See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/235303246

Future employment selection methods: Evaluating social

networking web sites

Article  in  Journal of Managerial Psychology · August 2009

DOI: 10.1108/02683940910974134

CITATIONS READS
154 5,121

2 authors:

Donald Herbert Kluemper Peter A. Rosen

University of Illinois at Chicago Pittsburg State University
33 PUBLICATIONS   907 CITATIONS    19 PUBLICATIONS   453 CITATIONS   

SEE PROFILE SEE PROFILE

All content following this page was uploaded by Peter A. Rosen on 05 September 2014.

The user has requested enhancement of the downloaded file.

 The current issue and full text archive of this journal is available at
www.emeraldinsight.com/0268-3946.htm

Employment
Future employment selection selection
methods: evaluating social methods
networking web sites
567
Donald H. Kluemper
Louisiana State University, Baton Rouge, Louisiana, USA, and
Peter A. Rosen
University of Evansville, Evansville, Indiana, USA

Abstract
Purpose – The use of social networking web sites (SNWs), like Facebook and MySpace, has become
extremely popular, particularly with today’s emerging workforce. Employers, aware of this
phenomenon, have begun to use the personal information available on SNWs to make hiring decisions.
The purpose of this paper is to examine the feasibility of using applicant personal information
currently available on SNWs to improve employment selection decisions.
Design/methodology/approach – A total of 378 judge ratings (63 raters £ 6 subjects) are
evaluated to determine if raters can reliably and accurately determine the big-five personality traits,
intelligence, and performance based only on information available on SNWs. Interrater reliability is
assessed to determine rater consistency, followed by an assessment of rater accuracy.
Findings – Based solely on viewing social networking profiles, judges are consistent in their ratings
across subjects and typically able to accurately distinguish high from low performers. In addition,
raters who are more intelligent and emotionally stable outperformed their counterparts.
Practical implications – Human resource (HR) professionals are currently evaluating social
networking information prior to hiring applicants. Since SNWs contain substantial personal
information which could be argued to cause adverse impact, academic studies are needed to determine
whether SNWs can be reliable and valid predictors of important organizational criteria.
Originality/value – This paper is the first, as far as the authors are concerned, to address the use of
SNWs in employment selection, despite their current utilization by HR practitioners.
Keywords Selection, Recruitment, Social networks, Internet
Paper type Research paper

Within the past few years, the phenomenon of social networking web sites (SNWs) on
the internet has exploded into the mainstream. Further, this online information has
begun to be used for purposes beyond its intended use. Owing to the vast amount of
personal information on these web sites, employers have begun to tap into this
information as a source of applicant data in an effort to improve hiring decisions. This
study evaluates the use of the SNWs in employment selection. Specifically, can trained
judges consistently and accurately assess important organizational characteristics
such as personality, intelligence, and performance using only a target’s SNW
information? In addition, the use of this information may lead to discrimination against Journal of Managerial Psychology
Vol. 24 No. 6, 2009
applicants, given the wide range of available personal information such as gender, pp. 567-580
race, age, religion, and disability status otherwise illegal to use when making q Emerald Group Publishing Limited
0268-3946
employment decisions. DOI 10.1108/02683940910974134
JMP Social networking web sites
24,6 SNWs focus on building online communities of people who share interests and
activities, or who are interested in exploring the interests and activities of others. Most
provide a variety of ways for users to interact, such as e-mail and instant messaging
services. SNWs are designed to connect users to each other and to visually display each
individual’s network of friends. The number of users for these web sites and the daily
568 traffic created by these web sites are staggering. According to the “About Us” section
of the various sites, MySpace is the largest with over 248 million registered users.
Other SNWs also have millions of users registered, such as Facebook (110 million),
Friendster (85 million), Hi5 (80 million), Orkut/Google (37 million), and LinkedIn
(25 million).
While these sites differ in the features that are available, most have a mechanism for
posting pictures, music and videos, keeping blogs, sharing links, and displaying
interests. These sites vary in user demographics. For example, although Facebook is
currently open to anyone, it started as a high school and college web site exclusively,
with about 90 percent of students registered for the site (van der Werf, 2006).

Social networking web sites in selection

Owing to the increasing prevalence of SNWs in conjunction with the large volume of
information available to the viewer, employers have begun using SNWs to assist in the
selection process for new employees. About 50 percent of the employers attending
college career fairs use online technology, including both search engines and SNWs to
screen candidates (Shea and Wesley, 2006). Currently, between 20 (Framingham, 2008)
and 25 percent (Taylor, 2007; NACE, 2006) of managers are using SNWs to screen
candidates, with 40 percent indicating that they are likely to use them within the next
year (Zeidner, 2007). These employers apparently feel justified in electronic screening
using SNWs. The president of a small Chicago consulting firm, when asked about an
applicant that applied for an internship and had questionable content on his page,
replied, “A lot of it makes me think, what kind of judgment does this person have? Why
are you allowing this to be viewed publicly?” (Finder, 2006).
While it may be common practice to monitor web site content it may not be legal.
Lance Chou, Director of Career Development at Stanford University, noted that:
[. . .] some employers might try and learn something about the student’s personality and
whether it would be appropriate for the job. However, there is information on Facebook that is
not relevant to the job, but may be used inappropriately by employers to assess a candidate
(Fuller, 2006).
According to George Lenard, an employment attorney, employers can inadvertently
learn about matters such as candidates’ age, marital status, and other topics typically
are off limits in job interviews, and organizations can be sued for discrimination if
these candidates are not hired (Frauenheim, 2006). The question of whether employer
monitoring of SNWs is illegal may relate to equal employment opportunity (EEO) law.
EEO severely limits the type of information an interviewer may ascertain and use. For
example, during interviews employers may not ask questions regarding race, religion,
sexual preference, or marital status. However, all of this information can be easily
located using SNWs (Kowske and Southwell, 2006). The ethics and legality behind
electronic screening using SNWs has become an important issue for human resource
(HR) practitioners (Zeidner, 2007).
 Since no academic studies, to our knowledge, have assessed whether using SNWs in Employment
employment selection are a reliable and valid predictor of important organizational selection
outcomes, their value in employment screening is unknown. Until the reliability and
validity of information on SNWs is established, employers should use caution when methods
using SNWs to make hiring decisions. Just as with structured interview methods
(Campion et al., 1997), focusing on job-related information in SNWs should minimize
the use of less job-relevant information that might bias the hiring decision. This study 569
focuses on showing the validity and reliability of three such categories of job-relevant
characteristics which have potential to be rated consistently and accurately and to
serve as valid predictors of job performance. The assessment of personality,
intelligence, and a global measure of performance through SNWs provide multiple
possible avenues regarding validity generalization of job-relevant characteristics, and
thereby justifying the use of SNWs in the screening process.

Assessment of personality
The 1990s have seen a huge growth in the use of personality assessment within
personnel selection practice and research (Barrick and Mount, 1991; Frei and McDaniel,
1997; Ones et al., 1993; Salgado, 1998; Tett et al., 1991). These studies provide positive
evidence for the criterion-related validity of personality. When it comes to the
prediction of overall job performance, conscientiousness was found to be the best
predictor, showing consistent predictions across all occupational groups. In addition,
extroversion, agreeableness, and neuroticism were shown to predict job performance in
certain jobs. Finally, although typically unrelated to job performance, openness to
experience has been found to predict training performance.
Beyond the predominant focus on self-reported personality assessment, personality
can also be measured using other-ratings. It is well established that people can assess
the personality of others, even after relatively short exposure, but that the accuracy of
these assessments depends on the information available to the observer (Barrick et al.,
2000). Observer ratings may be more valuable than self-ratings in employment
selection, particularly when targets are not available for self-reports, self-reports are
untrustworthy, and researchers wish to improve accuracy by aggregating multiple
raters (Hofstee, 1994; McCrae and Weiss, 2007). Other-rated personality via SNWs
seems quite promising in the selection context, since other-ratings of personality have
been found to predict job performance. Motowidlo and colleagues (1996) found that
interviewer-rated extroversion and conscientiousness (r ¼ 0.27 and 0.20, respectively)
significantly predict supervisor rated job performance. In addition, Mount et al. (1994)
show that observer ratings of extroversion and conscientiousness from supervisors,
coworkers, and customers significantly predicts sales performance, even beyond
self-rated personality.
The validity of other-rated personality, however, can depend on the relationship the
subject has with the rater and the quality of the information available to the rater.
A meta analysis conducted by Connolly and Viswesvaran (1998) show low accuracy
for strangers in predicting personality and moderate prediction by peers for each of the
big-five traits. In addition, Barrick et al. (2000) developed a personality-based job
interview for the purpose of assessing the personality of the applicant. They found that
personality-based job interviews could be used to accurately predict three of the
big-five. Considering the average interview is approximately 40 minutes in length
JMP (Campion et al., 1997), it appears that an interviewer can assess some aspects of
24,6 personality as effectively as a close acquaintance of the applicant. We propose that the
information available in SNWs provides a similar means to assess personality.
Furthermore, since SNW ratings are obtained from a wide range of personal
information that is a reflection of ongoing behaviors and interactions with other users
of the networks, web sites may actually provide unique information not found with
570 other selection methods. Recent issues of Personnel Psychology (Morgeson et al., 2007a,
b; Ones et al., 2007; Tett and Christiansen, 2007) and Industrial and Organizational
Psychology (Hough and Oswald, 2008; Griffith and Peterson, 2008) have focused on the
limited criterion validity of self-reported personality measures as well as the complex
issue of social desirability and faking of self-reported personality measures. To
increase validity and address issues of socially desirable responding “we should look
at other ways of assessing personality. There are a variety of ways of finding out about
people’s stable pattern of behavior” (Morgeson et al., 2007a, b, p. 719). SNWs may prove
to be an appropriate means of assessing personality in this way. Beyond personality
assessment, similar issues emerge in relation to a host of employment selection
methods. Resumes, interviews, job applications, and many other forms of employee
selection include a certain element of self-presentation, reflecting “maximal” instead of
“typical” work performance (Sackett, 2007; Sackett et al., 1988). Employment selection
methods using social networking are likely to be based on “typical” behaviors, and
therefore may be more accurate than other selection methods. At a minimum, this
method should provide information that is distinct from “maximal” selection methods,
thereby allowing for a stronger likelihood that using SNWs will yield incremental
validity beyond established methods of employment selection[1].
This is not to say that SNWs are not susceptible to manipulation and faking, similar
to that of self-report personality measures and job interviews. In fact, as users of SNWs
become more aware that their profiles are being evaluated by potential employers,
information provided on profiles is likely to be skewed in an effort to be viewed more
favorably. However, there are aspects of SNWs which would make the process of
skewing information difficult. Much of the information present in a given social
networking profile is submitted by other members of the network, such as tagged
photos and writing on another’s wall[2]. Although some negative information can be
deleted by the user, the user has more limited control over this aspect of their profile. In
addition, some of the information controlled by the users themselves would be difficult
to fake. For example, extroversion may be tied to the number of friends a user has in
the social network. Artificially inflating a substantial number of friends in the network
would pose great difficulty, as the user cannot control who accepts their friend request.
As another example, rater assessment of personality traits might be drawn in part
from photos, which are similarly difficult to fake. Thus, while faking would appear to
be an issue, it is likely that the impact of faking is less than with other selection
methods. Future research should assess the impact of faking in the context of SNWs.
Despite the potential for faking, Vazire and Gosling (2004) used personal web sites
to accurately assess personality. Although personal web sites are similar in some ways
to SNWs, they are used by such a small percentage of potential applicants that they are
impractical for purposes of employment selection. In addition, SNWs provide
additional information not included in personal web sites, such as a list of the user’s
friends and a list of the interest groups a user has joined. However, Vazire and Gosling
provide initial evidence for the accurate prediction of personality using personal Employment
information available on the internet. selection
The types of information available on SNWs may be particularly effective in
predicting the big-five personality traits. SNWs contain various sources of information methods
which could be used to assess behaviors related to personality. For example, the types
and number of interest groups the user has joined, comments that have been left for the
user, comments made by the user on other people’s “walls,” “tagging” photos, updating 571
“status messages”[3], and listing books and intellectual interests in the “personal
information” section[4]. These examples provide only a very preliminary introduction
to the various sources of personal information available in SNWs which might indicate
an individual’s personality. It should also be noted that the use of SNWs may vary
based on characteristics beyond personality. For example, an individual who is more
adept with this form of technology may be more likely to participate in social
networking and may post more information more frequently than those individuals
lacking in these technological skills. Since age has been shown to relate to technology
acceptance and use (Morris and Venkatesh, 2000), this also brings the issue of age into
the possible groups which could be adversely impacted by the use of this type of
technology by HR departments in hiring. Future research should explore this issue
further.

Assessment of intelligence
Since the very earliest research on personnel selection, cognitive ability has been one of
the major methods used to attempt to discriminate between candidates and to predict
subsequent job performance (Robertson and Smith, 2001). Intelligence is the single
most effective predictor known of individual performance at school and on the job
(Gottfredson, 1998), accounting for approximately 25 percent of the variance in job
performance (Hunter and Hunter, 1984). Cognitive ability provides criterion-related
validity that generalizes across more or less all occupational areas (Robertson and
Smith, 2001). In addition, judge ratings of intelligence have been shown to predict
intelligence test scores (Borkenau et al., 2004). Furthermore, biographical data have
been shown to predict intelligence (Schmidt and Hunter, 1998). These results provide
some evidence that assessment of intelligence should be viable within the context of
SNWs.

Assessment of global performance

Beyond personality and intelligence, SNWs may also contain additional information
which may be useful in employment selection. Owing to the large volume of
information contained in SNWs, information may also be obtained which relate to the
user’s writing skills, job experiences, or a variety of knowledge, skills, abilities, or other
criteria which might relate to job or organizational fit in a given employment selection
context. A more global assessment of performance would include a variety of
information. In fact, due to the broad range of information available on SNWs and the
lack of consistency in this information across individuals, the approach of assessing
broad characteristics is likely to be more practical than assessing more narrow aspects
of social networking profiles that may be unavailable and/or inconsistent for a large
segment of the profiles.
JMP Rater consensus/interrater agreement
24,6 Furr and Funder (2007) identify rater consensus as the first psychometric concern when
assessing characteristics through judge-rated behavioral observation. Thus, the
primary focus of this paper, as the first to address the potential use of SNW information
for the purpose of employment selection, is to focus on the issue of rater consensus.
To the degree to which SNWs provide a distinguishable and consistent basis of
572 evaluation, judges will reach a certain level of consensus in their impressions of these
SNWs. Thus, consensus is expected across judges:
H1. Ratings of the big-five dimensions of personality, intelligence, and
performance based on SNW information are consistent across raters.

Accuracy of ratings
A second psychometric concern is that of validity (Furr and Funder, 2007). If rater
consensus is established, the next step will be to establish various forms of validity.
This study evaluates validity by assessing whether judges are accurate in their
assessment of personality intelligence, and performance.
Personality, intelligence, and performance impressions based on judge assessments
have been shown to be quite accurate, even when the amount of information is limited
(Borkenau et al., 2004). This growing body of research suggests that people have a
natural talent for judging one another accurately (Vazire and Gosling, 2004). Given the
high volume of information available to assess behavioral cues on SNWs, a suitable
level of accuracy is expected:
H2. Raters assessing an individual’s personality, intelligence, and performance
through SNWs are able to distinguish between those individuals who are high
on each characteristic from those who are low on that characteristic.

Method
Participants and procedures
This study was conducted at a large public university in the southern USA. A total of
63 students enrolled in an employment selection course participated in this project for
course credit. Participants were 49 percent male, 90 percent Caucasian, averaged
24 years of age, and worked an average of 26 hours per week. Participants had
prerequisites in HRs and statistics. As part of the employment selection course, these
participants were trained in both personality/intelligence testing and effective
utilization of rating scales, participated in a one hour training session for this project
(reviewing the definitions of the big-five personality traits, general mental ability, and
academic performance; viewing Facebook profiles to identify specific information which
could be used to assess the focal characteristics of the study; and familiarizing the
participants with the rating form to be used when conducting the assessments), and
participated in a series of practice assessments prior to conducting the ratings for this
study (an assignment to evaluate SNWs and identify specific information that could be
used to assess each of the focal characteristics of the study, follow-up class discussion of
these observations, and a practice session in which each participant conducted two
assessments of current Facebook users by using the researcher designed rating form).
All participants had personal involvement with SNWs. Participants were asked to spend
ten minutes evaluating each of the six social networking profiles, consider multiple
aspects of the profiles which could relate to a specific trait, then complete the rating form Employment
based on their overall impression of the social networking profile. selection
The choice of the three male and three female social networking profiles was
randomly generated from a list of volunteers in an introductory management course. methods
Along with volunteering to have their SNWs evaluated[5], the volunteers also completed
demographics and personality questionnaires, an intelligence test, and consented to
allow the researchers to obtain their grade point average (GPA) from the university 573
registrar.

Measures
Judge ratings of big-five personality traits – measured with 25 items from the bipolar
adjective checklist (Goldberg, 1992) on a nine-point scale.
Judge ratings of general mental ability – a single item measure was used to measure
intelligence quotient (IQ) based on Reilly and Mulhern (1995). Judges were asked to
“Estimate the user’s IQ. Remember that the average IQ is 100, and one-sixth of the
population have IQs less than 85, with one-sixth scoring over 115.”
Judge ratings of performance – a single item measure was used to measure academic
performance based on the format used to measure IQ. Judges were asked to “Estimate
the user’s GPA. Remember that an average GPA is 3.0 and the maximum is 4.0.”
Ratee self-reported big-five personality traits (referred to as true scores) – measured
with 150 items from the international personality item pool – IPIP (Goldberg et al.,
2006) on a five-point scale ranging from strongly disagree to strongly agree. The as
ranged from 0.92 for conscientiousness to 0.81 for openness.
Ratee general mental ability (IQ true score) – measured with the Wonderlic
personnel test (Wonderlic, 2000), a 12 minute/50 question timed test of intelligence.
Ratee performance (performance true score) – academic performance was obtained
via GPA from the University Registrar. Although academic performance is less ideal
than job performance in the context of employment selection, it represents an objective
measure to test the hypotheses presented.

Results
The as for the judge ratings of personality were calculated for each of the six ratees.
These six as were then averaged for each of the big-five to estimate the overall internal
consistency of the scales. In order to assess H1, interrater agreement in the form of
average measures intraclass correlation coefficients (ICCs) for the judge ratings are
included in Table I. The scaled scores for the big-five personality traits and the
single item scores for IQ and performance were evaluated for interrater agreement.
The 378 total ratings (63 raters £ 6 ratings each) were used to calculate the ICCs.
The ICC values were all adequate, ranging from 0.93 for extroversion to 0.99 for
conscientiousness and performance. Since ICCs are expected to be higher with a larger
number of raters, Table II also includes the number of raters for each characteristic
which would be necessary to achieve a 0.50 ICC value. Although there are no guidelines
for level of agreement, 0.50 was used in the analyses as it should provide a minimum
level of acceptable agreement across judges. The Spearman Brown prophecy formula
was used to determine how many raters would be required to obtain an adequate (0.50)
ICC value. Based on the 63 raters from this study, it was determined that between
two (for conscientiousness and performance) and six (for emotional stability
JMP
a ICC No. of raters
24,6
Conscientiousness 0.92 0.99 2
Emotional stability 0.83 0.94 6
Agreeableness 0.80 0.98 3
Openness to experience 0.87 0.98 3
574 Extroversion 0.85 0.93 6
IQ 0.98 3
Performance 0.99 2
Table I. Notes: n ¼ 378 ratings (63 raters £ 6 ratings each); number of raters indicates the number of raters
Judge rating as and ICCs which would be required in a future study in order to obtain a 0.50 ICC value

High ratee score Low ratee score

n S# SS RM S# SS RM T

Conscientiousness 63 6 4.07 8.03 2 3.50 7.65 2.77 *

Emotional stability 63 5 3.33 6.29 3 2.17 5.80 3.00 *
Agreeableness 63 6 4.27 8.04 5 3.57 7.26 6.12 *
Openness 63 3 3.78 5.51 1 3.46 6.37 6.34 *
Extroversion 63 1 4.10 7.65 2 3.40 6.85 5.99 *
Table II. IQ 63 6 26 110.4 3 17 94.7 13.53 *
Differences in judge Performance 63 6 3.94 3.57 2 1.81 3.46 2.78 *
ratings means comparing
high and low ratee scores Notes: *p , 0.05; S#, subject; SS, subject score (true scores); RM, rater mean score

and extroversion) raters would be required to obtain a satisfactory level of interrater

agreement.
H2 was evaluated by conducting t-tests on score means in order to determine whether
or not the means are statistically different from one another. In order to determine which
means to test, the true scores (self-reported big-five personality scores, intelligence
scores, and GPA) of the six rated subjects were evaluated. For each of the seven
characteristics, the individual with the highest true score and the individual with the
lowest true score were selected for analysis. Judge mean ratings for these subjects were
then compared to determine whether or not raters are able to distinguish individuals
high on a characteristic from those low on the same characteristic. This method also
allows for evaluation of the direction of the relationship, such that (in addition to
evaluating mean differences) the judge rating of the subject with the higher true score
should be higher than the judge rating of the lower true score. Results demonstrate that
the mean judge ratings for the subject highest on the seven characteristics were
statistically different from the subjects lowest on those characteristics. In addition, with
the exception of openness to experience, the judges mean ratings were higher for those
with the highest true score, indicating the ability of judges to distinguish the traits of
conscientiousness, emotional stability, agreeableness, extroversion, intelligence, and
performance by evaluating SNWs[6].
Post hoc analyses were conducted to determine the impact of intelligence and
personality on judge consistency and accuracy. Prior research has demonstrated mixed
findings related to the impact of personality traits of the rater on rating accuracy.
Ambady et al. (1995) found that less sociable (extroverted) raters were more accurate, Employment
while Lippa and Dietz (2000) found that only openness indicated more accurate raters. In selection
addition, narcissistic raters have been found to be less accurate (John and Robbins, 1994),
which may be relevant to the big-five since narcissism relates strongly to neuroticism. methods
Finally, intelligence has also been reported to positively relate to rater accuracy (Lippa
and Dietz, 2000). In the current study, the 63 judges were asked to take the same
intelligence and personality tests as the SNW subjects. The analyses conducted above 575
were then re-evaluated based on high versus low groups based on intelligence and the
big-five. Results show no difference in interrater agreement based on these
characteristics. However, judges who are more intelligent and more emotionally
stable were shown to be more accurate in their judgments. More specifically, when the
raters were split into high and low groups based on intelligence scores (the 31 highest
scores versus the 31 lowest scores), the high intelligence group significantly and more
accurately differentiated between high and low characteristics for conscientiousness,
emotional stability, openness, and performance. For example, with all 63 raters
combined, the difference between rater means for conscientiousness in Table II is 0.38
(8.03 for the high ratee score and 7.65 for the low ratee score). When assessing high and
low intelligence raters independently, the mean difference for the 31 high intelligence
raters is 0.61, but only 0.14 for the 31 low intelligence raters. Thus, more intelligent raters
seem to be more capable of assessing this trait than less intelligent raters. Similarly,
raters who are the most emotionally stable also rate more accurately for
conscientiousness, emotional stability, openness, and performance. For example, the
mean difference across raters for high and low ratee conscientiousness is again 0.38, but
is 0.73 for the 31 raters who are the most emotionally stable and 0.03 for the 31 raters who
are the least emotionally stable. These results indicate the potential need for researchers
to consider intelligence and emotional stability when selecting individuals who will
serve as raters of characteristics such as personality.

Discussion
Based on the large volume of personal information available on SNWs, judges’ ratings
of the big-five dimensions of personality, intelligence, and global performance were
consistent across the 63 raters in this study, demonstrating adequate internal
consistency reliability and interrater agreement. In addition, the trained raters were
able to accurately distinguish between individuals who scored high and individuals
who scored low on four of the big-five personality traits, intelligence, and performance,
providing initial evidence that raters can accurately determine these organizationally
relevant traits by viewing SNW information.
As stated earlier, other rated personality has been shown to predict job
performance. Considering that other methods of other-reported personality are unlikely
to be viable in an employment selection context, SNW ratings of personality may be a
practical approach. Owing to the theoretical and methodological differences between
self-reported and other-rated personality, it is likely that ratings of personality via
SNWs will provide a context for incremental prediction of job performance beyond the
predominant self-report approach. In addition, the differences in context between
SNWs and a job interview (i.e. socially desirable responding in the job interview as well
as the unique nature of information contained in SNWs) should similarly allow for
unique prediction of job performance beyond what can be evaluated through
JMP personality assessment in the employment interview. This approach may be
24,6 particularly valuable since these assessments take only a fraction of the time involved
with other selection methods.
This study is not without limitations. Although the analyses testing the consistency
of the relationships of SNW ratings are based on 378 judge ratings from 63 raters, the
analyses testing rater accuracy were conducted by testing for significant differences
576 between the high and low performer on the seven characteristics for only six subjects.
Future research should assess accuracy over a larger sample of subjects.
We hope that the results of this preliminary study will not be used by organizations
to support their use of SNWs in employment selection. Without further validation in a
variety of studies, with larger samples and in a variety of organizational contexts,
caution should be used when interpreting the implications of this study. This is
particularly true given the potential for employer legal liability due to the vast amount
of personal information available on SNWs. Information regarding gender, race, age,
disabilities, and other criteria which should not be used when making hiring decisions
will most certainly, consciously or not, influence who gets hired. Even if this
information does not bias the hiring decision, disparate impact issues may still exist.
Future research should also examine the potential issues of adverse impact and
potentially illegal information in hiring decisions using personal information from
SNWs. In addition, research should be conducted to compare assessments of SNWs to
other employment selection methods, such as personality assessment, intelligence
testing, and employment interviews.
Based on the relative absence of research evidence in this newly developing area,
particularly regarding the potential for adverse impact and the lack of validity evidence,
we believe the most important practical implication of this paper is for organizations to
use SNWs with these issues in mind. Organizational representatives assessing SNWs
should ask themselves two important questions. First, is the organization assessing (or
could be perceived as assessing) information which could lead to discrimination against
a legally protected group? Second, is the specific social networking information used to
help make a hiring decision valid in determining who will perform better on the job? The
approach used in this paper of assessing personality traits, intelligence, or general
performance begin to provide answers to these questions.

Notes
1. Special thanks to an anonymous reviewer for pointing out the “maximal”/“typical”
performance distinction.
2. A wall, similar to a guestbook on other web sites, is a forum for the friends of the user to post
comments to the user in an open forum where all of the user’s friends can see. This is
different from a message that goes only to the user, and cannot be seen by others. By tagging
a photo, both the user and the user’s friends have the ability to indicate that the particular
user appears in a photo. Both user posted photos and photos where the user has been tagged
appear on the users profile in the photos section. The user has the ability to “untag” or
remove the link between the picture and his or her profile, but the picture will still remain on
the page of the friend who uploaded the picture.
3. Status messages are a way for the user to briefly tell their friends what they are currently
doing, how they are feeling, or any other short message that they would like to convey to
their friend list. Status messages are often times the area of the social networking page that
 gets updated most frequently, with some users changing their status message multiple times Employment
a day.
selection
4. Personal information lists the user’s activities, interests, and favorite movies, books,
television shows, and quotations, and is a way for the user’s friends to get a better methods
understanding of the user.
5. It is unknown whether or the degree to which these participants made modifications to their
Facebook profile after agreeing to allow their profiles to be evaluated for research purposes. 577
However, the potential for altering a profile parallels situations in which these Facebook
users might choose to alter their profile when applying for a job, since it is now widely
known that employers may potentially assess these profiles during the employment
selection process. Future research should assess the degree to which SNW users attempt to
modify their profiles in these situations.
6. Based on a suggestion from one of the anonymous reviewers, we also assessed H2 based on
the magnitude of the correlation coefficients between Facebook-ratings and “true scores,”
since this approach includes all six of the Facebook users in the sample instead of just the
high and low performer for each characteristic. Owing to the extremely small sample size,
this approach is problematic, even considering the large number of items and raters used to
generate the scores. However, due to the novelty of the research question and the exploratory
nature of this study, we agree that this analysis may provide additional insight into the
proposed relationships. Results indicate that four of the seven proposed relationships have
medium to large effect sizes (Cohen, 1988). Specifically, conscientiousness (0.40),
agreeableness, (0.38), extroversion (0.52), and performance (0.32). Emotional stability and
IQ were small (0.07 and 2 0.01, respectively), while openness was once again negatively
associated. These results provide additional evidence that measuring job-relevant
characteristics using SNWs may be a valid method of assessment.

References
Ambady, N., Hallahan, M. and Rosenthal, R. (1995), “On judging and being judged accurately in
zero-acquaintance situations”, Journal of Personality and Social Psychology, Vol. 69,
pp. 518-29.
Barrick, M.R. and Mount, M.K. (1991), “The big five personality dimensions and job performance:
a meta-analysis”, Personnel Psychology, Vol. 44, pp. 1-26.
Barrick, M.R., Patton, G.K. and Haugland, S.N. (2000), “Accuracy of interviewer judgements of
job applicant personality traits”, Personnel Psychology, Vol. 53, pp. 925-51.
Borkenau, P., Mauer, N., Rieman, R., Spinath, F.M. and Angleitner, A. (2004), “Thin slices of
behavior as cues of personality and intelligence”, Journal of Personality and Social
Psychology, Vol. 86, pp. 599-614.
Campion, M.A., Palmer, D.K. and Campion, J.E. (1997), “A review of structure in the selection
interview”, Personnel Psychology, Vol. 50, pp. 655-702.
Cohen, J. (1988), Statistical Power Analyses for the Behavioral Sciences, Lawrence Erlbaum
Associates, Hillsdale, NJ.
Connolly, J.J. and Viswesvaran, C. (1998), “The convergent validity between self- and observer
ratings of personality”, poster (71-08) presented at the 13th Annual Conference of the
Society for Industrial and Organizational Psychology, Dallas, TX.
Finder, A. (2006), “For some, online persona undermines a resume”, New York Times, available
at: www.nytimes.com/2006/06/11/us/11recruit.html?scp¼1&sq¼ForþSome%2CþOnline
þPersonaþUnderminesþa þ R%E9sum%E9 (accessed January 10, 2008).
JMP Framingham, H.H. (2008), “Employers use social networks in hiring process”, Computerworld,
available at: http://computerworld.co.nz/news.nsf/care/63C6E9BE6A2CD920CC2574C9000
24,6 3ADDD (accessed October 16, 2008).
Frauenheim, E. (2006), “Caution advised when using social networking web sites for recruiting,
background checking”, Workforce Management, Vol. 24, available at: www.workforce.
com/archive/feature/24/58/49/245851.php (accessed October 15, 2008).
578 Frei, R.L. and McDaniel, M.A. (1997), “Validity of customer service measures in personnel
selection: a review of criterion and construct evidence”, Human Performance, Vol. 11,
pp. 1-27.
Fuller, A. (2006), “Employers snoop on Facebook”, Stanford Daily, available at: http://daily.
stanford.edu/article/2006/1/20/employersSnoopOnFacebook (accessed January 10, 2008).
Furr, R.M. and Funder, D.C. (2007), “Behavioral observation”, in Robins, R.W., Fraley, R.C. and
Krueger, R.F. (Eds), Handbook of Research Methods in Personality Psychology, Guilford
Press, New York, NY.
Goldberg, L., Johnson, J., Eber, H., Hogan, R., Ashton, M., Cloninger, C., Gough, H. and Johnson, J.
(2006), “The international personality item pool and the future of public-domain
personality measures”, Journal of Research in Personality, Vol. 40, pp. 84-96.
Goldberg, L.R. (1992), “The development of markers for the big-five factor structure”,
Psychological Assessment, Vol. 4, pp. 26-42.
Gottfredson, L.S. (1998), “The general intelligence factor”, Scientific American Presents
Intelligence, Vol. 9, pp. 24-9.
Griffith, R.L. and Peterson, M.H. (2008), “The failure of social desirability measures to capture
applicant faking behavior”, Industrial and Organizational Psychology: Perspectives on
Science and Practice, Vol. 1, pp. 308-11.
Hofstee, W.K. (1994), “Who should own the definition of personality?”, European Journal of
Personality, Vol. 8, pp. 149-62.
Hough, L.M. and Oswald, F.L. (2008), “Personality testing and industrial-organizational
psychology: reflections, progress, and prospects”, Industrial and Organizational
Psychology: Perspectives on Science and Practice, Vol. 1, pp. 272-90.
Hunter, J.E. and Hunter, R.F. (1984), “Validity and utility of alternative predictors of job
performance”, Psychological Bulletin, Vol. 96, pp. 72-98.
John, O.P. and Robins, R.W. (1994), “Accuracy and bias in self-perception: individual differences
in self-enhancement and narcissism”, Journal of Personality and Social Psychology, Vol. 66,
pp. 206-19.
Kowske, B. and Southwell, M. (2006), “E-screening proves ‘e-resistible’: but at what cost?”,
Human Resource Executive Online, available at: www.hreonline.com/HRE/story.jsp?
storyId¼6835642 (accessed January 10, 2008).
Lippa, R.A. and Dietz, J.K. (2000), “The relation of gender, personality, and intelligence to judges’
accuracy in judging strangers’ personality from brief video segments”, Journal of
Nonverbal Behavior, Vol. 24, pp. 25-43.
McCrae, R.R. and Weiss, A. (2007), “Observer ratings of personality”, in Robins, R.W., Fraley, R.C.
and Krueger, R.F. (Eds), Handbook of Research Methods in Personality Psychology, Guilford
Press, New York, NY.
Morgeson, F.P., Campion, M.A., Dipboye, R.L., Hollenbeck, J.R., Murphy, K.R. and Schmitt, N.
(2007a), “Are we getting fooled again? Coming to terms with limitations in the use of
personality tests for personnel selection”, Personnel Psychology, Vol. 60, pp. 1029-49.
Morgeson, F.P., Campion, M.A., Dipboye, R.L., Hollenbeck, J.R., Murphy, K.R. and Schmitt, N. Employment
(2007b), “Reconsidering the use of personality tests in personnel selection contexts”,
Personnel Psychology, Vol. 60, pp. 683-729. selection
Morris, M.G. and Venkatesh, V. (2000), “Age differences in technology adoption decisions: methods
implications for a changing workforce”, Personnel Psychology, Vol. 53, pp. 375-403.
Mount, M.K., Barrick, M.R. and Strauss, J.P. (1994), “Validity of observer ratings of the big five
personality factors”, Journal of Applied Psychology, Vol. 79, pp. 272-80. 579
National Association of Colleges and Employers Newsletter (2006), “Quick poll results: more than
one-quarter of organizations have ‘Googled’/reviewed candidate profiles on social
networking sites”, available at: www.naceweb.org/pubs/spotlightonline/2006/c070606.
htm#1 (accessed January 10, 2008).
Ones, D.S., Viswesvaran, C. and Schmidt, F.L. (1993), “Comprehensive meta-analysis of integrity
test validities: findings and implications of personnel selection and theories of job
performance”, Journal of Applied Psychology, Vol. 83, pp. 35-42.
Ones, D.S., Dilchert, S., Viswesvaran, C. and Judge, T.A. (2007), “In support of personality
assessment in organizational settings”, Personnel Psychology, Vol. 60, pp. 995-1027.
Reilly, J. and Mulhern, G. (1995), “Gender difference in self-estimated IQ: the need for care in
interpreting group data”, Personality and Individual Differences, Vol. 18, pp. 189-92.
Robertson, I.T. and Smith, M. (2001), “Personnel selection”, Journal of Occupational &
Organizational Psychology, Vol. 74, pp. 441-72.
Sackett, P.R. (2007), “Revisiting the origins of the typical-maximum performance distinction”,
Human Performance, Vol. 20, pp. 179-85.
Sackett, P.R., Zedeck, S. and Fogli, L. (1988), “Relations between measures of typical and
maximum job performance”, Journal of Applied Psychology, Vol. 73, pp. 482-6.
Salgado, J.F. (1998), “Big five personality dimensions and job performance in army and civil
occupations: a European perspective”, Human Performance, Vol. 11, pp. 271-88.
Schmidt, F.L. and Hunter, J.E. (1998), “The validity and utility of selection methods in personnel
psychology: practical and theoretical implications of 85 years of research findings”,
Journal of Applied Psychology, Vol. 124, pp. 262-74.
Shea, K. and Wesley, J. (2006), “How social networking sites affect students, career services, and
employers”, National Association of Colleges and Employers Journal, Vol. 66 No. 3, pp. 26-32.
Taylor, A. (2007), “Careless online talk can cost candidates jobs”, Financial Times, available at:
www.ft.com/cms/s/0/0dd40300-dcca-11db-a21d-000b5df10621.html (accessed January 10,
2008).
Tett, R.P. and Christiansen, N.D. (2007), “Personality tests at the crossroads: a response to
Morgeson, Campion, Dipboye, Hollenbeck, Murphy, and Schmitt (2007)”, Personnel
Psychology, Vol. 60, pp. 967-93.
Tett, R.P., Jackson, D.N. and Rothstein, M. (1991), “Personality measures as predictors of job
performance: a meta-analytic review”, Personnel Psychology, Vol. 44, pp. 703-42.
van der Werf, M. (2006), “Beware of using social-networking websites to monitor students,
lawyers say”, Chronicle of Higher Education, Vol. 53 No. 26, p. A28.
Vazire, S. and Gosling, S.D. (2004), “E-perceptions: personality impressions based on personal
websites”, Journal of Personality and Social Psychology, Vol. 87, pp. 123-32.
Wonderlic & Associates (2000), Wonderlic Personnel Test and Scholastic Level Exam: User’s
Manual, Wonderlic & Associates, Libertyville, IL.
Zeidner, R. (2007), “How deep can you probe?”, HR Magazine, Vol. 52 No. 10, pp. 57-62.
JMP Further reading
24,6 McCain, J. (2008), “McCain says using Google to vet VP candidates”, Yahoo News, available at:
http://news.yahoo.com/s/nm/20080609/wr_nm/usa_politics_mccain_google_dc (accessed
June 12, 2008).
Motowidlo, S.J., Burnett, J.R., Maczynski, J., Witkowski, S., Wojtachnio, A. and Chelpa, S. (1996),
“Predicting managerial job performance from personality ratings based on structured
580 interview: an international replication”, Polish Psychological Bulletin, Vol. 27, pp. 139-51.

About the authors

Donald H. Kluemper is an Assistant Professor and Robert and Patricia Hines Professor of
Management at Louisiana State University. He received his PhD from the Department of
Management at Oklahoma State University. He is a member of the Academy of Management,
American Psychological Association, Southern Management Association, and the Society for
Industrial and Organizational Psychology. His research interests include employment selection,
personality testing, intelligence testing, and emotional intelligence. His research has been
published in the Journal of Organizational Behavior, International Journal of Selection and
Assessment, and Personality and Individual Differences. Donald H. Kluemper is the
corresponding author and can be contacted at: [email protected]
Peter A. Rosen is an Assistant Professor of Management Information Systems at the
University of Evansville, Indiana. He received his PhD from the Department of Management
Science and Information Systems at Oklahoma State University. He is a member of the Decision
Sciences Institute, Association for Information Systems, INFORMS, and the Academy of
Management. His research interests include social networking technology, data privacy and
security, technology acceptance, personal innovativeness, and statistics in sports. His research
has been published in the Journal of Database Management, the International Journal of
Information and Computer Security, the Journal of Quantitative Analysis in Sports, and the
Journal of the Academy of Business Education.

To purchase reprints of this article please e-mail: [email protected]

Or visit our web site for further details: www.emeraldinsight.com/reprints

View publication stats