribute to increased interrater reliability. Overall, the

methodological aspects of structured formats seem to improve

the I' I linkage, resulting in improved reliability and

construct validity of the interview ratings, and subsequently









strengthen the I' C' linkage, reported as the criterion-

related validity (Cronshaw and Wiesner, 1989).

Aspects of a structured format, such as the job-

relatedness of the questions, the presence of predetermined

rules for conducting the interview, and the use of

behaviorally anchored scales to make ratings, may each

contribute to the construct and empirical validity of the

ratings. This explanation, however, may be only part of a

more comprehensive explanation of the advantages of the

structured interview. One can not forget that between the

time the interview questions are asked and the interviewer

makes a rating or judgement, an important phenomenon occurs--

the processing of information. The structure of the

interview, particularly the methodologies employed, may

improve the effectiveness of the interviewers' information

processing capabilities, that, in turn, affect the validity

of the interview (Eder, Kacmar and Ferris,1989).


The Information Processing Model


The interviewer processes information by progressing

through five steps (Lord, 1985). The first three steps

constitute the input of information, including (1) selective

attention to informational cues, (2) the comprehension and

encoding of those cues, and (3) storage and retention into

memory. After all information has been collected, at some

later point, the judge must retrieve the information (step 4)









from memory and translate it into the required judgment or

decision (step 5) (Lord, 1985).

The methodology of a structured format may affect each

step of the information processing system. First, the

interviewer must obtain a sample of information from the

population of positive and negative information available

about the job applicant (Motowidlo, 1986). This sample of

information is obtained through interviewers' automatic

and/or controlled attentional mechanisms (Ilgen and Feldman,

1983; Feldman, 1981; Schneider and Shiffrin, 1977; Shiffrin

and Schneider, 1977). The automatic process of attention

occurs without conscious monitoring or awareness. Relevant

information must be detected within the context of other,

irrelevant pieces of information. When under controlled

attentional conditions, individuals are aware of their

cognitive processes and consciously seek out relevant

applicant information. Both the automatic and controlled

attention mechanisms are influenced by the salience of the

informational cues. The salience of the cues determines

whether or not they are utilized by interviewers (Ilgen and

Feldman, 1983; Feldman, 1981; Schneider and Shiffrin, 1977;

Shiffrin and Schneider, 1977).

In the structured interview the interviewers are

typically trained to attend to information that is relevant

to the dimensions on which they will rate the applicant. For

example, in the Situational Interview (Latham et al., 1980)

raters are aware of the dimensions that each question is









intended to tap, and they are also provided with a scoring

guide that gives them examples of a good, moderate, and poor

response for each question. Motowidlo et al. (1992) also

trained interviewers using the Structured Behavioral

Interview format to identify information provided by the

applicant that was relevant to the dimension to be measured.

The interviewers made ratings on each dimension that were

anchored with specific descriptions of behaviors at high,

moderate and low levels. This training and the tools

provided to interviewers (specific dimensions, structured

questions, scoring guides) make their attentional processes

more controlled. Additionally, by providing interviewers

with carefully developed tools and predetermined rules to

follow they are relieved of the responsibility of making

those decisions themselves. They then have more time and

freedom to attend to interviewees' responses, rather than

worry about what question to ask next, or how to score it.

Overall, the controlled attention of interviewers using a

structured interview format enables them to distinguish

relevant, from irrelevant, pieces of information more

readily.

Second, the rules and tools of the structured format

make it easier for interviewers to encode the information to

which they have attended. Encoding occurs when an external

stimulus, such as applicants' verbal and nonverbal responses

to interview questions, is translated into an symbolic code

within the mind of the perceiver (Lord, 1985). Each piece of









information is not encoded independently, but instead, is

matched with more general, preexisting categories.

Information that is incongruent with existing categories

takes longer to encode, and may require the development of a

new category for novel stimuli (Lord, 1985). Ilgen and

Feldman (1983) stated that the categories that judges use to

process information are a function of their experience and

education. Accordingly, the structured interview may improve

interviewers' encoding process because they are provided with

behavioral dimensions (i.e. categories), trained to use them,

and gain more experience with using those same dimensions

over time. The dimensional scoring guides that use

behaviorally-anchored scales give them a relevant bases upon

which they can identify and categorize interviewees'

responses. Therefore, they are better able to make sense of

and integrate the wealth of information presented to them

during the interview by matching it with predetermined

behavioral categories.

After information is encoded it is stored in short-term

memory, or a "workspace" (Wyer and Srull, 1980). In many

cases of social perception, that information may not be

needed until some period of time has elapsed. In those

situations the information is eventually transferred from

short-term to long-term memory until retrieval. In the

context of the interview, however, there is very little time,

often a matter of minutes or, at most, hours, between when

the information is received and when it is retrieved for









evaluation. Therefore, we will assume that, under most

interviewing conditions, it does not enter long-term memory.

The short retention period helps to minimize the loss of

detailed information, and the interviewer is not limited to

using only the prototypes of the encoded categories, for

information output.

For information output, the interviewer must retrieve

and integrate encoded information, and then formulate a

judgment of the applicant based on that information. These

steps often occur contiguously (Lord, 1985). The information

recalled could be both behavioral and dispositional (Feldman,

1981). There has been much criticism about using traits to

make personnel decisions, rather than behaviors, due to the

subjective assessment of personality attributes and judges'

tendencies to define traits differently Instead, more

objective, behavioral information is recommended for

personnel decisions. When making hiring decisions, the

behaviors of the job candidate are of greatest interest,

since these should predict how they will perform on the job.

In the structured interview, since the scoring guides with

which interviewers make their ratings are based on specific

behaviors, their ability to recall and rate behavioral

information may be improved.

Overall, the rules and methods used in structured

interviews have the potential to influence each step of the

interviewers' information processing system. Motowidlo

(1986) noted the importance of the information processing









system for making personnel decisions and stated that "the

accuracy of an evaluative judgment depends upon how well the

input sample of impressions represents the population of

information, how well the retrieved sample of impressions

represents the input sample of impressions, and how well the

judgment represents the retrieved sample of impressions" (p.

6). If a structured format results in more efficient

processing of information by the interviewer, then the

validity of those decisions is improved.

Within the context of both an information processing

model and a validation model, we have theoretically implied

reasons for the advantage of structured interview formats

over unstructured formats. It would be infeasible, if not

impossible, to demonstrate all of these implications at once.

Instead we would like to narrow our focus for this study to

one stage of the information processing system--that of

attention and the utilization of cues. In order to do this

we turn to another model, by Brunswik (1956), that limits

itself exclusively to the process of cue utilization.


The Brunswik Lens Model


The Brunswik lens model provides an approach for

researchers to examine (1) the influence of applicant

attributes on interviewers judgments (cue utilization), (2)

the actual relationship between each attribute and the

criterion (ecological validity), and (3) the overall









relationship of interviewers' judgments of the applicant to

the criterion (achievement).


Hierarchy of Cues


Achievement


Figure 1.2. A modified version of the Brunswik lens model.



The level of achievement is a function of ecological

validity and cue utilization. In other words, the accuracy

of an interviewers' judgment varies with whether he or she

utilizes valid cues, and the consistency with which cues are

utilized (Dipboye, 1992). If an interviewer has knowledge

about which pieces of verbal and nonverbal information

presented during the interview are relevant for predicting

the criterion, and uses those cues to make hiring decisions,

then he or she has identified valid cues and is utilizing

them appropriately.









Few studies in the employment interview literature have

used the Brunswik lens model. In their 1982 review, Arvey

and Campion encouraged more use of the lens model, yet ten

years later, there has been only one known application of

this model by Gifford et al. (1985) for interview research.

Borkenau and Liebler (1992) claimed they were the second to

use the entire lens model in a study of person perception.

Although they focused on situations where targets and raters

are strangers, much like an interview, they did not

explicitly use an employment interview setting. Furthermore,

though many studies have investigated interviewers'

utilization of nonverbal cues, or variables that correspond

to only the right side of the lens model, only one study,

conducted by Anderson and Shackleton (1990), explicitly used

the Brunswik lens model. This body of literature will now be

reviewed.




Attention to and Utilization of Nonverbal Cues in the
Interview


The interview is, by definition, a face-to-face

discussion between the applicant and the interviewer, but

interviewers do not attend only to what applicants say. Much

evidence indicates that interviewers also attend to and

evaluate nonverbal cues presented by interviewees (Gifford et

al., 1985; Imada and Hakel, 1977; McGovern and Tinsley, 1978;

Raza and Carpenter, 1987; Young and Beier, 1977). Several









researchers have investigated whether those cues provide

information that influences hiring decisions, above and

beyond that provided by verbal and more objective (i.e.

resume) information. Schmitt (1976) suggested that nonverbal

sources of information were more important than verbal cues,

and a combination of both verbal and nonverbal information

had the greatest influence on differences in interviewers'

ratings of job applicants. Amabile and Kabat (1982) go so

far as to suggest that when subjects' verbal responses

conflict with their behavior, interviewers tend to base their

evaluations more on actions than words. Even when there is

no conflict, the variance accounted for by nonverbal cues has

exceeded that of verbal content in some studies. Young and

Beier (1977) found that eye contact, smiling and head

movement accounted for 87% of the variance in interviewers

ratings of applicant qualifications. Sigelman and Davis

(1978) found that nonverbal behavior accounted for 56% of the

variance in the interviewers' evaluations. In two other

studies (Argyle, Salter, Nicholson, Williams and Burgess,

1970; Walker, 1977), researchers reported that nonverbal cues

accounted for 10.3 and 10 times more variance, respectively,

than that due to verbal content.

Systematic research on the effect of nonverbal behavior

on social interaction has been conducted in only the last

thirty to thirty-five years, and specifically within the

context of the interview only in the last twenty years.

Before reviewing this body of literature different types of









nonverbal behaviors will be identified and operationally

defined.


Identifying and Defining Nonverbal Cues


There are three specific types of nonverbal cues that

have been studied as having an effect on interviewers'

employment decisions--dynamic, static and paralinguistic

cues. Dynamic cues are those that involve some type of

movement that can be changed from moment to moment. Examples

of dynamic cues include smiling, gesturing, gaze, head

movement or nodding, body orientation, and posture. Static

cues are characteristics of an individual that cannot readily

be altered in a given situation. Age, gender, physical

attractiveness, dress, and cleanliness are examples of static

cues. Obviously the latter three can be altered over time,

but, for example, in a job interview the applicant has the

same clothes, hairstyle, scent, etc. for the duration of the

interview. Paralinguistic cues refer to characteristics of

vocalizations. The rate, volume and tone with which someone

speaks, their ability to articulate words, as well as the

time spent talking and length of pauses before answers, are

examples of paralinguistic cues.

There is not a consensus regarding which nonverbal

behaviors are important for predicting interviewers'

judgments. Edinger and Patterson (1983) identified eleven

"nonverbal involvement behaviors" that determine the degree









of involvement between individuals in social settings. These

behaviors include interpersonal distance, gaze, touch, body

orientation lean, facial expressiveness, talking duration,

interruptions, postural openness, gestures of relational

nature, head nods, and paralinguistic cues (volume, speech

rate, and intonation). Within the context of an interview,

where the applicant's behavioral strategy is designed to

create some positive impression on the rater, Edinger and

Patterson noted that the most important nonverbal involvement

behaviors seemed to be divided, depending on the criterion.

First, gaze and facial expressions (i.e., smiling, eye

contact) were the most influential on perceptions of

likability, interest, pleasantness, credibility, and the

favorability of the applicant. Second, interpersonal

distance and touch were most influential on perceptions of

assertiveness, social activity, and favorability.

Another way to determine the possible importance of cues

may be to simply identify which cues are being studied. A

review of 19 studies (Anderson, 1960; Anderson and

Shackleton, 1990; Brunswik, 1956; Cann et al., 1981; Cash and

Kilcullen, 1985; Forbes and Jackson, 1980; Gifford et al.,

1985; Hollandsworth et al., 1979; Imada and Hakel, 1977;

Kinicki and Lockwood, 1985; Kinicki et al., 1990; McGovern et

al., 1979; Parsons and Liden, 1984; Rasmussen, 1984; Raza and

Carpenter, 1987; Sterrett, 1978; Wexley et al., 1975; Young

and Beier, 1977; Zuckerman et al., 1990), in which at least

one nonverbal cue was found to have an effect on









interviewer's judgments, shows that some researchers study

only one nonverbal cue, others study several, but rarely ever

are the same combination of cues studied. Eleven of these 19

studies measured or manipulated eye contact, making it the

most commonly studied dynamic cue. Smiling, gesturing, and

posture have also been frequently studied (7, 5, and 5

studies, respectively). Of the static cues, physical

attractiveness has received the most attention (7 of the 19

studies), with gender and appropriateness of dress also of

interest to researchers (4 and 3 studies, respectively).

Paralinguistic cues of voice articulation and modulation have

been included in three studies each, and voice loudness,

pauses before answering, and time spent talking have been

investigated twice.

This review of 19 studies only reveals what cues

researchers have identified as worthy of investigation. It

cannot determine from this information which cues are the

most significant. Also, it is unclear whether two

researchers who both studied the effects of gesturing, for

example, defined "gesturing" in the same way. Few of these

researchers explicitly defined the cues under study. Many

researchers rely on the assumption that behaviors such as eye

contact, smiling, and head nodding are commonly defined and

understood. As Burgoon and Baesler (1991) stated, "The

result is that many measurement decisions in the nonverbal

arena are governed more by happenstance, history, or cost

than by traditional measurement criteria" (p. 57).









Other researchers prefer to use the definitions set

forth by others, such as Mehrabian (1972), a psychologist

studying nonverbal behaviors in various social settings.

Mehrabian set forth scoring criteria for several categories

of nonverbal and implicit verbal behaviors. He noted that

these scoring criteria were derived from a variety of

experimental findings, and included only those cues that

yielded significant findings. He divided the behaviors into

five dimensions that he identified as the immediacy

dimension, the relaxation dimension, movements, facial

expressions, and vocalizations. The immediacy dimension

included behaviors such as touching, physical distance,

forward lean, eye contact, observation and orientation. The

relaxation dimension included arm position asymmetry,

sideways lean, leg position asymmetry, hand relaxation, neck

relaxation, and reclining angle. The movement dimension

included head nodding, head shaking, gesticulation, trunk

swivel, rocking, self-manipulation, object-manipulation, leg

movement, foot movement, and duration walking. The facial

expression dimension included such behaviors as facial

pleasantness, facial activity, and facial dominance. And,

finally, the verbalization dimension included speech error

rate, halting quality, speech volume, length and number of

statements, number of questions, verbal reinforcers,

pleasantness, positive and negative verbal content, speech

rate, and vocal activity.










Mehrabian (1972) operationally defined these behaviors.

For example, eye contact was defined as occurring when the

communicator and the addressee look into each other's eyes,

and would be measured as the duration of time that the

behavior occurs. Or, as another example, gesticulation was

defined as the number of movements of the hands or fingers,

cyclical, side-to-side, forward-back, and up-and-down

movements, scored as one unit each.

Rasmussen (1984) was one of the few researchers who gave

a detailed description of how the nonverbal cues of interest

were measured. He investigated five nonverbal cues--eye

contact, smiling, hand gestures, head nodding, and tone of

voice. He defined and quantified them in the following way:

Eye contact: any period during which the interviewee
looked directly at the interviewer (i.e., camera),
quantified as a percentage of total time.

Smiling: not specifically defined, but quantified as a
percentage of total time.

Hand gesture: a back and forth movement of the hands
equals one occurrence of a gesture, quantified as the
number of occurrences.

Head nodding: an up and down movement of the head equals
one occurrence of a nod, quantified as the number of
occurrences.

Tone of voice: an enthusiastic versus flat tone,
quantified dichotomously.

Not only have researchers failed to explicitly indicate

how the nonverbal cues were operationally defined, but they

also often failed to explain why they chose to take either a

macroscopic or microscopic approach to the measurement or









manipulation of those cues. Some researchers combined a

number of cues together in a macroscopic approach (Imada and

Hakel, 1977; Kinicki et al., 1990; McGovern et al., 1979;

Rasmussen, 1984; Wexley et al., 1975). Multiple behaviors

were combined into a single variable called nonverbal

"involvement", "immediacy", "enthusiasm" or other terms.

Other researchers have taken a microscopic approach by using

single, discrete nonverbal cues in order to identify their

individual effects on the criterion of interest (Gifford et

al., 1985; Hollandsworth et al., 1979; Parsons and Liden,

1984; Raza and Carpenter, 1987; Sterrett, 1978). The

microscopic approach usually measures behaviors as frequency

counts, durations or judgments within brief time frames, by

using highly reliable coder observations, or a physical

apparatus (Burgoon and Baesler, 1991). These latter studies,

however, are few, and more emphasis on the effects of

specific cues has been encouraged. Imada and Hakel (1977)

suggested that additional research was needed to "tease out

the effects of specific nonverbal cues as well as different

combinations of cues on selection decisions" (p. 300).

McGovern, Jones, and Morris (1979) also suggested "further

research needs to be done to identify the weightings given to

specific visual and vocal components of interviewees

nonverbal behavior" (p.178). These challenges must still be

met.










Research of Nonverbal Cues Presented in the Interview


A handful of studies have focused mainly on the effect

of physical attractiveness, a static cue, on hiring

decisions. A few other studies have exclusively investigated

paralinguistic cues. The bulk of the nonverbal cue

literature, however, is focused on the effects of dynamic

cues, or some combination of the three types. First, studies

of dynamic and combination cues will be reviewed and

discussed. Next, research that has been done on solely

physical attractiveness and paralinguistic cues will be

presented. Finally, contrary findings will be presented.

Studies of dynamic and combination cues in the interview


In many of the studies that have investigated dynamic

cues, or a combination of cues, researchers have used

videotapes of simulated job applicants and manipulated

nonverbal behaviors presented in those videos. A study by

Imada and Hakel (1977) provided the impetus for more

laboratory research using videotaped applicants when they

found no significant difference in hiring decisions between

live interviewers and observers of recorded interviews.

Imada and Hakel (1977) were not only interested in the the

medium through which judges observed nonverbal cues, but also

the effect of those nonverbal cues on interview impressions

and decisions. They manipulated nonverbal cues by training a

female applicant to vary her nonverbal behavior as either









high ("immediate") or low ("nonimmediate") while verbally

responding in a consistent manner to the interviewer's

questions. In the immediate condition, the applicant

exhibited behaviors such as greater eye contact, more

smiling, more attentive posture, less interpersonal distance,

and a direct body orientation. In the nonimmediate condition

the applicant had no eye contact, no smiling, a slouched body

posture, greater interpersonal distance, and and indirect

body orientation. Seventy-two subject were assigned to either

the immediate or nonimmediate condition, and to one of three

conditions of either participating in a live interview as the

interviewer, observing the live interview from the same room,

or observing the live interview on a television monitor from

another room. The main effects for the three conditions of

rater proximity, and interaction effects were not

significant. However, results indicated that nonverbal cues

had a significant effect on interviewers impressions and

decision to hire. Particularly, the immediate applicant was

rated as being more likely to be accepted, more successful,

more qualified, better liked, having more desirable

characteristics, more motivated, more competent, more

satisfied if given the position, and, as a result, was

recommended more often for the position. Nonverbal

information accounted for an average of 43% of the variance.

McGovern (1976; McGovern and Tinsley,1978) conducted a

study in which 52 professional personnel representatives

rated one of four videotaped employment interviews. McGovern









manipulated the level of nonverbal behaviors and the sex of

the job applicant in each videotape, while keeping verbal

content the same. Low levels of nonverbal behaviors were

defined as minimal eye contact, low energy, lack of affect

and voice modulation, lack of speech fluency, and frequent

speech disturbances. A high level of nonverbal cues included

maximal eye contact, high energy, exhibited affect and voice

modulation, fluid speech and no speech disturbances. The 10

factors identified by the researchers as critical for the

interviewer's judgment were the applicant's ability to

communicate, aggressiveness and initiative, self-confidence,

enthusiasm and motivation, intelligence, leadership

potential, maturity, persuasiveness, pleasant personality and

sociability, and positive attitude. The level of nonverbal

involvement had a significant and favorable effect on ratings

made by the judges on all 10 dimensions measured, but

particularly on ratings of applicants' enthusiasm and

motivation, self-confidence, persuasiveness, and pleasantness

of personality. Eighty-eight percent (23 of 26) of those who

saw the high nonverbal applicant reported that they would

have asked him or her back for a second interview. One

hundred percent (all 26) of the subjects who saw the low

nonverbal candidate reported that they would not have

recommended inviting the applicant for a second interview.

McGovern et al.'s (1979) replication of the prior study

confirmed these findings using students rather than

professional personnel representatives as observers.









Although students were more lenient, of those who viewed the

interviewee with a high level of nonverbal behaviors, 87%

reported that they would accept him or her. Of those who

rated the interviewee exhibiting low levels of nonverbal

behaviors, 70% reported that they would have rejected the

applicant.

Wexley, Fugita, and Malone (1975) used videotaped,

simulated loan application interviews in their study of

nonverbal cue utilization. They manipulated two levels of

nonverbal enthusiasm, represented by either high or low eye

contact, gesturing, smiling, and appropriateness of voice

tone, on the tapes. Three levels of suitability to receive a

loan (high, average, or low) was also manipulated through the

provision of objective loan application information.

Seventy-eight student raters read the loan applications,

observed the videotapes, and rated applicants' suitability to

receive a loan. A significant main effect was found for both

nonverbal enthusiasm and suitability. Those applicants who

exhibited the high levels of nonverbal behaviors received

more favorable evaluations by raters.

In another study by Young and Beier (1977), applicants

in a simulated, videotaped employment interview varied their

nonverbal behaviors in response to instructions from the

researcher to either exhibit more eye contact; more eye

contact and smiling; more eye contact, smiling and head

movement; or minimal eye contact, smiling and head movement.

Applicants followed scripts to control verbal content. Those









who were rated by independent judges as actually engaging in

more eye contact, smiling, and head movement were evaluated

by a different set of judges as more deserving of the job.

An independent rating of physical attractiveness was not

found to be a significant predictor of hiring evaluations.

In response to the popularity of using videotaped

interviews to manipulate nonverbal behavior Rasmussen (1984)

argued that the problem with the these studies was that

nonverbal behaviors were isolated, thereby researchers were

ignoring other important variables, such as verbal behavior

and resume credentials. Rasmussen used simulated interviews

recorded on videotapes in order to vary the levels of both

verbal and nonverbal behaviors. Four tapes were created to

represent either appropriate and relevant, or inappropriate

and irrelevant scripts for verbal responses, and either high

or low levels of nonverbal behaviors. Nonverbal behaviors

which were manipulated included the degree of eye contact,

smiling, hand gesturing, head nodding, and tone or modulation

of voice. In addition to viewing the videotapes, subjects

were provided with simulated resumes that included either

excellent academic achievement and highly relevant work

experience (high) or low academic achievement and little

previous work experience (low). Eighty subjects rated the

applicant on the videotape on their qualifications for the

job, on a scale from poor (1) to superior (25). Resume

credentials accounted for the greatest amount of variance,

and both resume information and verbal content had









significant main effects on the criteria. The nonverbal

behaviors alone had a relatively small effect. Rasmussen

concluded that these findings are not necessarily contrary to

prior research of the effects of nonverbal cues, but merely

indicate that when there is variability in resume credentials

and verbal responses then the additional information provided

by nonverbal cues may not be relevant. On the other hand,

when members of an applicant pool are relatively similar in

regard to verbal content and credentials, nonverbal behaviors

could have an important impact on interviewers' decisions, as

other studies that control for those variables have

indicated. In addition to those main effects, Rasmussen

reported a significant and interesting interaction effect

between verbal content and nonverbal behavior. When verbal

content was good, high levels of nonverbal behavior resulted

in even higher ratings, but when verbal content was poor,

high levels of nonverbal behavior produced lower ratings. In

other words, the presence of more nonverbal behaviors seemed

to enhance the magnitude of the effect of verbal content, not

the direction.

These six laboratory studies using videotaped, simulated

interviews have received some criticism for their lack of

generalizability to a real interview situation

(Hollandsworth, Kazelskis, Stevens, Dressel,1979; Parsons and

Liden, 1984). The variance in nonverbal behaviors

manipulated in those taped interviews may have been









artificially inflated by the researcher in comparison to how

real job applicants behave when interviewed.

One important field study that addressed this criticism

was conducted by Parsons and Liden (1984). These researchers

investigated the effect of nonverbal cues within the context

of an actual, structured employment interview. Eight

interviewers rated a total of 517 applicants on both

nonverbal cues and a final judgment of overall

qualifications. The eight cues included poise, clothing

neatness, personal cleanliness, posture, articulation of

speech, voice audibility and understandability, pauses before

answers, and eye contact. Two separate scales were used to

make the judgment, or final rating. For the interviews in

which the first scale was utilized, interviewers rated

applicants on a scale from not qualified (1) to highly

qualified (5). The other applicants were rated on a 3-point

scale as rejected, accepted, or conditionally accepted.

There was substantial multicollinearity among ratings of

nonverbal cues, with correlations ranging from .54 to .90.

The correlations between nonverbal cues and interviewer

ratings of applicant qualifications ranged from .64 to

.81(N=251). These researchers also found that nonverbal

variables of articulation and eye contact were significant

predictors of each type of hiring decision on the independent

samples (N=251, N=266). Clothing and cleanliness were the

least influential cues. Furthermore, nonverbal cues added

significantly to the rating of applicant qualifications, even









after accounting for variance of objective data, such as

application blank information, and applicants' verbal

responses to questions regarding their scholastic

performance, extracurricular activities, and previous job

experiences (N=232). Nonverbal information alone predicted

73% of the variance of interviewers judgments and, when

subsequently added, objective information increased the

variance accounted for by only 2%. Parsons and Liden's

(1984) findings are similar to those of another field study,

previously conducted by Forbes and Jackson (1980), who used

real job applicants, and found that they were most favorably

rated when they engaged in more eye contact, smiling, and

head movement.

Slightly different results were reported by

Hollandsworth et al. (1979) who also conducted a field study,

and included verbal content information as well as nonverbal

behaviors as variables of interest. Eighteen on-campus

college recruiters were asked to rate real candidates (N=338;

mean per recruiter = 4.6) on six nonverbal behaviors, the

appropriateness of the content of their verbal responses to

questions, and to indicate whether they would hire the

candidate (l=not a chance, 4=definitely hire). The six

nonverbal cues included eye contact, appropriate loudness of

voice, body posture, speech fluency, personal appearance, and

composure. Hollandsworth et al. reported that the content of

the verbal responses was the single most important variable

in a discriminant analysis. The applicants' fluency of









speech and composure were second and third in importance,

respectively. Body posture, eye contact, loudness of voice,

and personal appearance were also significant, in that order.

These findings are not totally contrary to the literature

supportive of the favorable effects of nonverbal cues. They

do suggest, however, that it is the verbal content, not the

nonverbal behaviors, on which interviewers primarily base

their judgments.

Overall, these laboratory and field studies provide

overwhelming evidence that several types of nonverbal

behaviors may influence interviewers' judgments. Next,

research studies that have exclusively studied one type of

cue will be reviewed. First, research on the effects of

physical attractiveness on interviewers' impressions and

decisions will be presented, then research on only

paralinguistic cues.

A closer look at physical attractiveness


As defined earlier, physical attractiveness is a static

cue that the applicant does not, or possibly can not, vary

during the course of the interview. Much of the research

conducted on the effect of static nonverbal cues on

interviewer decisions has focused mainly on the variable of

physical attractiveness.

One of the earliest studies of the effect of appearance

or attractiveness on others' judgments was conducted in 1956

by Brunswik. He used 25 psychology students to rate









photographs of 46 Army personnel on variables of

intelligence, energy, likability, and good looks. The

students rated the soldiers, and the soldiers rated each

other. Brunswik investigated, but found no support for, the

mediating effect of external facial characteristics (i.e.

nose height, forehead height, and overall body height) on the

relationship between actual intelligence and student raters'

estimated intelligence and personality.

As mentioned previously, an early study by Young and

Beier (1977) failed to find support for the effect of

physical attractiveness on interviewers' decisions to hire.

Those researchers noted that this finding was interesting,

considering previous, supportive results (Dipboye, Fromkin,

and Wiback, 1975; Dipboye, Arvey, and Terpstra, 1977; Cash,

Gillen, and Burns, 1977). Young and Beier suggested that

perhaps when nonverbal behaviors were present that physical

appearance was less important as a source of information for

the interviewer.

In more recent studies, Kinicki and Lockwood (1985) did

find support for the effects of physical attractiveness.

They had 24 professional recruiters conduct 3 to 4 real

interviews of 91 students as part of an interviewing skills

workshop. The recruiters rated interviewees on five

variables, including occupational knowledge, personal drive,

ability to express ideas, appearance, and attractiveness.

The first four ratings were combined into an overall

"interview impression" score, based on a subsequent factor









analysis. Other predictors of work experience,

extracurricular activities, professional objective, college

major and grade point average, and honors received, were

collected using personal data sheets filled in by applicants.

These were also reduced to two predictors--work experience

and academic achievement--as the result of the factor

analysis. The same recruiters also indicated applicants'

suitability for hire and interviewing skills, that were

treated as the criterion variables. Recruiters' interview

impressions of applicants significantly correlated .69

(p<.05) with suitability for hire and .75 (p<.05) with

interviewing skill. The independent variable of attraction

also correlated significantly with the criteria, .66 and .65

(p<.05), respectively. Relevant work experience and academic

achievement correlated significantly, however less strongly,

with just one criterion, interviewing skill, .27 (p<.05) and

.20 (p<.05), respectively. Therefore, Kinicki and Lockwood

concluded that recruiters were relying more on subjective

information of interview impressions and attraction, rather

than concrete, objective information, when making their

employment decisions.

Another study by Kinicki, Lockwood, Hom, and Griffeth

(1990) supported the previous findings. These researchers

reported that 3 nursing directors' judgments of 186 nursing

applicants relied more on interview impressions, including

applicants' appearance, attitude, job interest, and

communication skills, than resume credentials. In this study









the independent variable of attraction was not included.

Interview impressions combined with interviewers' perceptions

of applicants' work experience and education explained 42%

(p<.05) of the variance in hiring recommendations, whereas

resume data only accounted for 1%. With regard to the effect

of that information on the validity of the interview, Kinicki

et al. found that resume credentials did not predict

performance any better than interview impressions, even

though the resume cues predicted 14% (p<.05) of the variance

in performance.

Many other studies have investigated the effect of

attractiveness on interviewers' decisions, two of which are

Cash and Kilcullen (1985) and Cann, Siegfried and Pearce

(1981). Cash and Kilcullen found that applicants who were

attractive, well-qualified for the job, and male were most

preferred when judges made hiring decisions from their

fictitious resumes and attached photographs. Cann et al.

also found that the attractiveness and sex (males preferred)

of the applicant positively affected the hiring decision made

by raters. The sex of the applicant, sex of the rater, and

the sex-stereotype of the job for which they are interviewing

all seem to moderate that relationship (Cash, Gillen, and

Burns, 1977; Heilman and Saruwatari, 1979; Jackson, 1983).

One of the few field studies that investigated the

effects of applicant attractiveness on actual employment

interviews of 171 applicants for a variety of different jobs

was conducted by Raza and Carpenter (1987). These









researchers collected demographic information (i.e.,

interviewer age and sex, applicant age and sex, job) for each

interview, as well as having 8 professional interviewers rate

applicants on physical attractiveness, intelligence,

likability, skill level for the job, hirability and

acceptability of the applicant as an employee. Hirability

and employability were the criterion variables. These two

hiring decision variables correlated significantly with

ratings of intelligence, physical attractiveness, likability,

and skill level. Hirability was most strongly related to

perceived skill level, and employability was most strongly

associated with intelligence and likability.

Overall, these studies provide evidence that the

physical attractiveness of the applicant can influence

ratings made by interviewers. However, whether or not

physical attractiveness has the same impact when relevant

verbal content is taken into account is still unclear. These

researchers were mainly interested in studying only one

static nonverbal cue, that of physical attractiveness, much

like researchers who are only interested in studying a single

paralinguistic cue.

A closer look at paralinguistic cues


A rather different paralinguistic complement to the

research on physical attractiveness has been recently

introduced by Zuckerman, Hodgins, and Miyake (1990) who

investigated the effects of vocal attractiveness on









interviewers' judgments. They found strong support for the

notion of vocal attractiveness between judges who agreed on

what characteristics constitute attractive versus

unattractive voices. These researchers reported that

applicants rated as vocally attractive received more

favorable personality ratings than vocally unattractive

applicants. Thirty years earlier, Anderson (1960)

investigated the effect of a very different type of

paralinguistic characteristic on an interviewer's decision to

hire. He was interested in whether the amount of time that

the interviewer or interviewee each spent talking had any

effect on interviewer judgments. Anderson reported that in

the interviews where applicants were selected for the job

(N=70), the interviewer spent significantly more time

talking, and there was significantly less silent time, when

no one spoke, than in interviews where applicants were

rejected (N=45). Applicants who were selected also spent

less time talking and the total duration of the interview was

less than for rejected applicants, but those differences were

not statistically significant.

Contrary findings


Up until this point we have presented research that

confirms that nonverbal cues, whether they be dynamic,

static, or paralinguistic, have a favorable and significant

impact on employment decisions or judgments of applicant

suitability made by interviewers. One study in which









researchers reported evidence to the contrary was by Sterrett

(1978). This researcher created eight videotapes of a male

job applicant displaying high and low levels of eye contact,

hand gestures, formality of dress, and length of pause before

answering. The verbal content of the applicant's answers was

held constant. One hundred and sixty managers from the

insurance industry observed these tapes and rated the

applicant on eight traits that Sterrett claimed were typical

traits considered in the hiring process. These traits

included ambition, motivation, self confidence, self

organization, responsibility, verbal ability, intelligence,

and sincerity. No significant relationships were found

between the nonverbal cues and traits assessed. It should be

noted that the criteria in this study differed from much of

the other nonverbal cue research that used specific

employment criteria, such as decision to hire or

qualifications for the job. Here, observers of the videos

assessed several traits that may, or may not, have been

related to the job to be filled.


Issues of Concern and Future Directions


A careful look at the body of research on nonverbal cues

has helped to illustrate some gaps that must be filled by

further research in this area. There are three particular

issues that will be discussed here. First, the effect of

nonverbal cues on interview validity, rather than only









favorability, must be investigated. Second, ratings of

nonverbal behaviors must be made by independent, unbiased

judges. Third, the measurement of nonverbal cues should take

place at a microscopic level. These matters will

subsequently be addressed in the current study.


Validity vs. Favorability


First, the research reviewed has had as its criterion of

interest the favorability of hiring decisions. The main

issue has been whether or not applicants are hired, are

recommended for hire, are more suitable for the job, or

whether they possess attributes that are deemed important to

be successful on the job. This research has shown that

nonverbal cues have a significant impact on the favorability

of interview judgments. But, favorability is only one of two

important properties of the decision to hire an job

applicant. The other property is the accuracy of the

decision. Favorability is the degree to which the job

applicant is positively evaluated. Accuracy is the

correctness of the judgment (Motowidlo, 1986). It is

appropriate to consider the accuracy of the selection

decision since it requires a cognitive estimation of

probability, or behavioral prediction, rather than simply an

evaluative judgment of favorability. If we treat performance

ratings as "true scores", then accuracy is how well the

decision to hire corresponds with how the applicant









eventually performs on the job. A necessary condition for

accuracy is validity (Sulsky and Balzer, 1988). Therefore,

the interest here is not on an actual measure of accuracy,

but rather on validity. In other words, what effect do

nonverbal cues have on interview validity?

Only Gifford, Fan Ng, and Wilkinson (1985) have

investigated interview validity, rather than the favorability

of interviewers judgments. They videotaped 38 interviews for

a real job opening of a temporary, part-time position as a

research assistant in a university. The tapes were viewed

and applicants evaluated on motivation and social skill by 18

judges who had training and experience in interviewing.

Applicants completed a questionnaire to measure their

perceptions of their own motivation and social skill. The

applicant's self-reports were the criteria assessed in this

study. Motivation and social skill were chosen because they

were considered important for this job. Seven dynamic

nonverbal cues were measured by two trained raters and

included the time spent talking, facial regard, smiling,

gesturing, trunk recline, self-manipulation, and object-

manipulation. Static nonverbal cues also recorded included

age, sex, formality of dress and physical attractiveness.

Using the Brunswik lens model, Gifford et al. computed a

correlation coefficient, or ecological validity coefficient,

between applicant's self-assessed qualities of motivation and

social skill and the static and dynamic nonverbal cues.

Applicants who perceived themselves as very motivated were









mostly males, dressed more formally, and reclined more during

the interview. Applicants who perceived themselves as more

socially skilled also dressed more formally, gestured more,

and spent more time talking. These applicants were also

mostly male and older.

Interviewer cue utilization was computed as the

correlation between the nonverbal cues and the judges'

ratings of the applicants' motivation and social skill.

Applicants who smiled, gestured, and talked more were

perceived to be more motivated to work. Those who dressed

more formally, gestured more often, and talked more were

perceived as having more social skill by the judges.

The accuracy of the judges' attributions, as a

correlation of those ratings with applicants' self-assessed

qualities, was not significant for motivation (r=.09), but

was for social skill (r=.29, p=.05). In other words, raters

were not using relevant cues when assessing applicants'

motivation, but were using the correct cues to assess social

skill, and, as a result, were more accurate in their ratings

of social skill.

The primary shortcoming of this study by Gifford et al.

(1985) is that the measure of accuracy, or achievement score,

uses self-reported trait data as a "true" score. Judges were

not attempting to predict applicants' behavioral performance

on the job. Alternatively, instead of using self-reported

traits as the true score, researchers should use the

criterion (variable C in Figure 1.1) that interviewers are









actually trying to predict when they make a decision to hire,

that is performance on the job.

Kinicki et al. (1990) also addressed the issue of

interview validity using performance and other job-related

criteria. This study also had many shortcomings, however.

They investigated whether interview trait impression

predicted actual job success less accurately than did resume

credentials, by assessing interviewers' ability to identify

interviewee job performance, as well as attitudinal and

withdrawal predispositions. As mentioned, the interview

ratings consisted of an overall "interview impression" score

that included ratings on interviewee attitude, job interest,

job-related experience, job-related training or education,

communication skill, and the only nonverbal cue assessed,

appearance. Performance and retention information was

obtained from personnel files, and job satisfaction and

organizational commitment was measured using a survey

completed by subjects after their third week of employment.

The combined interview impression score was used for all the

regression analyses in this study. Therefore, no evidence

was reported regarding the unique effects of each individual

factor. However, correlations between all variables were

reported, and ratings of appearance significantly, and

negatively, correlated with self-reported job satisfaction

(r=-.15, p<.05), but not the other criteria.

Although Kinicki and his colleagues addressed the issue

of validity, it is deficient at meeting the present concerns.









First, they only studied one nonverbal cue, appearance, that

was buried within the macroscopic variable of "interview

impression". Second, the measure of performance used was not

independently collected, and it suffered from range

restriction. Third, the attitudinal measures were self-

reported and taken after only three weeks on the job.

In a recent study, Ambady and Rosenthal (1993)

recognized the importance of using an ecologically valid

criterion. They found that ratings made by judges exposed to

only small clips of teachers' nonverbal behavior were

significantly related to student evaluations of college

teachers' and principal's evaluations of high school

teachers' effectiveness. This study, however, was not

concerned with selection, and ratings did not take place in

the context of a selection interview. Rather, clips of

nonverbal behaviors were taken from silent videotapes of

subjects teaching classes in a university or high school

setting.

In another study which addressed interview validity and

the relevance of nonverbal cues, Motowidlo et al. (1992)

concluded that the information obtained from visual nonverbal

behaviors was not necessary for the interview's validity. In

other words, the content of applicants' answers was a

significant source of validity, since judgments by

interviewers who did not have access to nonverbal cues were

at least as valid as judgments by interviewers who did. This

finding does not entirely contradict previous research that









has supported the importance of nonverbal cues, since it has

also been reported that when appropriate and relevant verbal

information is available that it is the primary, but not

only, influence on interviewers' decisions (Hollandsworth et

al., 1979; Parsons and Liden, 1984; Rasmussen, 1984). Of

course, additional research is necessary to determine the

direction of the effect of nonverbal cues when they are

present.

Nonverbal cues may act as either enhancers or

suppressors of the interview's validity. If cues are

utilized that are irrelevant, in the sense that they do not

correspond with the interviewees' true scores on the

criterion of interest, then interviewers' judgments will be

biased and distorted. If certain nonverbal cues affect

interview outcomes, but do not predict job performance, then

those cues are a source of invalidity and interviewers must

be trained to be less influenced by them when making

judgments (Rasmussen, 1984). Schuh (1980) stated "...any

preoccupation of the interviewer with non-task relevant cues

could clutter primary memory and interfere with the

perception of the applicants verbal report...and thereby

change the applicant's information before it is needed for

recall and prediction" (p. 125). Prevailing practices

support this position, because interviewers are often trained

to attend only to the content of applicants' responses,

thereby ignoring nonverbal characteristics of the applicant

when making interview ratings (e.g. Motowidlo et al.,1992).









Conversely, if certain cues legitimately represent the

interviewees' true score on dimensions of performance,

interviewers may be justified and encouraged to use those

cues, in that validity will be enhanced (Arvey and Campion,

1982; Harris, 1989; Motowidlo et al., 1992; Rasmussen, 1984).

Especially if certain information cannot, or will not, be

explicitly verbally expressed in the interview, then it is

legitimate for interviewers to seek and utilize relevant cues

to determine the true character of the applicant (Edinger and

Patterson, 1983; Schlenker, 1980).

There is a real need to investigate the effect of

nonverbal cues on the validity of the interview, using

relevant performance, or other job-related, criteria. By

doing so one can get at the core of the purpose of the

interview--that of selecting applicants who will be good

performers on the job.


Measurement of Nonverbal Cues


The second issue of concern is the precision and

accuracy with which nonverbal cues are measured. More than a

decade ago Hollandsworth et al. (1979) criticized researchers

of nonverbal behavior for using artificial rather than real

interview settings. Researchers were criticized for

artificially inflating the variability of cues displayed by

applicants in laboratory settings. Researchers who were

anxious to get out of the laboratory and conduct real









selection interviews were willing to discard that "control"

for more generalizable results, but failed to take into

consideration how to measure the nonverbal cues in the real

setting. Parsons and Liden (1984) noted in the conclusion of

their field study that caution should be used in the

interpretation of their results since the nonverbal cues of

applicants were not controlled, as they are in the

laboratory, nor was an independent measurement taken of those

cues.

One year after Parsons and Liden's (1984) study, Gifford

et al. (1985) addressed this problem. They trained two

independent judges to score occurrences of applicants'

nonverbal behaviors and evaluate static cues. For example,

gesturing, manipulation of objects, and facial orientation

were measured as the proportion of time spent displaying

these behaviors. Physical attractiveness and formality of

dress were scored on 7-point and 3-point scales,

respectively.

Unfortunately, other field researchers did not take the

lead from Gifford and his colleagues. It is more common to

find studies in which the same person who is making the

hiring decision, or judging the appropriateness of verbal

responses, is also the one rating the applicant on nonverbal

behavior (Hollandsworth et al., 1979; Kinicki and Lockwood,

1985; Kinicki et al., 1990; Parsons and Liden, 1984; Raza and

Carpenter, 1987).









Scherer (1982), in discussing the requirements for

applying the Brunswik lens model to the study of nonverbal

communication, emphasized the need for independent

measurement of variables. He noted the necessity of

obtaining independent measures of the criterion, the

nonverbal cues, and the judgments of observers. Accordingly,

independent judges should be trained to accurately identify

and measure specific nonverbal cues, in order to obtain more

reliable and unbiased measures of those cues.


Level of Measurement and Analysis of Nonverbal Cues


The third issue of concern is the level of measurement

and analysis of nonverbal behavior. Nonverbal behavior may

be measured at either the microscopic or macroscopic level.

At the microscopic level each nonverbal cue is measured

individually. At the macroscopic level individual cues are

clustered together to define a general variable of nonverbal

behavior. Individual cues are often not defined for a

macroscopic analysis.

For example, Imada and Hakel (1977) operationally

defined nonverbal "immediacy", a macroscopic variable, as

greater eye contact, smiling, attentive posture, gestures,

smaller interpersonal distance and a direct body orientation.

Wexley et al. (1975) defined their macroscopic nonverbal

variable of "enthusiasm" as high or low amounts of eye

contact, gesturing, smiling, and appropriate tone of voice.









In both cases, individual cues were neither individually

defined, nor individually measured. Since macro-level

variables do not necessitate the careful and specific

definition of each nonverbal cue, analyses can only take

place at the macroscopic level.

With the measurement of microscopic cues, both the

individual cues and some combination of those cues can be

analyzed. This level of measurement is essential to

determine not only whether nonverbal behavior, as a whole, is

utilized in the interview and related to performance, but

also to determine which particular cues are the most

relevant. For example, an interviewee's hand movement may

not be related to performance, but the time they spend

smiling is related to performance. These findings would have

important implications for training interviewers.

Interviewers could be taught which relevant cues to attend to

and utilize when making interview judgments, rather than

being taught to ignore all nonverbal cues and attend only to

verbal content. These distinctions can only be determined by

measuring and analyzing each cue separately.

Even when nonverbal cues are measured individually, it

is the combination of these cues which is of greatest

interest. Knapp and Hall (1992) suggested that it is the

combination of individual parts of the nonverbal behavioral

system which provides the best understanding of the system's

purpose. People do not only smile, or only move their hands.

Rather, their behavior consists of multiple signals which









together contribute to the message sent to the receiver of

those cues.

Van Hooff (1982) suggested that in order to interpret

nonverbal behavior we should use larger functional units of

behavior rather than molecular behavioral elements or acts.

These molar units, he suggested, enable researchers to record

and compile more meaningful, complex behavioral processes.

He cautioned that the connections between behavioral

elements, or acts, tend to get lost if a molecular

description of the behavior is used (Van Hooff, 1982).

Brunswik's (1956) lens model also indicates that a

combination of nonverbal cues, rather than individual cues,

influence person perceptions and judgments. Brunswik's

hierarchy of cues (see Figure 1.2) suggests that certain cues

are more important than others, or, if combined, individual

cues would carry different weights.

In summary, it is important to recognize that

naturalistic nonverbal behavior is not exhibited in

unconnected parts, but rather as an orchestrated whole.

Measuring individual nonverbal cues at the microscopic level

allows for analyses of both composite and molecular nonverbal

cues.


Summary of Issues to be Addressed in the Current Study


In reaction to the research that has been conducted thus

far, three issues have been presented regarding the purpose





58



of these studies and the methodologies used to achieve those

purposes. These three issues highlight (1) the need to

investigate the effects of nonverbal cues on interview

validity rather than favorability, (2) the need to use

independent ratings of those cues for analysis, and (3) the

need to measure nonverbal behavior at the microscopic level.

In the current study these needs will be met.














PRELIMINARY STUDIES AND
THE CURRENT STUDY RESEARCH QUESTIONS

In preparation for the current study it was necessary to

conduct three preliminary studies. In the first study, the

structured interview was developed, actual interviews were

collected on videotape, and the interview was validated.

This study provided evidence of the soundness of the

interview questions for predicting management effectiveness.

Second, a sample of the taped interviews was used to

determine if visual cues, in general, provide sufficient

information for raters to make valid judgments of

interviewees. The results of this study were quite

compelling, in that, raters who were exposed to only visual

cues were able to make valid ratings, and that the sex of the

interviewee seemed to play some role in the relationship

between interview and performance ratings (Motowidlo and

Burnett, 1992).

A third study was conducted to identify which specific

nonverbal cues are most likely to be related to behavioral

dimensions of management effectiveness. Related literature

was reviewed, and empirical evidence was collected. These

two sources helped to identify nonverbal cues for

investigation in the current study.

After each preliminary study is presented here, their

implications for the current study will be discussed. Next,









the research questions for the current study will be

presented, as well as reasons why the answers to those

research questions are both interesting and important.


Study #1: The Development and Validation of the Interview


A structured selection interview will be used as the

context of the current dissertation research. The interview

dimensions, questions, and format for use in the current

study were developed as part of the first preliminary study.

Supervisory ratings of performance will be the criterion for

the current dissertation research. The measures needed to

collect those supervisory ratings were also developed as part

of this first study. Additionally, in this preliminary

study, interviews were conducted and recorded on videotape,

interviewees were rated by the interviewer, and supervisory

ratings were collected. The collection of this data allowed

for a preliminary validation of the interview.

A Structured Selection Interview

Development of interview questions and dimensions

An interview system was developed to measure general

managerial skills for entry-level managers. Performance

dimensions of managerial skill were identified according to

prior research (Motowidlo et al., 1992). They included:

leadership, forcefulness, teamwork, open-mindedness,

consideration, planning and organization, thoroughness,

drive, and results orientation. These nine dimensions were









then combined to form four dimensions. This was accomplished

by having three doctoral management students carefully read

each dimension definition, and then sort the nine dimensions

into fewer, meaningful categories based on their similarities

and degree of overlap. The four new dimensions are:

Leadership (leadership + forcefulness), Teamwork (teamwork +

open-mindedness + consideration), Drive (drive + results

orientation), and Planning and Organizing (planning and

organization + thoroughness).

Rating scales were developed for each dimension. These

scales were behaviorally anchored with general behavioral

descriptions at the High, Moderate, and Low level of each

scale. These were 7-point scales, where 7 and 6 are at the

High level, 5, 4, and 3 at the Moderate level, and 2 and 1 at

the Low level. These anchors went through several revisions

before arriving at the final scales (see Appendix A).

Next, interview questions were developed to tap each

behavioral dimension. As discussed previously, in the

structured interview literature there are two basic types of

interview questions. Janz (1982) introduced past-oriented

questions where interviewees are asked, "Tell me about a time

when...". Another type of question, developed by Latham et

al. (1980), is oriented toward future behavior and

interviewees are asked, "Tell me what you would do if...".

There is no sufficient evidence regarding whether one type of

question is better than another. As a result, in this study,

both types of questions were included. A total of eight









interview questions were developed. Two questions, one of

each type (i.e., past-oriented and hypothetical), were

developed for each of the four dimensions.

The interview questions were pilot tested using

undergraduate students in an elective human resource

management class at a large Southeastern university. Twelve

students participated in the first pilot test, where two

doctoral management students, including this researcher,

acted as the interviewers. Questions were revised as a

result of the first pilot. The same interviewers conducted a

second pilot with 14 different students from the same class.

Additional revisions were made before arriving at the final

eight interview questions (See Appendix A).

Collection of interviews

In order to collect interviews of real managers the

cooperation of four utility and telecommunications companies

was solicited. Company representatives contacted

interviewees to obtain their participation, and schedule the

interview appointments. Seventy-three interviews were

collected.

Interviews were conducted in three day periods in each

company. Approximately eight people were interviewed each

day, at hour intervals. Interviews took, on the average,

thirty to forty minutes. Each interview was videotaped.

At the beginning of the interview session, before the

videotaping began, the interviewer briefly described the









research project and the purpose for which the videotapes

would be used. Interviewees were asked to imagine that the

session was a real job interview, and to provide real and

truthful answers. The videocamera was then turned on.

At this point, the interviewer told the interviewee

about the types of questions that they would be asked (past

versus future). The interviewer then asked the eight

interview questions, as well as probing questions. Half of

the interviewees were asked the past-oriented questions

first, and the other half were asked the future-oriented

questions first.

At the conclusion of each interview the interviewer

collected demographic information and asked the interviewee

if they had a preference for one or another type of question.

Demographic information was obtained by directly asking the

interviewee for their job title, job tenure, organizational

tenure, a brief description of their job duties, and their

age. Additional demographics recorded by the interviewer

included race and sex.

Interviewees were also asked to compare the two types of

questions, past- versus future-oriented. The actual

questions were: "As you know we used two types of questions

in this interview, past-oriented and the hypothetical. Which

type do you think applicants would prefer? Which type seemed

more relevant to your job?". Responses were recorded as

either past, future, both, or none (see Appendix A).









Interview ratings

The interviewer rated interviewees using the four

behaviorally anchored rating scales of leadership, teamwork,

drive, and planning and organizing, for each type of

question. This resulted in eight ratings for each

interviewee.


Supervisory Ratings of Job Performance


Development of performance rating scales

The same four dimensions and accompanying scales used

for the interview were also used for the supervisors' ratings

of their subordinate's performance on their current job. The

supervisors' form has a more narrative format for the

behavioral anchors on each scale, than the bullet-type

sentences on the interview form. The supervisors' form also

includes a brief description of the project and detailed

instructions on how to fill-out the form .

In addition to the four dimensions of Leadership,

Teamwork, Drive, and Planning and Organizing, supervisors

were asked to rate individuals on a dimension of Overall

Performance. They also provided information regarding how

important they felt each dimension was for the job that their

subordinate performs (see Appendix B).

Collection of performance ratings

Each company representative distributed the performance

rating forms to the appropriate supervisor of each









interviewee. The forms were accompanied by instructions that

they be returned directly to the researchers. Sixty-five

supervisory ratings, or 89% of the original 73 interviews,

were returned.

Supervisors reported, on average, that Leadership skills

were "important" (mean=3.2, s.d.=1.13), Teamwork was "very

important" (mean=4.2, s.d.=0.72), Drive was "very important"

(mean=4.2, s.d.=0.67), and Planning and Organizing was "very

important" (mean=4.0, s.d.=0.78). None of the four

dimensions was rated "not at all important" by any

supervisor. These responses indicated that these dimensions

were appropriate for assessing the general effectiveness of

entry-level managers.


Validation of the Interview


In order to determine the validity of the entire

interview, and of each type of question, interview ratings by

the interviewer were correlated with supervisory ratings.

First, the four dimension ratings were correlated within

question type. The average correlation for past-oriented

questions was .40 and for hypothetical questions was .39.

This justified the combination of dimension ratings into

total scores for each type of question. Also, an interview

total score was computed across question type. This involved

averaging the past-oriented and future-oriented score on each

dimension, and then summing the four averages.









Next, the total scores for each type of question and the

overall interview totals were correlated with performance.

The correlation of performance with past-oriented questions

was .31 (p1.01), with future-oriented questions r=.22 (p.05)

and with the total interview score r=.30 (p<.01). These

results should be interpreted with caution since the

interview ratings were made by only one rater.


Summary of Preliminary Study #1


The development and collection of interview and

performance ratings for 65 interviewees were described here.

The correlational analysis of interview ratings with

performance ratings indicated that both past-oriented and

future-oriented questions are significantly correlated with

performance.


Study #2: Aural and Visual Sources of Validity
(Motowidlo and Burnett. 1992)


The current dissertation research study will investigate

the effect of nonverbal cues on the validity of the

structured interview. But first, it was necessary to

determine what, if any, effect visually-based information had

on interview ratings. The results of this second preliminary

study legitimized the further investigation of the role of

nonverbal cues on the validity of the interview.










Purpose of Study and Research Ouestions


The purpose of this second preliminary study was to

determine the effect of visual cues on the validity of a

structured interview. The primary research question was:

How valid are judgments based only on visual cues for

predicting job performance? Secondary research questions

included: What is the relationship between judgments made

when only visual cues are available and when both aural and

visual cues are available?; and, What is the relationship

between aurally based judgments and visually based judgments?


Methods and Procedures


The videotaped interviews, described in Study #1, were

the stimulus in this study and the criterion was the

supervisory performance ratings. Interview ratings were

collected from 194 undergraduate students. These student

raters either a) watched and listened, b) only listened

(with no picture), or c) only watched (with no sound) to 40

of the 65 videotaped interviews. Additionally, raters either

saw and/or heard the portion of the original interview where

interviewees responded to past-oriented question, or to the

responses to future-oriented questions. This design resulted

in 6 (3 forms of cue availability x 2 types of questions)

different conditions.









There were an average of 32 raters per condition. All

40 interviews, combined into 10 sets of 4 interviews, were

presented in each condition. Each rater rated 4 interviews

and each interview was rated by an average of 3.2 raters in

each condition.


Summary of Results


The correlations between interview ratings of the past-

oriented questions and performance for the aural/visual,

aural only, and visual only conditions were .45 (p<.05), .32

(p..05), and .27 (p<.05), respectively. The correlations

between the hypothetical questions and performance for the

aural/visual, aural only, and visual only conditions were .16

(n.s.), .24 (n.s.),and .29 (p<.05), respectively. Pooled

ratings, across question type also correlated significantly

with performance for the aural/visual (r=.36, p<.05), aural

only (.33, p<.05), and visual only (.32, p.05) conditions.

These results address the primary research question, and

confirm that visually based interview judgments can be valid

predictors of performance.

Second, significant correlations were found between

visually based judgments and judgments made when both visual

and aural cues are available. This was true for both ratings

based on past-oriented questions (r=.51, pS.01) and future-

oriented questions (r=.46, p..01). The pooled visually based

ratings correlated significantly (r=.68, p.01) with the









pooled ratings made when both visual and aural cues were

available. Furthermore, both visual and aural cues

contributed independently to the variance explained in

ratings made when both types of cues were available. These

results address the second research question, and indicate

that there is a strong positive relationship between visually

based ratings and ratings made from both aural and visual

information.

Third, judgments made by raters who were presented with

only visual cues and those presented with only aural cues

were significantly correlated for the past-oriented questions

(r=.49, p..01) and marginally significant for the future-

oriented questions (r=.25, p=.06). The pooled ratings of

raters who only watched the interviews correlated .53 (p<.01)

with pooled ratings of those who only heard the interviews.

This result addresses the third research question and may

indicate that there exists some redundancy between these two

types of cues.

As an exploratory analysis, the sex, age and race of

interviewees were examined as potential influences on the

validity findings. Although age and race were not related to

performance or interview ratings, the sex of the interviewee

had a significant relationship with both performance (r=-.33,

p.05) and visually based interview ratings (r=-.40, p<.05).

This may indicate that sex differences play some role in

correlations between interview and performance ratings,

particularly when visual cues are available.










Summary of Preliminary Study #2


The results of this study were quite intriguing. First,

raters who were exposed to only interviewees' visual cues

were able to make valid ratings. It was concluded that

visual cues play a potentially important role in interview

validity.

Second, visually based ratings and aurally based ratings

were highly related. One explanation for this finding may be

that similar personality traits underlie both sets of

judgments. In this case, aural and visual cues would share

common variance. Alternatively, it may be systematic error

variance which is shared by raters of these two types of

cues.

Third, the sex of the interviewee seemed to play an

important role in the relationship between interview and

performance ratings. One possibility for this result may be

that women exhibit different visual cues than do men. The

nonverbal cues women tend to show may be judged negatively by

raters who are assessing potential management effectiveness.

On the other hand, men and women may be behaving in similar

ways, but interpretations of those behaviors by raters may

differ by sex.










Study #3: Identifying Important Nonverbal Behaviors


Before conducting the current study of the effect of

nonverbal cues on the validity of the interview, the specific

nonverbal cues had to be chosen. This third preliminary

study served to identify those nonverbal behaviors that were

the most relevant for the purpose of the current study.


Overview


In order to choose the specific cues for the current

study those cues must be identified which are considered the

most likely to be relevant for predicting job performance.

Rather than associating the actual nonverbal behaviors

directly with performance, it is presumed that nonverbal

behaviors reflect personality traits which, in turn, are

related to managerial performance. In other words, rather

than stating that effective managers smile more often than

ineffective managers, one can state that smiling is

associated with agreeableness, a trait considered important

for effective management.

To justify the choice of certain dynamic, static, and

paralinguistic cues for the current study, traits believed to

underlie the four job-related dimensions leadership,

teamwork, drive, and planning and organizing must be linked

to cues that existing research indicates are relevant

reflections of those same traits.










Linking Performance Dimensions with the Big Five Personality
Traits


Researchers who have investigated the co-occurence of

behaviors and traits have mainly used a five factor model of

trait assessment. These five factors were derived in several

studies across different samples that found five fairly

strong factors of personality assessment (see Tupes and

Christal, 1992). These factors are Extroversion,

Agreeableness, Conscientiousness, Emotional Stability, and

Intellect.

Extroversion, or Surgency, is defined by traits such as

talkativeness, frankness, adventurousness, assertiveness,

sociability, energetic, composed, interest in opposite sex,

and cheerfulness. The second factor, Agreeableness, is

defined by several traits including good-natured, not

jealous, emotionally mature, mildness, cooperativeness,

trustfulness, adaptability, kindliness, attentiveness to

people, and self-sufficiency. Traits that load negatively on

the Agreeableness factor include assertiveness, talkativeness

and orderliness. The third factor, called Conscientiousness

or Dependability, is defined by traits of orderliness,

responsibility, perseverance, and conventionality, as well as

smaller loadings on the variables cooperativeness, mildness

and emotional stability. The fourth factor is called

Emotional Stability or the opposite label used is

Neuroticism. If using the first label the variables that









load highest on this factor include placid, poised, non-

hypochondriacal, calm, and self-sufficient. Secondary

variables on this factor are lack of jealousy, emotional

maturity, cooperativeness, trustfulness, adaptability,

responsibility, perseverance, and independent-mindedness. The

fifth factor is called Culture or Intellect and is defined by

traits such as intellectual, imaginative, artistically

sensitive, openness and polishedness.

To link these five personality factors with the job-

related dimensions six doctoral students were asked to

determine which traits seemed most indicative of each

dimension of leadership, teamwork, drive, and planning and

organizing. Each judge was provided with a list of 34 trait

adjectives which have been used to define the Big Five

personality factors (Tupes and Christal, 1992). The order of

the traits was scrambled. Judges were asked to rate the

degree of association between each trait for each of the four

performance dimensions on a 5-point scale (5=definitely

related to l=not at all related) (see Appendix C).

There were a different number of trait adjectives

representing each personality factor. For each personality

factor, mean scores were calculated by averaging the ratings

made by judges on those trait adjectives within each

performance dimension. For example, for the Leadership

dimension nine trait adjective ratings were averaged to form

a score for Extraversion. This resulted in five personality









factors scores within each of the four performance dimensions

for each of the six judges.

To determine the reliability of judges, intraclass

correlations were computed according to the model by Shrout

and Fleiss (1979). The corrected intraclass correlation

(ICC) for Leadership was .68, for Teamwork .59, for Drive

.68, and for Planning and Organizing .72. Based on these

estimates of reliability, it was concluded that there was

agreement and consistency among judges when they indicated

their perception of the relationship between personality

traits and performance dimensions.

Additionally, a two-way analysis of variance was

computed. The dependent variable was the mean scores on each

personality factor within each dimension for each judge

(5x4x6=120). The two independent variables were the

personality factors and the judges. There was a main effect

for the personality factors analyzed separately for each of

the four dimensions, as well as overall. This indicates that

there is a significant difference in the pattern of means

among personality factors within each performance dimension.

Next, a cut-off was established at 3.0. The top two

mean scores within each performance dimension which were

greater than or equal to 3.0 would be considered

representative of that dimension. The mean scores for each

personality factor on each dimension are presented in Table

2.1.









Leadership was most related to Extroversion and

Conscientiousness; Teamwork was highly related to

Agreeableness, and less so to Conscientiousness; Drive was

related to Conscientiousness and Extroversion; and Planning

and Organizing was highly related to Conscientiousness, as

well as Agreeableness. Note the mean scores for the factors

of Emotional Stability and Culture/Intellect were the lowest



Table 2.1. Means and standard deviations of judges' ratings
of personality factors across performance dimensions.

Performance Dimensions


Personality
Factors

Extroversion

Agreeableness

Conscientious

Emotional
Stability

Culture/
Intellect


Leadership

3.91 (0.81)

3.55 (1.10)

3.77 (1.01)


3.53 (1.16)


2.92 (1.02)


Teamwork

3.20 (1.00)

4.02 (1.06)

3.40 (1.22)


3.14 (1.15)


2.80 (1.06)


Drive

3.70 (1.14)

3.12 (1.13)

3.87 (1.22)


Planning/
Organizing

2.83 (0.99)

3.30 (1.23)

4.27 (0.98)


3.09 (1.24) 3.11 (1.30)


2.75 (1.07)


2.85 (1.01)


for each dimension, and did not meet the cut-off criteria for

any of the dimensions.

After the performance dimensions were linked to the

traits, the next task was to link the traits to nonverbal

cues. The outcome is the list of nonverbal cues that will be

investigated in this study.










Linking the Big Five Personality Traits with Nonverbal
Behaviors


Strong relationships are frequently found between

judgments of personality and objectively measured behaviors

or physical attributes (Albright, Kenny, and Malloy, 1988;

Borkenau and Liebler, 1992; Dion, Berscheid, and Walster,

1972; Funder, 1983; Funder and Colvin, 1988; Funder and

Sneed, 1993; Gangestad, Simpson, DiGeronimo, and Biek, 1992;

Kenny, Horner, Kashy, and Chu, 1992). Even when the only

information available to judges is the target's physical

attributes, trait assessments have been found to be accurate

and consensus exists among judges (Cleeton and Knight, 1924;

Hunt and Lin, 1967; Norman and Goldberg, 1966; Passini and

Norman, 1966). This context, where there is no interaction

between judges and targets, and no prior knowledge of one

another, is called "zero acquaintance" (Albright et al.,

1988). Evidence indicates that judgments made at zero

acquaintance are stable even after the judge and target have

interacted (Kenny et al., 1992), and the accuracy of trait

assessments increases directly with the degree of

acquaintanceship (Paunonen, 1989).

While the degree of acquaintance in the context of a

selection interview is not zero, it is low. Before the

applicant walks through the door for the interview the only

information an interviewer typically has is the individual's

resume. Interaction between the interviewer and applicant









takes place within a limited period of time, typically 30

minutes. During that brief period the interviewer must

obtain adequate information in order to make judgments

regarding the applicant's potential to perform on the job.

The interviewer is initially presented with the physical

attributes of the applicant (attractiveness, dress, etc.) but

as the interview progresses the applicant provides dynamic

behavioral and audible cues. All of these cues, researchers

have argued, influence the assessment of personality traits,

and those traits seem to underlie the behavioral dimensions

upon which the interviewer makes ratings.

In order to make these inferences between traits and

cues it is essential to measure the behaviors or cues that

are indicative of relevant traits (Gangestad et al., 1992).

Researchers are only beginning to understand the cues that

judges use to make trait assessments, and, therefore, there

are no specific guidelines for defining the co-occurrences

between traits and behaviors, or as Borkenau called it, "what

goes with what" (1992, pp. 297-298). A summary of related

literature is presented here which has identified the

relation between certain nonverbal cues and three of the big

five personality factors of Extroversion, Agreeableness, and

Conscientiousness. These relations are illustrated in Figure

2.1.

First, dynamic cues such as smiling, eye contact and

gaze, and rapid body movement seem to reflect Extroversion.

Kenny et al. (1992) noted that rapid body movement was










correlated at .47 (p5.005) with Extroversion. They also

reported a correlation of .49 (p5.005) between smiling and

this factor. Borkenau and Liebler (1992) reported

significant correlations between Extroversion and extensive

smiling (r=.33, p5.O1) and friendly expression (r=.39,

pS.O1). These researchers also found less eye contact to be

negatively, and significantly correlated with this factor

(r=-.33, p5.O1). (Argyle, 1988) noted that smiling, gaze, and

spatial proximity were correlates of Extroversion. Static


Performance
Dimensions

Leadership \


Personality
Trait Factors

Extroversion.A


Teamwork


Agreeableness


Drive


Cons


ness


Planning and
Organizing


Nonverbal Cues

Gaze/Eye Contact

Smiling

Body Movement

Hand Gestures

Spatial Proximity

Posture

Physical
Attractiveness

Appearance

Aspects of Dress

Vocal
Characteristics

Amount of Speech


Figure 2.1. Associations between nonverbal cues, traits and
performance dimensions.









cues that seem to be related to Extroversion include

attractiveness and appropriateness of dress. Borkenau and

Liebler (1992) found that attractiveness correlated .37

(p5.O1) with Extroversion, and Albright et al. (1988)

reported a correlation of .74 (p5.05) between the same

variables. Kenny et al. (1992) replicated and confirmed

Albright et al.'s findings. Borkenau and Liebler (1992) also

found that Extroversion was significantly related to an

unrefined appearance (-.28, p<.01), showy dress (.34, p<.01),

and unfashionable dress (-.27, p5.O1). Two paralinguistic

cues seems to also reflect Extroversion. Borkenau and Liebler

(1992) found a significant correlation between this factor

and softness of voice (-.26, p5.01), and (Argyle, 1988)

identified amount of speech as a correlate of Extroversion.

Second, dynamic cues that have been found to be related

to the second factor, Agreeableness, include smiling, hand

movements, and rapid body movement. Kenny et al. (1992)

reported a significant correlation of .66 (p5.005) between

smiling and Agreeableness, and Borkenau and Liebler (1992)

found a relationship between friendly expression and this

factor (r=.23, pS.05). In the study by Borkenau and Liebler

(1992) a significant relationship was also found between

Agreeableness and frequency of hand movement (r=-.23, p5.05).

Rapid body movement was reported by Kenny et al. (1992) as

related to this factor (r=-.35, p5.005). The Agreeableness

factor is also related to paralinguistic cues, including

effortful reading (r=.31, p5.01), easy to understand (r=.28,









p9.01), and hectic speaking (r=-.25, p5.01) (Borkenau and

Liebler, 1992).

One dynamic cue has been found to be related to the

third factor, Conscientiousness. Borkenau and Liebler (1992)

reported a significant, negative correlation between relaxed

posture and Conscientiousness (r=-.29, p5.01). Static cues

related to this factor are associated with dress. Borkenau

and Liebler (1992), Kenny et al. (1992) and Albright et al.

(1988) all noted significant correlations between dress

formality and the Conscientiousness factor (r=-.25, pS.05

reversed scored; r=.44, p5.005; and r=.76, p5.05

respectively). Albright et al. (1988) also found that

neatness of dress was related to this factor (r=.73, p5.05).

Borkenau and Liebler (1992) found one paralinguistic cue,

effortful reading, was significantly related to

Conscientiousness (r=.32, p5.01).


Summary of Preliminary Study #3


Evidence and research have been presented regarding

"what goes with what". More specifically, judges' ratings

indicate what traits go with what performance dimensions, and

existing research has been used to indicate what cues go with

what traits. These linkages will be used to justify the

choice of nonverbal cues for investigation in the current

study.










Implications of Preliminary Studies on the Current Study


These three studies had important implications for the

current study. As a result of each preliminary study,

important decisions were made regarding the variables of

interest and methods of the current study.


Decision Based on Study #1: Use of the Structured Interview


The structured behavioral interview format developed in

Study #1 will be used in the current study. First, the

training and tools provided to interviewers using this type

of format make the processes of cue attention and utilization

more controlled, thereby enabling interviewers to more

readily distinguish relevant from irrelevant cues. Second,

the validation results confirm the soundness of the interview

questions developed to predict general managerial

effectiveness. Accordingly, the structured behavioral

interview is the appropriate context for this study.


Decisions Based on Study #2: Type of Interview Ouestion and
Sex Differences


First, the results of Study #2 indicated differences in

validity of past-oriented and future-oriented types of

interview questions. Similar differences were found between

interview and performance ratings in Study #1. In both

cases, correlations with performance were lower for future-

oriented questions than for past-oriented questions. As a









result, it was decided that only the portion of the original

interviews where interviewees responded to the past-oriented

questions would be included in the current study.

Second, the results of Study #2 imply that the sex of

the interviewee may play an important role in the validity of

the interview. Accordingly, sex differences will be explored

in the current study. The detailed information which will be

collected regarding interviewees' nonverbal behavior may help

to explain why sex differences were found in Study #2.


Decision Based on Study *3: The Choice of Nonverbal Cues


The nonverbal cues that will be investigated in the

current study are based on the findings in Study #3 and are

presented in Table 2.2. The four dynamic and two static cues

were found to be significantly correlated with at least one

of the personality factors determined to underlie the

behavioral dimensions of leadership, teamwork, drive, and

planning and organizing. The paralinguistic cue chosen for

investigation, vocal attractiveness, is more global than the

paralinguistic cues noted previously (i.e. softness of voice,

effortful reading, easy to understand, and hectic speaking).

However, Zuckerman, Hodgins, and Miyake (1990) found vocal

attractiveness to be significantly related to the

favorability of personality ratings. Accordingly, vocal

attractiveness is considered, by this researcher, to be more

appropriate for the interests of the current study.











Table 2.2. Nonverbal cues of interest in the current study.

Dynamic Static Paralinguistic
Gaze Physical attractiveness Vocal attractiveness
Smiling Dress
Hand movement
Back/side lean
Body orientation



It is assumed that all of these nonverbal cues if

exhibited by an interviewee would be considered favorable by

interviewers, except for back/side lean. This dynamic cue is

often interpreted as a sign of relaxation or boredom (Argyle,

1988; Mehrabian, 1969; Mehrabian, 1972). A relaxed posture

also communicates dominance in interpersonal interaction

(Argyle, 1988; Mehrabian, 1972). These characteristics are

not considered appropriate, nor judged favorably in the

context of a selection interview. Alternatively, when an

interviewee leans forward they appear attentive, interested,

and involved in the interviewers' questions (Bull, 1987;

Mehrabian, 1972). Attentiveness and involvement are more

appropriate behaviors in an interview. Accordingly, the

variable of back/side lean will be reverse scored for

analyses in the current study.


Summary of the Implications of the Preliminary Studies on the
Current Study


As a result of the three preliminary studies the

videotaped interviews of the 65 interviewees responding to









the four past-oriented structured interview questions will be

used in the current study. Furthermore, the nonverbal cues

to be measured in the current study were identified.

Finally, sex differences will be investigated as possible

influences on the validity of the interview.


Current Research Study


The focus of the current study is the effect of

nonverbal cues on the validity with which interviewers make

predictions of interviewees' job performance. The specific

research questions of this study will be stated in this

section. Also, the reasons for why this study is both

interesting and important to researchers and practitioners

will be discussed.


Current Research Ouestions


Four specific research questions will be asked to

determine how nonverbal cues affect interview validity. The

first question asks: Do raters of interviewees attend to and

utilize nonverbal cues in the interview? This question

corresponds to the right side of the Brunswik lens model (see

Figure 1.2), or the utilization of nonverbal cues. Cue

utilization has been the focus of several research studies on

the effect of nonverbal cues on the favorability of interview

ratings, but investigating this question in the context of a

structured interview is unique. As part of structuring an









interview, researchers and practitioners have emphasized the

importance of focusing interviewers' attention exclusively on

the content of applicants' answers. This implies that visual

cues do not carry important information and should be

ignored. The answer to this first research question will

help to determine whether or not interview ratings, made

using a structured format, are based on information carried

by nonverbal cues as well as verbal content.

The second research question asks: Are nonverbal cues

associated with performance on the job? This question

corresponds to the left side of the Brunswik lens model, or

the ecological validity of nonverbal cues. As discussed in

the previous chapter, only three research studies by Gifford

et al. (1985), Kiniki et al.(1990), and Ambady and Rosenthal

(1993) have addressed whether nonverbal cues are related to

job performance. However, there were several deficiencies

in those studies which will be overcome in the current study.

The third and fourth research questions ask: Do

nonverbal cues contribute incrementally to validity after

interview ratings are taken into account?; and do interview

ratings contribute incrementally to validity after nonverbal

cues are taken into account? These two questions correspond

to Brunswik's concept of achievement. Both questions are

necessary because there is no existing evidence regarding

these relations. Again, the issue of how nonverbal cues

influence interview validity has not yet been sufficiently









addressed, and these research questions are an attempt to

confront that issue.


Important and Interesting Implications of the Current Study


Results of this study can contribute to the increasing

research effort to improve the interview as a selection

device. The structured techniques that have been studied,

and are now being marketed in various forms, should not be

treated as the only way to improve the interview. We cannot

ignore a history of research that has shown that other

factors, including nonverbal cues, influence interviewers'

judgments.

The investigation of the effect of nonverbal cues on

interviewer validity is itself important because, until now,

research has only used criteria such as the favorability of

ratings or personality traits. This study will take an

important step further by using job-related criteria (i.e.,

supervisory ratings of performance on the job) to assess the

overall effect of nonverbal cues. The issue of validity is

important to managers and practitioners, as well. They are

interested in significant ways to improve interview validity,

since validation is the primary criteria used when

establishing the defensibility and appropriateness of

selection procedures (Cronshaw and Wiesner, 1989).

Last, by independently measuring microlevel nonverbal

cues we will be able to "tease out" the individual and





87



combined effects of those behaviors on validity. Researchers

and practitioners are interested in knowing not only whether

nonverbal cues are appropriately utilized, but also exactly

which cues are used to make valid judgments. Findings of

this study could have an important impact on the training of

interviewers.
















HYPOTHESES, METHODS AND PROCEDURES

This chapter begins with a statement of the research

hypotheses and ends by describing the data analysis which

will be used to test those hypotheses. Additionally, methods

and procedures will be carefully explained, and the

reliabilities of interview, performance, and nonverbal cue

ratings will be reported.



Research Hypotheses

Hypotheses will be presented here which are derived from

the previously stated research questions. The four research

questions to be investigated are: (1) Do raters of

interviewees attend to and utilize nonverbal cues in the

interview?, (2) Are nonverbal cues associated with

performance on the job? (3) Do nonverbal cues contribute

incrementally to validity after interview ratings are taken

into account?; and (4) Do interview ratings contribute

incrementally to validity after nonverbal cues are taken into

account? These research questions correspond to Brunswik's

concepts of cue utilization, ecological validity, and

achievement, respectively.











Cue Utilization


With regard to the first research question, it is

expected that raters will attend to and utilize nonverbal

cues when making interview ratings. This expectation is

consistent with the findings in the nonverbal cue literature

that interviewers attend to and utilize cues, and that their

use affects the favorability of their judgments (Forbes and

Jackson, 1980; Gifford et al., 1985; Hollandsworth et al.,

1989; Imada and Hakel, 1977; McGovern, 1976; McGovern and

Tinsley, 1978; Parsons and Liden, 1984; Young and Beier,

1977).

Arvey and Campion (1982) suggested that interviewers may

not view verbal and nonverbal variables as independent. In

fact, researchers who have investigated the effect of both

verbal and nonverbal information on rating favorability have

not suggested that nonverbal information be eliminated

completely. Rasmussen (1984) and Hollandsworth et al. (1979)

found that verbal content was the primary source of

information used by interviewers, but nonverbal behavior

still had a significant, albeit smaller effect in both

studies. Therefore, it is expected that interviewers will

use both types of information cues. Accordingly, the first

hypothesis is:









Hypothesis 1. Interviewees' nonverbal cues will be

significantly and positively correlated with interview

ratings.



Ecological Validity


Not only is it expected that interviewers will use

nonverbal cues, but also that these cues will be related to

performance ratings. According to recent research, visual

cues can be remarkably accurate indicators of personality

traits. If these same traits are important for job

performance, then nonverbal cues may be correlated with

supervisory ratings of performance. Motowidlo and Burnett

(1992, preliminary study #2) found that raters who were

exposed to only visual sources of information were able to

make valid interview ratings. Ambady and Rosenthal (1993)

reported similar results where judgments of college

teachers', based on only teachers' nonverbal behavior, were

significant predictors of student evaluations of teacher

effectiveness. Thus, information communicated through

nonverbal behavior seems to be an important indicator of job

performance, and should not be ignored. The specific

hypothesis is stated as follows:

Hypothesis 2. Interviewees' nonverbal cues will be

significantly and positively correlated with supervisory

ratings of job performance.











Achievement


The third and fourth research questions are exploratory

in nature and difficult to hypothesize due to a lack of prior

evidence or theory to support a specific expectation.

However, following the logic of the Brunswik lens model, if

cues are being utilized by those who make interview ratings,

and those same cues are related to the criterion of job

performance, then the validity of the judgments, as a

function of these two relations, may be enhanced.

Assuming raters of the interview are utilizing nonverbal

cues and those cues are related to performance, two outcomes

are still possible for the third research question. The

first possibility is that raters attend to and utilize

nonverbal cues, but do not weight them sufficiently when

making their ratings. In this case, the addition of

independently measured nonverbal cues to interview ratings

will explain significantly more variance in the criterion

even though raters already took those nonverbal cues into

account. Alternatively, if interview raters utilized those

cues appropriately, and the information they provide was

sufficiently incorporated into the interview ratings, then

nonverbal cues would not significantly add to the validity of

interview ratings alone.

Again assuming that raters of the interviews are

utilizing nonverbal cue information, and that those cues are