Hiring For Retention and

HIRING FOR RETENTION AND
PERFORMANCE
M U R R AY R . B A R R I C K A N D R YA N D . Z I M M E R M A N
This study evaluated the usefulness of several pre-hire variables to predict

voluntary turnover and job performance. Analyses showed that applicants
who knew current employees, had longer tenure with previous employers,
were conscientious and emotionally stable, were motivated to obtain the job,
and were confident in themselves and their decision making were less likely
to quit, and had higher performance within six months after hire. Results also
indicated that pre-hire attitudes (employment motivation and personal con-
fidence) did not predict turnover and performance beyond biodata (pre-hire
embeddedness in the organization and habitual commitment) and the per-
sonality traits (conscientiousness and emotional stability). For all predictors
but personality, the strength of the relationships weakened over time up to
two years after hire. Nonetheless, organizations can avoid voluntary turnover
and increase performance by basing hiring decisions on the set of predictors
analyzed in this study. © 2009 Wiley Periodicals, Inc.
Keywords: turnover, job performance, conscientiousness, emotional

stability, confidence, personality traits, biodata, selection, staffing
hen making hiring decisions, Hand, & Meglino, 1979; Schmidt & Hunter,
W organizations historically
have focused on determin-
ing which job candidate will
likely be the best performer.
As it becomes more difficult to retain em-
ployees, organizations have also started to
try to determine which candidates are most
1998), it is important to consider which
individual difference variables have significant
effects on both criteria.
The vast majority of selection research
has focused on pre-hire predictors of job
performance. In contrast, very little re-
search, excluding biographical information
likely to stay with the organization. Finding (biodata), has investigated whether employ-
constructs that predict both performance ers can prevent turnover before employees
and turnover enables organizations to use start their jobs. One recent exception is a
fewer resources when selecting applicants. study by Barrick and Zimmerman (2005),
Since individual differences play an im- which investigated the predictive validity of
portant role in established models of both three theoretically relevant sets of turnover
turnover and performance (Johnson, 2003; predictors. Analyses confirmed that biodata
March & Simon, 1958; Mobley, Griffeth, and other retention-related predictors were
Correspondence to: Murray R. Barrick, Mays Business School, Texas A&M University, 440 Wehner, 4113 TAMU,
College Station, Texas 77843-4113, Phone: 979-845-0329, E-mail: mbarrick@mays.tamu.edu
Human Resource Management, March–April, Vol. 48, No. 2, Pp. 183– 206
© 2009 Wiley Periodicals, Inc.
Published online in Wiley InterScience (www.interscience.wiley.com).
DOI: 10.1002/hrm.20275
184 HUMAN RESOURCE MANAGEMENT, MARCH–APRIL 2009
related to employee turnover (R=.33) within Sablynski, & Erez, 2001) have examined how
six months after hire. What is virtually un- employee embeddedness can influence their
mentioned in the literature is whether pre- desire to leave, whether through the employ-
hire predictors of turnover would also be ees’ perceptions of how well they fit with
effective predictors of work performance. their jobs and organizations or the number of
By examining the validity of pre-hire pre- links the employees have with other individu-
dictors of voluntary turnover, this study builds als within their work environments. Other
on prior research, particularly Barrick and researchers (Maertz & Campion, 2004; Maertz
Zimmerman (2005). It differs, however, in & Griffeth, 2004) have focused on examining
that it also examines whether predictors that motivational forces that influence employees’
are theoretically relevant to turnover also desire to remain in their jobs. These forces
would relate to job performance as measured include organization-related factors, such as
by supervisory ratings. Since job performance employees enjoying the jobs they hold, liking
and voluntary turnover are two of the most the organizations for which they work, or
important criteria affected by em- wanting to continue working with coworkers
ployee behavior, they warrant si- with whom they have close relationships.
The vast majority of
multaneous study in determining These forces also include individual-related
selection research joint causes of both outcomes. factors, such as feelings of obligation to an
This study also investigates a employer who provides them with a job, an
has focused on broader set of predictors, including innate sense of responsibility not to quit their
relevant personality traits and an job, or even an escalation of commitment to
prehire predictors
additional biodata construct to test their decision to work for the organizations
of job performance. whether individuals who have a that employ them. Although some factors
history of job hopping will be that influence the desire to leave (or stay) can
In contrast, very more likely to leave their current only be determined once individuals start
organization (Ghiselli, 1974). It their job and experience their work environ-
little research. . . . also examines whether these rela- ment, other factors may be ascertained even
has investigated tionships exist long after the ap- before they are hired. Finally, some of these
plicant is hired (up to two years), same researchers have suggested that the at-
whether employers not just shortly after hire (up to six tributes that influence employees’ attachment
months). Thus, the purpose of this to their organizations may also influence
can prevent turnover study is to systematically explore their motivation to perform well in their jobs
the ability of each pre-hire variable (Lee et al., 2004; March & Simon, 1958).
before employees
to predict both voluntary turnover Therefore, the predictors used in this study
start their jobs. and performance over an extended were selected based on their theoretical rela-
period of time. tionship to factors that would influence em-
ployees’ desire to stay, as well as affect their
Theoretical Foundation level of job performance.
for Pre-hire Predictors of Voluntary
Turnover and Job Performance Effects of Biodata on Employees’
Turnover Decisions and Job
An overarching theme common to most turn-
Performance
over theories is that some employees have
greater feelings of attachment to their organi- Recent research using biodata has attempted
zations than others do. Perhaps the first time to create a theoretical basis for the items used
this idea was codified in the management lit- (Dean, Russell, & Muchinsky, 1999). Barrick
erature was with March and Simon’s (1958) and Zimmerman (2005) examined three the-
discussion of factors influencing employees’ oretically relevant biodata items (number of
desire to leave their organizations. Later friends and family working at the firm,
researchers (Lee, Mitchell, Sablynski, Burton, referral by an employee, and tenure in prior
& Holtom, 2004; Mitchell, Holtom, Lee, job) and found that the items predicted
Human Resource Management DOI: 10.1002/hrm

Hiring for Retention and Performance 185
voluntary, avoidable turnover (R =.31). The seeking new jobs, as represented by how long
theoretical basis of these hypothesized effects they stayed in the previous job and the num-
is tied to the job embeddedness literature ber of jobs held over the previous five years,
(Mitchell et al., 2001), realistic job preview could be expected to do so again. According
literature (Breaugh & Dossett, 1989; Premack to the unfolding model of turnover (Holtom,
& Wanous, 1985; Rynes, 1991), and the find- Mitchell, Lee, & Inderrieden, 2005; Lee &
ing that past behavior is the best predictor of Mitchell, 1994), some individuals are likely
future behavior (Owens & Schoenfeldt, 1979; to be impulsive quitters who terminate their
Wernimont & Campbell, 1968). Supplement- employment spontaneously. Similarly, Ghis-
ing the three items that Barrick and Zimmer- elli (1974) noted that some individuals are
man used, we add a fourth biodata item: the more likely to develop a habit of quitting job
influence of frequent job changes. In this after job, which he termed “hobo syndrome.”
study, two measures composed of two bio- Empirical research by Judge and Watanabe
data items each are used to reflect pre-hire (1995) has supported this contention.
embeddedness in the organization (employee The four biodata items used in
referral; number of friends and family) and this study also are likely to corre-
Since past behavior
habitual commitment (tenure in prior job; late with job performance. In a
number of jobs in last five years). recent study on job embedded- is the best predictor
Biodata items assessing the impact of ness, Lee et al. (2004) theorized
whether current employees referred the appli- that employees with a greater of future behavior,
cants and whether they have friends or rela- number of links to other cowork-
individuals who
tives working at the organization have their ers would be more likely to be
theoretical basis tied to the job embeddedness motivated to perform than em- have a habit of
literature (Mitchell et al., 2001), which suggests ployees with fewer links. Further-
that the greater the number of links to the or- more, they suggested that the seeking new jobs, as
ganization, the less likely employees are to more individuals are socially en-
represented by how
quit. In addition, if applicants have contacts meshed in an organization, the
within the organization, they are more likely more likely they are to engage in long they stayed
to understand the advantages and disadvan- contextual performance. Embed-
tages of the position for which they are apply- ded employees are likely to have in the previous job
ing. Hence, they are better able to engage in larger social networks from which
self-selection (Rynes, 1991; Wanous, 1980), to obtain assistance in performing and the number of
because their perceptions of fit with the job their jobs effectively (Settoon, Ben- jobs held over the
and the organization (Hom & Griffeth, 1995; nett, & Liden, 1996). Lee et al.
Premack & Wanous, 1985) are better informed. found that on-the-job embedded- previous five years,
The ability to better assess fit, in addition to ness was a significant predictor of
more embedded social links, will likely increase both in-role and extra-role job per- could be expected
the probability that employees will remain formance. Positive performance
to do so again.
with the organization. Thus, these two biodata not only helps avoid termination,
items are combined into one measure to reflect but also helps preserve the reputa-
pre-hire embeddedness. tion of the coworkers closest to the employee,
The other two biodata items, tenure in specifically friends and family employed at
last job (Barrick & Zimmerman, 2005) and the organization. Therefore, the more friends
number of jobs over the past five years (Price and family members an employee has work-
& Mueller, 1986), reflect commitment to ing at an organization, the more likely it is
prior employers. These two items were com- that the employee will perform well. In addi-
bined into one biodata measure to reflect tion, there is evidence that current employees
habitual commitment. Since past behavior is are likely to refer more capable applicants
the best predictor of future behavior (Owens (Breaugh & Starke, 2000; Rynes, 1991).
& Schoenfeldt, 1979; Wernimont & Camp- An applicant’s tenure in prior jobs and
bell, 1968), individuals who have a habit of the number of jobs recently held may be

influenced by several factors, including pre- they found that personal confidence scales
vious performance. An applicant who left (called disguised-purpose scales) added incre-
previous jobs because of poor performance mental validity to the prediction of voluntary
will likely exhibit a similar level of perfor- turnover beyond the biodata predictors, while
mance in the new position. Because poor employment motivation scales (termed clear-
performers tend to be terminated before purpose scales) did not. This study will exam-
good performers, poor performers will have ine whether this distinction also matters in
less tenure in their previous job. If poor per- this setting for turnover and extends to pre-
formers are not fired, they are more likely to dictions of job performance.
decide to leave their organizations (Griffeth,
Hom, & Gaertner, 2000) because of decreased
Personal Confidence Scales
job satisfaction (Judge, Thoresen, Bono, &
Patton, 2001). In addition, a meta-analysis Barrick and Zimmerman (2005) found that
by Meyer, Stanley, Herscovitch, and Topol- two measures of confidence, confidence with
nytsky (2002) showed a positive relationship self and confidence with decisions, predicted
between affective organizational commit- voluntary, avoidable turnover (r = −.17 and
ment and supervisor-rated job performance −.22, respectively), because employees with
( =.17), as well as between affec- higher confidence will be more persistent in
An applicant’s tive organizational commitment striving to adapt to novel job demands or the
and contextual performance work setting and will be less likely to with-
tenure in prior jobs ( =.32). Since past job-hopping draw from work because of anxiety over low
behaviors also indicate a histori- performance or ineffective adjustment (Lee,
and the number of cal lack of commitment to em- Ashford, Walsh, & Mowday, 1992). Higher
jobs recently held ployers, employees who have a confidence also should be related to higher
track record of job hopping are job performance, as employees with greater
may be influenced also more likely to have lower job confidence are likely to be more involved in
performance if they are uncom- their jobs, exert greater effort, and persist at a
by several factors. mitted to their current employer. task longer. Kanfer and Ackerman (2005)
There are two hypotheses for theorized that self-confidence affects job per-
these two biodata measures in reference to formance through self-regulatory behaviors
both turnover and performance: because individuals with higher self-confi-
dence are more likely to persist in perfor-
H1: Employees who have greater pre-hire em- mance-related goals, even in the face of
beddedness in the organization will be (a) more obstacles. Furthermore, self-confidence can
likely to remain with their current employers and affect employees’ motivation in such a way
(b) better performers than those with lower em- that employees low in self-confidence per-
beddedness. ceive the effort-performance relationship to
be flat even if objective information indicates
H2: Employees who had greater habitual com- the contrary (Kanfer & Ackerman, 2005). In
mitment to their former employers will be (a) addition, previous empirical research has de-
more likely to stay with their current employers termined that job performance is positively
and (b) better performers than those with less affected by both self-confidence (Linnehan,
commitment. 1998) and confidence with decisions (or deci-
siveness) (Kipnis & Glickman, 1962; Pynes &
Effects of Pre-hire Attitudes on Bernardin, 1992). Finally, both self-efficacy
Employees’ Turnover Decisions and generalized self-efficacy have been shown
to moderately correlate with job performance
and Job Performance
(r=.38 for self-efficacy, Stajkovic & Luthans,
Employee pre-hire attitudes were found to 1998; =.23 for generalized self-efficacy, Judge
differentially predict turnover in the Barrick & Bono, 2001). Although confidence is a
and Zimmerman (2005) study. Specifically, broader, more distal construct than

self-efficacy, both reflect the impact of one’s relationship between intent to quit and job
perceived competence in a setting. performance ( = −.14). For these reasons, the
employment motivation scales (desire for
H3: Employees with higher confidence will be the job and intent to stay) are likely to be
(a) more likely to stay with the organization related to turnover, even when assessed be-
and (b) better performers than those with lower fore hire, as well as to overall job perfor-
confidence. mance.
Employment Motivation Scales H4: Employees who are more motivated to ob-
tain the job will be (a) more likely to stay and
Barrick and Zimmerman (2005) showed that (b) better performers than those who are not as
applicants differ in their attraction to the job motivated.
or organization even before they are hired,
and these differences were predictive of vol- Effects of Personality Traits on
untary, avoidable turnover. Specifically, they Employees’ Turnover Decisions
found that two measures of the applicant’s
and Job Performance
motivation to obtain the job, the applicant’s
desire for the job and pre-hire intent to quit, Theoretically, both conscientiousness and
were related to turnover (R=.20). These results emotional stability should be negatively re-
reveal that one of the best predictors of vol- lated to voluntary turnover (Bar-
untary turnover for current employees, that rick & Mount, 1996). Maertz and Previous empirical
is, intent to quit (Griffeth et al., 2000), may colleagues (Maertz & Campion,
also predict turnover when assessed before 2004; Maertz & Griffeth, 2004) research has
the employee is hired. suggested that conscientiousness,
Employment motivation is also expected partially defined as being depend- determined that
to relate to job performance. Breaugh and able and reliable (Barrick & job performance
Mann (1984) posited that active job seekers Mount, 1991), is a factor in the
tended to be better performers than passive contractual and moral/ethical is positively
job seekers because they were more moti- motivational forces that affect
vated to obtain a job with that particular employees’ turnover decisions. affected by both
employer. Also, new employees who have a Specifically, Maertz and Griffeth self-confidence
strong desire to work for an organization stated that individuals who con-
likely require less time to be socialized into sider leaving their employers may and confidence
its culture (Van Maanen & Schein, 1979). reflect, “Do I owe any obligation
Employees who are more effectively social- to the organization that I would with decisions (or
ized into an organization should achieve break by leaving?” Therefore, in a
decisiveness).
higher performance (Wanous & Colella, contractual situation, such as ac-
1989). In addition, according to the theories cepting a job offer, conscientious
of social exchange (Van Dyne & Ang, 1998), employees are more likely to perceive that
norms of reciprocity (Gouldner, 1960), and obligations exist to their employers and are
perceived organizational support (Rhoades & more likely to adhere to these obligations by
Eisenberger, 2002), applicants who are more staying at the organization. These perceived
attracted to the job are more likely to repay obligations have been termed normative com-
the organization through greater effort (Lee mitment (Meyer & Allen, 1991) and have
et al., 2004). been found to negatively relate to intent to
Applicants who intend to quit or who quit (Shore, Tetrick, Shore, & Barksdale,
lack desire for their job will have lower com- 2000). For moral/ethical motivational forces
mitment to the job or organization, which discussed by Maertz and Griffeth (2004),
may lead to lower job performance (Meyer conscientious individuals are more likely to
et al., 2002). A meta-analysis by Zimmerman believe they have a moral obligation to stay
and Darnold (2009) indicated a negative with an organization. Maertz and Griffeth

posited that individuals with religious or conscientiousness and emotional stability

moral beliefs, such as the Protestant work were predictive of turnover of semi-truck
ethic, believe that perseverance is good re- drivers ( = −.26 for conscientiousness; = −.22
gardless of the circumstances (Blau & Ryan, for emotional stability). This evidence indi-
1997; Niles, 1999), and that switching jobs cates that both personality traits should be
indiscriminately is a sign of poor character. negatively related to voluntary turnover.
Conscientiousness also has well-established
relationships with job satisfaction ( =.26, H5: Employees who are more conscientious will
Judge, Heller, & Mount, 2002; Judge, Locke, be (a) more likely to stay with the organization
Durham, & Kluger, 1998; Organ & Lingl, and (b) better performers than those who are less
1995), an important predictor of employees’ conscientious.
turnover decisions.
There are also reasons to believe that H6: Employees who are more emotionally stable
emotional stability would be negatively will be (a) more likely to remain with the or-
linked to turnover. Individuals low in the ganization and (b) better performers than those
trait (or high in neuroticism) tend to have who are less stable.
negative perceptions of themselves and their
environment (Burke, Brief, & George, 1993; The full value of using FFM personality
Watson, Clark, & Tellegen, 1988). This leads traits during selection emerges when one
to an increased likelihood of expe- simultaneously considers the predictive va-
riencing negative states of mind lidity with both turnover and performance.
Theoretically, both or mood, which are associated Research (e.g., Barrick, Mount, & Judge, 2001)
with encoding and recalling nega- has shown that the two FFM traits examined
conscientiousness
tive information (Watson & Clark, here are valid predictors of performance in
and emotional 1984; Weiss & Cropanzano, 1996), all, or nearly all, jobs. Hence, these two traits
and higher levels of conflict with should be correlated with job performance,
stability should be coworkers (Organ, 1994). Because as well as with voluntary, avoidable turn-
neurotic individuals tend to be in over.
negatively related to
negative moods more often than
voluntary turnover. emotionally stable individuals are The Temporally Dynamic Nature of
and tend to have more conflicts
Turnover
with coworkers (Organ, 1994),
they are less likely to become effectively so- How do the effects of the variables examined
cialized into their organizations. Cote (2005) in this study endure over time? Theoretically
theorized that individuals exhibiting nega- the biodata measures and pre-hire attitude
tive emotions, such as sadness and anger, are scales are expected to primarily affect turn-
less likely to receive social support from co- over early in the job. When applicants begin
workers and more likely to experience inter- new jobs, socialization has been shown to
personal conflict, thereby increasing their have a disproportionately large effect on
stress levels and intentions to quit (Spector & turnover (Berlew & Hall, 1966). Socialization
Jex, 1998). promotes sense making, situational identifi-
In discussing the affective motivational cation, acculturation (Louis, 1980), and cre-
forces influencing voluntary turnover, Maertz ation of relationships and social integration
and Griffeth (2004) noted that employees (Louis, Posner, & Powell, 1983). Basing his
who have negative views of their work envi- work on field theory (Lewin, 1951), Allen
ronments are more likely to leave (Meyer (2006) noted that employees who fail to
& Allen, 1991). Judge et al. (2002) found that adapt to a new job environment may take
of all the five factor model (FFM) traits, the extreme response of leaving the organi-
emotional stability had the largest true score zation, particularly during the early stages of
correlation with job satisfaction, at .29, socialization. Learning how to do the job,
while Barrick and Mount (1996) noted that meeting the “right” people from whom to

learn about the organization, and figuring habitual commitment are expected to have
out the power structure of the firm and the a continuous effect on turnover and perfor-
organization’s goals and values are impor- mance. In keeping with this reasoning, we
tant to employee success and lead to lower propose:
turnover (Chao, O’Leary-Kelly, Wolf, Klein,
& Gardner, 1994; O’Reilly, Caldwell, & Bar- H7: The impact of more pre-hire embeddedness,
nett, 1989). attraction, and confidence on turnover and
Employees who were referred by other performance will be greater earlier in employees’
employees and who have more friends and job tenure than later in their tenure.
family in the organization are more likely
to have been provided with a realistic job Research Methodology and
preview and therefore are more familiar Measures Used in the Study
with the job requirements. These employ-
ees also have a pre-established social net- The sample consisted of job applicants at
work within the organization, which a large financial company in the Rocky
provides a means of social support early in Mountain region. The total sample was 354
their tenure. Similarly, the employment applicants, of whom 119 were hired as
motivation and personal confidence scales credit union tellers. The typical
relate to the internal resources employees participant was white (93%), fe-
Socialization
have to deal with these issues. Specifically, male (about 75%), in her early
greater confidence, desire for the job, and twenties (median age was 20 or promotes sense
intention to stay contribute to overcoming 21), with at least a high school
the new and uncertain situations inherent education. The respondents had making, situational
in the early stages of organizational entry. a stake in the outcome of the as-
Hence, employees who rate higher on these sessments and consequently identification,
constructs are better able to cope with the were motivated test takers who acculturation,
demands of a new job early in their tenure, thought their responses would
while employees who rate lower on these influence the hiring decision. and creation of
constructs are more susceptible to the psy- However, the results from these
chological upheaval that can occur during questionnaires were not used for relationships and
the early stages of employees’ socialization hiring purposes. Applicants who
social integration.
into an organization (Bauer, Morrison, & were not hired (N = 235) were
Callister, 1998). told so at that time and were ex-
According to attitude theory (Fishbein & cluded from consideration for the purposes
Ajzen, 1975), however, pre-hire attitudes of this study. Thus, all analyses were based
about a job are likely to change over time, on responses from hired applicants
based on on-the-job experiences. These ex- (N = 119).
periences may affect how employees feel As part of a formal job selection process,
about their positions, including their desire all applicants completed a pre-hire assess-
to stay, their confidence in having made the ment based on the scales used in this study.
right job choice, and their efficacy in com- All predictor measures, except the biodata
pleting their job duties. Furthermore, later measures, used a 5-point Likert rating for-
in their job tenure, employees without pre- mat (1=strongly disagree to 5=strongly
existing social networks will establish their agree).
own social circles with the organization.
Taken together, these predictors are ex-
Biodata Measures
pected to have stronger relationships with
turnover early in the job but weaker effects The first biodata measure, number of friends
over time, as the social and coping benefits and family, asked the applicant the number of
derived from these variables become less friends working at the organization and how
important. Conversely, personality and many family members work at the organiza-

tion (Breaugh & Dossett, 1989). The sum of committed to this company.” Intent to stay
both answers was used to reflect the number was assessed with five items ( =.80) from
of friends and family variable. Employee refer- Chatman (1991). These were based on items
rals, the second biodata measure (Breaugh & written for traditional intent-to-quit scales.
Mann, 1984), asked whether the applicant Examples were “I intend to remain with this
had been referred by an employee of the company for a long time,” and “If I have my
company. Those who were referred were own way, I will be working for this company
coded as a 1, and those who were not were six months from now.” Confirmatory factor
coded as a 0. Each biodata item was con- analyses indicated that a one-factor model fit
verted to a z-score, and these scores were the data better than a two-factor model. Fit
added together to reflect the biodata/pre-hire statistics for the one-factor model were: 2:
embeddedness measure. 151.78, 65 df; PNFI: .77; SRMR: .049; PGFI:
The number of months the applicant .67. Fit statistics for the two-factor model
had worked in his or her most recent job were: 2: 143.20, 64 df; PNFI: .76; SRMR: .047;
represented the third theoretically relevant PGFI: .66. The coefficient alpha for the com-
biodata item, time in prior job. The fourth bined scale was .70.
theoretically relevant item was number of
jobs held in the past five years.
Personal Confidence Scales
Again, these two items were con-
As part of a
verted into z-scores and then The measure reflecting applicant confi-
formal job combined to form the biodata/ dence was the sum of two scales: confidence
habitual commitment measure. with self and confidence with decisions.
selection process, Standardized scores for the num- Confidence with self was assessed with
ber of jobs in the last five years eight items from Lee et al. (1992) ( =.76).
all applicants variable were calculated to clas- Examples of confidence-with-self items in-
completed a pre- sify applicants into four groups: clude “I have always been able to do well in
those less than 18 years old, anything I have tried,” and “I expect to do
hire assessment those 18 to 19 years old, those well at this company.” Confidence with de-
20 to 21 years old, and those cisions was measured with a five-item scale
based on the scales over age 21. These scores were adapted from Lee et al. (1992) ( =.76). Ex-
compared only to those of other amples were “I never make major decisions
used in this study.
applicants in the same category quickly,” and “I always carefully weigh
to control for differences in the costs and benefits when making decisions
opportunity to switch jobs, depending on that affect my life.” Confirmatory factor
time in the workforce. The number of jobs in analyses indicated that a one-factor model
the last five years variable was reverse-coded fit the data either the same as or slightly
before being combined with the time in better than a two-factor model. Fit statistics
prior job variable, so high scores on both for the one-factor model were: 2: 281.12,
were equated with greater commitment to 65 df; PNFI: .69; SRMR: .074; PGFI: .64. Fit
prior employers. statistics for the two-factor model were: 2:
239.97, 64 df; PNFI: .69; SRMR: .074; PGFI:
.64. The coefficient alpha for the combined
Employment Motivation Scales
scale was .73.
The measure of applicant attraction to the
employment opportunity focused on the ap-
Personality Traits
plicants’ pre-hire desire for the job at the firm
and pre-hire intent to stay. Desire for a job was The personality assessment consisted of 60
assessed with eight items from the Lee et al. items designed to comprehensively measure
(1992) job desirability scale ( =.76). Examples conscientiousness and emotional stability,
were “I have a strong desire to be an em- with 30 items measuring each personality
ployee of this company” and “I feel very trait. Coefficient alpha reliabilities were .87

and .86, respectively. Examples of items for but only for comparative purposes with pre-
conscientiousness were, “I put a great deal of vious findings.
effort into my work,” and, “Others have de-
scribed me as a very disciplined person.” Ex- Job Performance Ratings
amples of emotional stability items were, “I
become irritated when others criticize me,” The employees’ supervisors rated their job
and “I tend to get over embarrassing situations performance after 30 days, 6 months, and 1
very quickly.” year. Employees were evaluated on nine di-
mensions: quality of work, quantity of work,
job knowledge, interpersonal skills, rule-fol-
Turnover
lowing behavior, communication skills, ini-
Turnover data were collected over a two-year tiative, punctuality, and customer service.
period after the applicants were hired and Performance was evaluated on a five-point
were categorized as having occurred within Likert scale ranging from “unsatisfactory” to
six months or after this period, up to two “far exceeds expectations” ( =.75–.84). Over-
years. Within the first six months, there were all performance was the mean of the ratings
95 stayers (coded as 0) and 24 leavers (coded across all dimensions. Correlations between
as 1). Of the employees who left during this other variables and job perfor-
period, 18 left for voluntary, avoidable rea- mance for time 1 (up to six
sons, and 6 left for voluntary, unavoidable months) used the 30-day perfor- The personality
reasons. Of those leaving between six months mance ratings, while correlations
assessment
and two years after hire, there were 70 stayers with job performance for time 2
and 25 leavers. Of the employees who left (up to two years after hire) used consisted of 60
during this period, one left because of the most recent performance rat-
involuntary reasons (was fired); 18 left for ing (either performance rated at items designed to
voluntary, avoidable reasons; and 6 left for six months or at one year) for em-
comprehensively
voluntary, unavoidable reasons. Reasons for ployees who remained after six
turnover decisions were coded according to months. measure
the classification scheme developed by Abel-
son (1987). The voluntary, avoidable turn- Study Results: Pre-hire conscientiousness
over category refers to turnover that reflected Predictors Are Related to and emotional
the individual’s decision to leave and that the Both Turnover and Job
organization may have been able to avoid stability.
Performance
(e.g., through raises or by providing better
work conditions). Voluntary, unavoidable Table I reports the means, stan-
turnover occurs when the employee chooses dard deviations, and zero-order correlations
to leave but the organization had no control among the variables. For variables 1 to 9, cor-
over it (e.g., quitting to trail a relocating relations greater than .15 are significant at the
spouse or to resume education). Abelson .05 level using a one-tailed test; for variables
(1987) found that those who leave the orga- 10 to 12, correlations greater than .17 are
nization for unavoidable reasons resemble significant. All the predictors were negatively
stayers more than they resemble the leavers related to voluntary, avoidable turnover (r
whose departure is avoidable. In fact, Hom ranges from −.18 to −.27) during the six-
and Griffeth (1995) stated that voluntary, month period, as hypothesized. These results
avoidable turnover is “a superior criterion for provide initial support for the hypotheses
testing prevailing turnover models”. For these that pre-hire embeddedness (r=−.22, hypoth-
reasons, this study examines the predictive esis 1a), habitual commitment (r=−.21, hy-
validity of the various selection variables (e. pothesis 2a), personal confidence (r=−.20,
g., biodata and personality) by focusing on hypothesis 3a), motivation for employment
voluntary, avoidable turnover. Correlations (r=−.18, hypothesis 4a), conscientiousness
with voluntary turnover also are reported, (r=−.19, hypothesis 5a), and emotional stabil-

192
TABLE I Correlations Among Variables

Variables M SD 1 1a 1b 2 2a 2b 3 4 5 6 9 12
1. Biodata Prehire Embeddedness .00 .74 —
Composite
1a. Employee Referral .41 .49 .65 —
1b. Number of Friends and 1.21 1.4 .57 .10 —
Family
2. Biodata Habitual Commitment .01 .71 .31 .28 .14 —
Composite
2a. Time on Prior Job (in 22.6 26.3 .28 .25 .17 .87 —
months)
2b. Number of Jobs in Last 2.70 1.4 −.07 −.09 −.01 −.73 −.54 —
Five Years
HUMAN RESOURCE MANAGEMENT, MARCH–APRIL 2009
3. Personal Confidence 3.68 .43 .09 .00 .11 .12 .09 −.08 —
4. Employment Motivation 4.15 .46 .09 −.04 .16 .06 .09 .04 .24 —
5. Conscientiousness 4.30 .34 .17 .09 .16 .09 .06 −.07 .40 .43 —
6. Emotional Stability 3.90 .36 .28 .17 .24 .07 .04 −.06 .45 .31 .73 —
Up to six months after hire
7. Voluntary, Avoidable Turnover 0.16 .37 −.22 −.17 −.15 −.21 −.05 .19 −.20 −.18 −.19 −.27 −.25
(N=113)
8. Voluntary Turnover (N=119) 0.20 .40 −.20 −.17 −.14 −.22 −.08 .19 −.16 −.13 −.18 −.27 −.29
9. Job Performance (N=119) 3.33 .61 .29 .26 .17 .22 .19 −.06 .03 −.05 .18 .19 — .37
Up to two years after hire
10. Voluntary, Avoidable Turnover 0.20 .41 .00 .17 −.15 −.06 −.03 .05 −.05 .04 −.22 −.21 −.14 −.27
(N=88)
11. Voluntary Turnover (N=95) 0.26 .44 .02 .16 −.12 −.01 .10 .12 −.05 .04 −.21 −.20 −.10 −.29
12. Job Performance (N=95) 3.50 .37 .11 .17 −.01 .07 .12 .06 .14 −.08 .18 .18 .37 —
Note: For variables 1–9, N=113–119, p <.05 for correlations >.15; for variables 10–12, N=88–95, p <.05 for correlations >.17.

ity (r=−.27, hypothesis 6a) predict who is pre-hire embeddedness and habitual com-
likely to remain working for a company six mitment are both related to supervisory
months after hire. Table I also reports the ratings of job performance at six months
correlations between these predictors and a with correlations of .29 and .22, respec-
broader turnover category, one that reflects all tively. These results provide initial support
voluntary turnover. This outcome reflects all for hypotheses 1b and 2b. In addition, hy-
turnover decisions when the employee volun- potheses 5b and 6b were initially supported,
tarily chooses to leave, regardless of whether given that conscientiousness and emotional
the company could reasonably do anything stability predicted job performance at six
about it (avoidable) or not (unavoidable). As months (r=.18 and .19, respectively). How-
expected, these correlations are comparable ever, personal confidence and motivation
to those reported for voluntary, avoidable for employment were not related to job
turnover (r ranges from −.13 to −.27). performance at six months, which fails to
After the six-month period and up to support hypotheses 3b and 4b. After the
two years later, conscientiousness and emo- six-month period, only the two personality
tional stability were still related to volun- traits (r=.18 for both) were significantly re-
tary, avoidable turnover, as expected (r=−.22 lated to performance. The other predictors,
and −.21, respectively). All other predictors including the two biodata mea-
(pre-hire embeddedness, habitual commit- sures and pre-hire attitude scales,
This study
ment, employment motivation, and per- were not significant predictors (r
sonal confidence scales) had correlations ranges from −.08 to .14). There- examines the
with voluntary, avoidable turnover that fore, hypothesis 7 was again only
were nearly equal to zero (r ranges from partially supported. predictive validity
−.06 to .04). Overall, there was initial par- Although the primary analy-
of the various
tial support for hypothesis 7. Specifically, ses involving the biodata items
the validity of pre-hire embeddedness, em- were for the broader construct- selection variables
ployment motivation, and personal confi- level composites, the results for
dence scales decreased over time for these the individual items comprising (e.g., biodata
predictors when predicting voluntary, the composites were also included
and personality)
avoidable turnover. However, the validity in the correlation matrix for infor-
of the habitual commitment measure, mational purposes. As can be seen by focusing on
which was expected to be a valid predictor in Table I, both items making up
during the later time period, also decreased the pre-hire embeddedness com- voluntary, avoidable
over time. As expected, personality was posite had fairly similar relation-
found to be a useful predictor of voluntary, ships with turnover and perfor- turnover.
avoidable turnover up to two years after mance up to six months after
hire. hire. However, for the habitual commitment
Table I also provides information as to composite, the number of jobs held over the
whether the turnover that occurs is func- previous five years was a better predictor of
tional (poor performers leave) or dysfunc- early turnover, while tenure on the most re-
tional (good performers leave). In accor- cent job was more predictive of early job
dance with the Griffeth et al. (2000) performance.
meta-analysis, voluntary, avoidable turn- These findings suggest that although all
over is negatively correlated with perfor- the predictors studied here are able to fore-
mance (r=−.25 at up to six months and cast who leaves within six months, only the
r=−.27 up to two years). This supports the biodata measures and personality traits pre-
contention that those who stay tend to be dicted employees’ job performance early in
better performers. The results in Table I also their jobs. Furthermore, only the personality
demonstrate that a number of the variables traits were correlated with performance after
that predict turnover also predicted job per- six months. Finally, the pre-hire attitude
formance at up to six months. Specifically, scales did not predict who is likely to be

successful in the job either early or up to two for performance. For each set of predictors
years later. (e.g., biodata), regressions were run three
Although the post-six-month analyses ways by entering the set of predictors in each
considered only the employees who were still step. In this way, it was possible to assess the
employed at the credit union at the time, the incremental gain provided by each set of pre-
decrease in the magnitude of the correlations dictors, after accounting for the effects of
was not due to restriction in the variability of other sets of predictors. For step 2, the results
the predictor variables. The degree of range for both possible orders of entry are presented
restriction between the time 1 and time 2 as steps 2a and 2b, with the subhead indicat-
samples was extremely small, with range re- ing which construct was entered first. Specifi-
striction values (SD for time 2/SD for time 1) cally, biodata were entered after either
of .96 to 1.01 across all the predictors, personality (2a) or pre-hire attitudes (2b),
although the means were higher at time 2. personality after biodata (2a) or pre-hire atti-
Although the organization did not use the tudes (2b), and pre-hire attitudes after per-
results of the assessments to make hiring de- sonality (2a) or biodata (2b).
cisions, it was still important to evaluate pos- For each regression, we assessed the over-
sible range restriction in the hired sample all variance accounted for (R2) and the rela-
versus the applicant sample. Similar to the tive change in prediction (R2) obtained by
results between the time 1 and adding the set of variables in each step of that
time 2 samples, the degree of regression. Thus, the relative change (R2)
The magnitude of
range restriction was very small demonstrates the incremental gain of that
the predictions (range restriction values of .90 to predictor set, once other sets (steps) of vari-
1.06) for most of the variables. ables are included in the regression. For the
obtained from the However, employment motiva- logistic regression results presented in Tables
tion and emotional stability had II and III, the R2 values are Cox and Snell
biodata and pre-hire higher levels of range restriction: analogs to ordinary-least-squares R2 estimates.
attitudinal predictors values of .76 and .82, respectively. In addition, the odds ratios are also pre-
This finding indicates that range sented. The odds ratio indicates the ratio of
of voluntary, restriction may be responsible for relative importance of the independent vari-
the somewhat lower correlations ables in terms of their effect on the depen-
avoidable turnover that employment motivation has dent variable’s odds of occurring.
with turnover and performance The overall regression equations for turn-
attenuate over time.
compared to the other predictors. over and performance, assessed at six months,
Similarly, the effects of personal- were of comparable magnitude (R2=.134 to
ity, especially the trait of emotional stability, .143), indicating that these three sets of vari-
on turnover and job performance are even ables (biodata, pre-hire attitudes, and per-
greater when accounting for range restric- sonality traits) are useful predictors of both
tion. outcomes at up to six months after hire. As
Another important purpose of this study shown in the logistic regression results pre-
was to establish the incremental validity of sented in Table II, each set of predictors sig-
these theoretically relevant predictors of vol- nificantly predicted voluntary, avoidable
untary, avoidable turnover and performance. turnover during the six-month period when
Hierarchical regression analyses tested the entered alone (step 1). The only significant
ability of each set of predictors (biodata, pre- gains in incremental validity occur when
hire attitudes, and personality traits) to explain personality is entered either second after bio-
separate portions of the variance in either data in the regression equation (R2=.050 for
turnover or performance. Each set of variables step 2a) or when biodata are added (R2=.059
was standardized before being entered into for step 2a or 3, R2=.066 for step 2b). The
the regressions. Tables II and III summarize odds ratios confirm these results. Specifically,
the results for predictions of voluntary, including biodata in the logistic regressions
avoidable turnover and Tables IV and V at steps 1, 2a, 2b, or 3 indicates that employ-

TABLE II Hierarchical Logistic Regression Results Between Voluntary, Avoidable Turnover at Up to Six Months After Hire and Various Sets of Predictors
Up to Six Months After Hire

Step 1 Step 2a Step 2b Step 3
2 2 2 2 2 2 2
Variable sets R R R R R R R R 2

Entered After Entered After
Personality Attitudes
Biodata .073* — .123* .059* .120* .066* .134* .059*
2
Full model , df 8.556, 1 14.815, 2 14.451, 2 16.296, 3
Step 2, df — 7.392, 1 8.127, 1 7.534, 1
Odds ratio .242* .245* .231* .236*
Biodata Attitudes
Personality .064* — .123* .050* .075* .021 .134* .014
Full model 2, df 7.423, 1 14.815, 2 8.762, 2 16.296, 3
Step 2, df — 6.259, 1 2.438, 1 1.845, 1
Odds ratio .443* .446* .566 .589
Personality Biodata
Pre-hire Attitudes .054* — .075* .011 .120* .047* .134* .011
2
Full model , df 6.324, 1 8.762, 2 14.451, 2 16.296, 3
Step 2, df — 1.339, 1 5.895, 1 1.481, 1
Odds ratio .366* .572 .356* .531
2 2
Note: R values are Cox and Snell R .
*p < .05
Hiring for Retention and Performance
195
196
TABLE III Hierarchical Logistic Regression Results Between Voluntary, Avoidable Turnover at Up to Two Years After Hire and Various Sets of Predictors
Up to Two Years After Hire
2 2 2 2 2 2 2
Biodata .001 — .053* .000 .001 .001 .071* .001
2
Full model , df .103, 1 4.777, 2 .103, 2 6.375, 3
Step 2, df — .020, 1 .102, 1 .026, 1
Odds ratio .859 .928 .859 .918

Biodata Attitudes
Personality .053* — .053* .052* .070* .070* .071* .070*
Full model 2, df 4.757, 1 4.777, 2 6.348, 2 6.375, 3
Step 2, df — 4.675, 1 6.347, 1 6.271, 1
Odds ratio .493* .493* .378* .379*
Personality Biodata
Pre-hire Attitudes .000 — .070* .017 .001 .000 .071* .018
2
Full model , df .001, 1 6.348, 2 .103, 2 6.375, 3
Step 2, df — 1.590, 1 .000, 1 1.597, 1
Odds ratio .991 1.786 .997 1.789
2 2
Note: R values are Cox and Snell R .
*p < .05

TABLE IV Hierarchical Regression Results Between Performance at Up to Six Months After Hire and Various Sets of Predictors
Up to Six Months After Hire
Variable Sets R2 R 2 R2 R 2 R2 R 2 R2 R 2

Biodata .099* — .118* .079* .103* .103* .143* .082*
Model F-value, df 12.838; 1, 117 7.749; 2,116 6.654; 2, 116 6.411; 3, 115
Step F-value, df — 10.339, 1 13.262, 1 11.015, 1
Biodata Attitudes
Personality .039* — .118* .019 .062* .062* .143* .040*
Model F-value, df 4.778; 1, 117 7.749; 2, 116 3.783; 2, 116 6.411; 3, 115
Step F-value, df — 2.496, 1 7.522, 1 5.419, 1
Personality Biodata
Pre-hire attitudes .000 — .062* .022 .103* .004 .143* .025
Model F-value, df 0.839; 1, 117 3.783; 2, 116 6.654; 2, 116 6.411; 3, 115
Step F-value, df — 2.717, 1 .522, 1 3.413, 1
Note: *p <.05
Hiring for Retention and Performance
197
198
TABLE V Hierarchical Regression Results Between Performance at Up to Two Years After Hire and Various Sets of Predictors
Up to Two Years After Hire
2 2 2 2 2 2 2
Biodata .011 — .045 .007 .012 .011 .051 .005
Model F-value, df .949; 1, 86 1.917; 2, 85 .499; 2, 85 1.448; 3, 84

Step F-value, df — .532, 1 .9331, 1 .460, 1
Biodata Attitudes
Personality .038* — .045 .033* .046 .045* .051 .039*
Model F-value, df 3.316; 1, 86 1.917; 2, 85 1.955; 2, 85 1.448; 3, 84
Step F-value, df — 2.864, 1 3.841, 1 3.318, 1
Personality Biodata
Pre-hire Attitudes .001 — .046 .008 .012 .001 .051 .006
Model F-value, df .067; 1, 86 1.955; 2, 85 .499; 2, 85 1.448; 3, 84
Step F-value, df — .610, 1 .060, 1 .532, 1
Note: *p <.05

ees who were one standard deviation above level of education as a proxy for ability. The
the mean had less than one-quarter the odds pattern of results for the regressions between
of turnover as employees at the mean. In- job performance at time 1 and the three sets
cluding personality in the logistic regressions of predictors still held even when controlling
at step 1 or 2a indicates that employees one for the employee’s level of education. The
standard deviation higher on personality variance in performance attributable to the
had less than half the odds of turnover as biodata scales decreased slightly (step 1
employees at the mean. As these findings il- R2 =.084; step 2a R2 =.061; step 2b R2 =.087;
lustrate, once personality and biodata are step 3 R2 =.064), but the results at each step
included in the regression of voluntary, were still significant. For personality and pre-
avoidable turnover, most of the available hire attitudes, the variance explained actually
variance is accounted for (R2=.123). increased by extremely small amounts (maxi-
Furthermore, the magnitude of the pre- mum increase in the R2 or change in R2 of
dictions obtained from the biodata and pre- .008). Education was not significantly corre-
hire attitudinal predictors of voluntary, lated with performance at either time 1 or
avoidable turnover attenuates over time. In time 2 (r=.15 and .11, respectively).
fact, as shown in Table III, only personality As shown in Table V, only per-
significantly added incremental validity to sonality significantly added incre-
The finding that
the prediction of voluntary, avoidable turn- mental validity to the prediction
over up to two years after hire, regardless of of job performance up to two pre-hire attraction
when personality was entered in the regres- years after hire, and this finding is
sion equation (R2=.053, .052, .070, or .070 true regardless of when personal- and confidence
for steps 1, 2a, 2b, or 3, respectively). For ity was entered into the regression
scales offer little
steps 1 and 2a, employees one standard de- (R2=.038, .033, .045, and .039 for
viation above the mean on personality have Steps 1, 2a, 2b, or 3, respectively). incremental
less than half the odds of turnover as em- Thus, by accounting for biodata
ployees at the mean. For steps 2b and 3, the and personality, one is able to ac- validity over the
odds of turnover are about 38% of those for count for nearly all the incremen-
employees at the mean. tal gain when predicting either broader traits of
In Table IV, two sets of predictors, bio- voluntary, avoidable turnover or conscientiousness
data and personality traits, were shown to performance after six months. In
have added significant incremental gains in contrast, only personality added and emotional
validity when predicting early job perfor- significant incremental validity
mance. The results reveal that biodata for either turnover or performance stability is important.
significantly improved the prediction, re- two years after hire.
gardless of when they were added (R2 =.079,
.103, or .082 in steps 2a, 2b, and 3, respec- Discussion of Implications of Find-
tively). The two personality traits also added ings, Study Limitations, and Future
significant incremental predictive validity
Research Opportunities
after pre-hire attitudes were entered (R2=.062
in step 2b) or after all other predictors (pre- The purpose of this study was to examine the
hire attitudes and biodata) were accounted extent to which employers can reduce turn-
for (R2=.040 in step 3). Consequently, once over and simultaneously increase perfor-
personality and biodata were included in the mance during the selection process by using
regression of early job performance, most of predictors related to applicants’ propensity
the available variance was accounted for to become attached to their organizations.
(R2=.118) Turnover researchers have historically ne-
In order to ensure that the regression re- glected this question with actual job appli-
sults were not affected due to a common an- cants (Griffeth et al., 2000). To fully evaluate
tecedent, such as differences in ability level, the utility of assessing pre-hire retention-
supplemental analyses were performed using related individual differences, however, it is

important to examine whether these predic- ness and emotional stability is important. For
tors of turnover are also simultaneously re- current employees, job-specific attitudes, par-
lated to job performance. This study brings ticularly intent to quit, are among the best
together biodata, pre-hire attitudes, and per- predictors of voluntary turnover. These re-
sonality traits for the first time to determine sults reveal that these measures are not as
their ability to jointly predict turnover and predictive when applicants provide the re-
performance. sponses. Furthermore, these pre-hire mea-
Consistent with earlier research (Barrick & sures are not related to later job performance.
Zimmerman, 2005), the results indicate that These findings indicate that general appli-
biodata measures that assess pre-hire embed- cant personality traits are more useful predic-
dedness in the organization and habitual tors of turnover and performance than
commitment and pre-hire attitude scales that job- or organization-specific predictors of
measure employment motivation, personal turnover. This may suggest that regardless of
confidence, and the traits of con- applicants’ pre-hire perceptions of how well
scientiousness and emotional the specific job meets their needs, it is more
Besides losing
stability predicted voluntary, important to select applicants with a positive
a potentially avoidable turnover during a six- propensity toward working in general. The
month period after hire. However, findings also indicate another potential ben-
productive these results also show that the efit of using personality-based integrity tests:
biodata and pre-hire attitude reduction of turnover. Possibly because of the
employee, scales’ validities attenuate rapidly, indirect measure of the personality constructs
turnover during the as only the personality traits were of conscientiousness and emotional stability
relevant predictors of turnover up (Ones, 1993), integrity tests predict not only
transition phase to two years later. The results also thievery and other deviant behaviors, but
indicate that biodata and person- also job performance (Ones, Viswesvaran, &
is likely to be ality predicted job performance, Schmidt, 1993) and, as implied by the results
although personality was the only of this study, turnover.
costly because the
valid predictor of performance Despite the presence of individual dif-
organization has after six months. Furthermore, the ferences in most models of turnover, they
study showed that the pre-hire at- are the least understood. The use of these
yet to recoup the titude scales (employment moti- distinct sets of individual differences be-
vation and personal confidence) gins to fill in the empirical gaps in theo-
direct and indirect
were not related to job perfor- retical models of turnover, particularly in
costs associated mance. Thus, this study goes be- relation to pre-hire antecedents of turn-
yond earlier findings to show that over. One of the most frequently examined
with hiring and the pre-hire attitudinal predictors pre-hire antecedents of turnover has been
did not add significant incremen- realistic job previews. However, as Phillips
training the
tal validity in predicting either (1998) illustrates, these effects are modest
employee. turnover or performance after at best (mean r=−.09 in field settings,
accounting for biodata and per- R2 =.03). The magnitude of the effects re-
sonality. These findings have im- ported here (R2 =.14 at up to six months)
plications for companies experiencing high far exceeds those obtained using realistic
turnover, as well as for models of the turnover job previews. Consequently, the under-
process. Taken together, these results illustrate standing that differences in employees’
that during selection, general personality likelihood of becoming attached to the or-
traits and life history experiences are better ganization influence their turnover deci-
predictors of voluntary, avoidable turnover sions provides future researchers with a
and job performance than pre-hire attitudes. foundation to examine how individual dif-
The finding that pre-hire attraction and ferences work through job satisfaction and
confidence scales offer little incremental va- organizational commitment to predict
lidity over the broader traits of conscientious- turnover. In fact, researchers (Hom & Grif-

feth, 1995) have been critical of structural with hiring and training the employee. Or-
equation model tests and accompanying ganizations would benefit from implement-
parameter estimates that do not include all ing programs (e.g., formal socialization,
the necessary variables. Our findings sug- mentoring, or training) that minimize
gest that omitting individual differences, turnover by reducing an employee’s uncer-
such as those examined in this study (par- tainty and ambiguity during the transition
ticularly the personality and biodata vari- phase.
ables), from the model may affect the Surprisingly, whether the individual was
accuracy of the parameter estimates of referred by an employee (one of the items
other variables studied. in the pre-hire embeddedness scale) changed
Another contribution of this study is the from being negatively related to turnover at
evidence that the validities of several pre- time 1 to being positively related to turn-
dictors of turnover decline over time. This over at time 2. The reason is unclear, but it
finding is consistent with the Griffeth et al. may be that people who are referred by em-
(2000) meta-analysis, which found that the ployees (particularly friends or family mem-
relationships of both performance and com- bers) often feel obligated to accept the job,
mitment with turnover were weaker when even if they are not sure that it is the best
turnover was measured more than 12 position for them. Although they may ini-
months after the measurement of the pre- tially stay because of such feelings of obli-
dictor. Researchers who develop turnover gation, they may later realize that they
models need to be explicit regarding whether made a wrong decision and leave for a job
their model explains turnover decisions with another organization. This explana-
shortly after hire or after an extended pe- tion is pure conjecture, however. Future
riod, as the antecedents of turnover during research should focus specifically on repli-
these two time frames may differ (Mitchell cating this finding and examining the un-
& James, 2001). Although there have been derlying causes.
repeated calls for the investigation of the One limitation of the study is the
effects of time in organizational research possibility that candidates engaged in im-
(Wright, 1997, 2002), such studies are still pression management. This would possibly
few and far between. attenuate the correlation between the mea-
In this study, the decline in predictive sured constructs and turnover (Barrick &
validity may have resulted from changes in Mount, 1996; Bernardin, 1987). In particu-
employees’ ability to deal with job require- lar, employment motivation items, such as
ments early in their tenure as opposed to “If I have my own way, I will be working
after having been on the job for an ex- for this company six months from now,”
tended period. This change in predictive could be influenced by applicant faking.
validity may be related to Murphy’s (1989) Nevertheless, even these predictors were
conceptualized division of job tenure into found to be significantly related to turn-
a “transition” phase, when job demands over in an actual applicant setting. Future
are uncertain and stressful, and a “mainte- research should examine the effects of im-
nance” phase, when the demands are no pression management on these measures.
longer novel after the employee learns how In addition, as a precondition for collect-
to effectively perform the job. While turn- ing enough data to provide reliable effect
over in either phase may be detrimental to size estimates, researchers typically study
organizational performance, turnover early jobs that have sufficient quit rates to afford
in an employee’s tenure may be particu- the large sample sizes needed to examine
larly harmful. Besides losing a potentially turnover. Although a strength of this study
productive employee, turnover during the was its departure from the use of health care
transition phase is likely to be costly be- professionals and military personnel as sub-
cause the organization has yet to recoup jects, the reliance on a lower-level job (credit
the direct and indirect costs associated union teller) may mean that the findings do

not necessarily apply to all types of posi- Although considerable research has been
tions. Similar to the finding that level of job conducted on the incremental validity of
complexity moderates the relationship be- predictors of performance (Schmidt &
tween general cognitive ability and perfor- Hunter, 1998), we know of no such research
mance (Schmidt & Hunter, 1998), job level for voluntary, avoidable turnover.
may also moderate the relationships between This study has identified a number of
the predictors in this study and turnover. individual differences related to organiza-
Future research should replicate these find- tional attachment that can be used to pre-
ings in upper-level jobs, where the costs as- dict both turnover and performance before
sociated with turnover are typically much the applicant is hired, with biodata and per-
higher (Hom & Griffeth, 1995). sonality being among the most effective.
Another limitation of the study is the However, pre-hire attitudes (employment
fairly homogeneous sample used, particu- motivation and personal confidence) did
larly in terms of minorities and individuals not predict turnover and performance be-
over age 40 (which is also likely a function yond biodata (pre-hire embeddedness in the
of the type of job in the sample), both of organization and habitual commitment)
whom made up 7% or less of the sample. and the personality traits (conscientious-
Because of this, meaningful adverse impact ness and emotional stability). Therefore,
analyses would be impossible. However, we organizations would benefit most from in-
refer readers to findings by Barrick and cluding the biodata and personality trait
Zimmerman (2005) in reference to the predictors in their hiring process. This study
minimal adverse impact caused by the pre- also illustrates that except for personality,
dictors included in both that study and this the importance of these predictors attenu-
one, as well as Hough (1998) and Hough, ates over time. The finding that the con-
Oswald, and Ployhart (2001) regarding the structs that best predicted early job turnover
low d-scores of personality traits. are related to employees having social and
Further research also should investigate psychological support provides indirect en-
the incremental validity of other pre-hire dorsement of developmental programs de-
methods of predicting turnover and perfor- signed to reduce employees’ uncertainty
mance, such as person-environment fit and ambiguity shortly after they begin their
(Kristof-Brown, Zimmerman, & Johnson, jobs. Future research should replicate these
2005) and extraorganizational loyalties (e.g., findings in higher-level jobs and include
moonlighting), to gain a more comprehen- other predictors related to job performance
sive picture of the value of each predictor. and voluntary turnover.
MURRAY R. BARRICK is the department head and Robertson Chair in Business at the
Mays Business School, Texas A&M University. He earned his PhD in industrial/organiza-
tional psychology from the University of Akron. His research interests include assessing
the impact individual differences in behavior and personality have on job and team per-
formance and on methods of measuring and predicting such differences. His work has
been cited more than 5,300 times (Google Scholar, Nov. 2008). Along with Mick Mount,
Barrick is the 2009 recipient of the Distinguished Scientific Contributions Award from the
Society for Industrial and Organizational Psychology.
RYAN D. ZIMMERMAN is an assistant professor of human resource management in the

Management Department at Texas A&M University. He earned his PhD in HRM from the
University of Iowa. His research interests include personnel selection, employee turnover,

and individual differences. He is a recipient, along with two coauthors, of the Academy
of Management Human Resources Division’s 2006 Scholarly Achievement Award for
the most significant paper in human resource management. Before earning his PhD,
Zimmerman worked as a manager, HR specialist, and internal HR consultant.
References value of theory-based biodata for predicting

turnover. Journal of Business and Psychology,
Abelson, M. A. (1987). Examination of avoidable and 3(4), 371–385.
unavoidable turnover. Journal of Applied Psychol-
Breaugh, J. A., & Mann, R. B. (1984). Recruiting
ogy, 72(3), 382–386.
source effects: A test of two alternative hypoth-
Allen, D. G. (2006). Do organizational socialization eses. Journal of Occupational Psychology,
tactics influence newcomer embeddedness 57(4), 261–267.
and turnover? Journal of Management, 32(2),
Breaugh, J. A., & Starke, M. (2000). Research on
237–256.
employee recruitment: So many studies, so many
Barrick, M. R., & Mount, M. K. (1991). The big remaining questions. Journal of Management,
five personality dimensions and job performance: 26(3), 405–434.
A meta-analysis. Personnel Psychology, 44(1),
Burke, M. J., Brief, A. P., & George, J. M. (1993).
1–26.
The role of negative affectivity in understanding
Barrick, M. R., & Mount, M. K. (1996). Effects of relations between self-reports of stressors and
impression management and self-deception on strains: A comment on the applied psychology
the validity of personality constructs. Journal of literature. Journal of Applied Psychology, 78(3),
Applied Psychology, 81(3), 261–272. 402–412.
Barrick, M. R., Mount, M. K., & Judge, T. A. (2001). Chao, G. T., O’Leary-Kelly, A. M., Wolf, S., Klein,
Personality and performance at the beginning of H. J., & Gardner, P. D. (1994). Organizational
the new millennium: What do we know and where socialization: Its content and consequences.
do we go next? International Journal of Selection Journal of Applied Psychology, 79(5), 730–743.
and Assessment, 9, 9–29.
Chatman, J. J. (1991). Matching people and organiza-
Barrick, M. R., & Zimmerman, R. D. (2005). Reducing tions: Selection and socialization in public account-
voluntary, avoidable turnover through selection. ing firms. Administrative Science Quarterly, 36(3),
Journal of Applied Psychology, 90(1), 159–166. 459–484.
Bauer, T. N., Morrison, E. W., & Callister, R. R. Cote, S. (2005). A social interaction model of the
(1998). Organizational socialization: A review effects of emotion regulation on work strain.
and directions for future research. In G. R. Ferris Academy of Management Review, 30(3), 509–530.
(Ed.), Research in personnel and human resources
Dean, M. A., Russell, C. J., & Muchinsky, P. M. (1999).
management (Vol. 16, pp. 149–214). Stamford,
Life experiences and performance prediction:
CT: JAI Press.
Toward a theory of biodata. In G. Ferris (Ed.),
Berlew, D. E., & Hall, D. T. (1966). The socialization Research in personnel and human resources
of managers: Effects of expectations on perform- management (pp. 245–281). Greenwich, CT: JAI
ance. Administrative Science Quarterly, 11(2), Press.
207–223.
Fishbein, M., & Ajzen, I. (1975). Belief, attitude,
Bernardin, H. J. (1987). Development and validation intention, and behavior: An introduction to theory
of a forced choice scale to measure job-related and research. Reading, MA: Addison-Wesley.
discomfort among customer service representa-
tives. Academy of Management Journal, 30(1), Ghiselli, E. E. (1974). Some perspectives for industrial
162–173. psychology. American Psychologist, 80(2), 80–87.
Blau, G., & Ryan, J. (1997). On measuring work Gouldner, A. W. (1960). The norm of reciprocity:
ethic: A neglected work commitment facet. Journal A preliminary statement. American Sociological
of Vocational Behavior, 51(3), 435–448. Review, 25(2), 161–178.
Breaugh, J. A., & Dossett, D. L. (1989). Rethinking Griffeth, R. W., Hom, P. W., & Gaertner, S. (2000). A
the use of personal history information: The meta-analysis of antecedents and correlates of

employee turnover: Update, moderator tests, and Kristof-Brown, A. L., Zimmerman, R. D., & Johnson,
research implications for the next millennium. E. C. (2005). Consequences of individuals’ fit
Journal of Management, 26(3), 463–488. at work: A meta-analysis of person-job, person-
organization, person-group, and person-
Holtom, B. C., Mitchell, T. R., Lee, T. W., & Inderrieden,
supervisor fit. Personnel Psychology, 58(2),
E. J. (2005). Shocks as causes of turnover: What
281–342.
they are and how organizations can manage them.
Human Resource Management, 44(3), 337–352. Lee, T. W., Ashford, S. J., Walsh, J. P., & Mowday, R.
T. (1992). Commitment propensity, organizational
Hom, P. W., & Griffeth, R. W. (1995). Employee turn-
commitment, and voluntary turnover: A longi-
over. Cincinnati: South-Western.
tudinal study of organizational entry processes.
Hough, L. M. (1998). Personality at work: Issues Journal of Management, 10(1), 15–32.
and evidence. In M. Hakel (Ed.), Beyond multiple
Lee, T. W., & Mitchell, T. R. (1994). An alternative ap-
choice: Evaluating alternatives to traditional testing
proach: The unfolding model of voluntary em-
for selection (pp. 131–166). Mahwah, NJ: Erlbaum.
ployee turnover. Academy of Management Review,
Hough, L. M., Oswald, F. L., & Ployhart, R. E. (2001). 19(1), 51–89.
Determinants, detection and amelioration of
Lee, T. W., Mitchell, T. R., Sablynski, C. J., Burton, J. P.,
adverse impact in personnel selection procedures:
& Holtom, B. C. (2004). The effects of job embed-
Issues, evidence and lessons learned. International
dedness on organizational citizenship, job perform-
Journal of Selection and Assessment, 9 (1–2),
ance, volitional absences, and voluntary turnover.
152–194.
Academy of Management Journal, 47(5), 711–722.
Johnson, J. W. (2003). Toward a better understanding
of the relationship between personality and indi- Lewin, K. (1951). Field theory in social science. New
vidual job performance. In M. R. Barrick & A. M. York: HarperCollins.
Ryan (Eds.), Personality and work. San Francisco: Linnehan, F. (1998). Examining selective determinants
Jossey-Bass. of job performance for entry and lower-level work-
Judge, T. A., & Bono, J. E. (2001). Relationship of core ers in Mexico. Applied Psychology: An International
self-evaluations traits—self-esteem, generalized Review, 47(4), 547–557.
self-efficacy, locus of control, and emotional stabil- Louis, M. R. (1980). Surprise and sense making: What
ity—with job satisfaction and job performance: newcomers experience in entering unfamiliar
A meta-analysis. Journal of Applied Psychology, organizational settings. Administrative Science
86(1), 80–92. Quarterly, 25(2), 226–251.
Judge, T. A., Heller, D., & Mount, M. K. (2002). Louis, M. R., Posner, B. Z., & Powell, G. N. (1983).
Five-factor model of personality and job satisfac- The availability and helpfulness of socialization
tion: A meta-analysis. Journal of Applied Psychol- practices. Personnel Psychology, 36(4), 857–866.
ogy, 87(3), 530–541.
Maertz, C. P., & Campion, M. A. (2004). Profiles in
Judge, T. A., Locke, E. A., Durham, C. C., & Kluger, quitting: Integrating process and content turnover
A. N. (1998). Dispositional effects on job and life theory. Academy of Management Journal, 47(4),
satisfaction: The role of core evaluations. Journal of 566–582.
Applied Psychology, 83(1), 17–34.
Maertz, C. P., & Griffeth, R. W. (2004). Eight motiva-
Judge, T. A., Thoresen, C. J., Bono, J. E., & Patton, G. tional forces and voluntary turnover: A theoretical
K. (2001). The job satisfaction–job performance synthesis with implications for research. Journal
relationship: A qualitative and quantitative review. of Management, 30(5), 667–683.
Psychological Bulletin, 127(3), 376–407.
March, J. G., & Simon, H. A. (1958). Organizations.
Judge, T. A., & Watanabe, S. (1995). Is the past New York: Wiley.
prologue? A test of Ghiselli’s hobo syndrome. Jour-
Meyer, J. P., & Allen, N. J. (1991). A three-component
nal of Management, 21(2), 211–229.
conceptualization of organizational commitment.
Kanfer, R., & Ackerman, P. L. (2005). Work competence: Human Resource Management Review, 1(1), 61–89.
A person-oriented perspective. In A. J. Elliot & C.
Meyer, J. P., Stanley, D. J., Herscovitch, L., & Topolnyt-
S. Dweck (Eds.), Handbook of competence and mo-
sky, L. (2002). Affective, continuance, and
tivation (pp. 336–353). New York: Guilford Press.
normative commitment to the organization: A
Kipnis, D., & Glickman, A. S. (1962). The prediction of meta-analysis of antecedents, correlates, and
job performance. Journal of Applied Psychology, consequences. Journal of Vocational Behavior,
46(1), 50–56. 61(1), 20–52.

Mitchell, T. R., Holtom, B. C., Lee, T. W., Sablynski, Pynes, J., & Bernardin, H. J. (1992). Mechanical vs
C. J., & Erez, M. (2001). Why people stay: Using consensus-derived assessment center ratings: A
job embeddedness to predict voluntary turno- comparison of job performance validities. Public
ver. Academy of Management Journal, 44(6), Personnel Management, 21(1), 17–28.
1102–1121.
Rhoades, L. & Eisenberger, R. (2002). Perceived
Mitchell, T. R., & James, L. R. (2001). Building better organizational support: A review of the literature.
theory: Time and the specification of when things Journal of Applied Psychology, 87(4), 698–714.
happen. Academy of Management Review, 26(4),
Rynes, S. L. (1991). Recruitment, job choice, and
530–547.
post-hire consequences: A call for new research
Mobley, W. H., Griffeth, R. W., Hand, H. H., & Meglino, directions. In M. D. Dunnette & L. Hough (Eds.),
B. M. (1979). Review and conceptual analysis of the Handbook of industrial and organizational psy-
employee turnover process. Psychological Bulletin, chology (pp. 399–444). Palo Alto, CA: Consulting
86(3), 493–522. Psychologists Press.
Murphy, K. R. (1989). Is the relationship between cog- Schmidt, F. L., & Hunter, J. E. (1998). The validity and
nitive ability and job performance stable over time? utility of selection methods in personnel psychol-
Human Performance, 2(3), 183–200. ogy: Practical and theoretical implications of 85
Niles, F. S. (1999). Toward a cross-cultural understand- years of research findings. Psychological Bulletin,
ing of work-related beliefs. Human Relations, 52(7), 124(2), 262–274.
855–867. Settoon, R. P., Bennett, N., & Liden, R. C. (1996). Social
Ones, D. S. (1993). The construct validity of integrity exchange in organizations: Perceived organizational
tests. Unpublished doctoral dissertation, University support, leader-member exchange, and employee
of Iowa, Iowa City. reciprocity. Journal of Applied Psychology, 81(3),
219–227.
Ones, D. S., Viswesvaran, C., & Schmidt, F. L. (1993).
Comprehensive meta-analysis of integrity test Shore, L. M., Tetrick, L. E., Shore, T. H., & Barksdale, K.
validities: Findings and implications for personnel (2000). Construct validity of measures of Becker’s
selection and theories of job performance. Journal side bet theory. Journal of Vocational Behavior,
of Applied Psychology, 78(4), 679–703. 57(3), 428–444.
O’Reilly, C. A., Caldwell, D. F., & Barnett, W. P. (1989). Spector, P. E. & Jex, S. M. (1998). Development of four
Work group demography, social integration, and self-report measures of job stressors and strain:
turnover. Administrative Science Quarterly, 34(1), Interpersonal Conflict at Work Scale, Organizational
21–37. Constraints Scale, Quantitative Workload Inventory,
and Physical Symptoms Inventory. Journal of Oc-
Organ, D. W. (1994). Personality and organizational cupational Health Psychology, 3(4), 356–367.
citizenship behavior. Journal of Management,
20(2), 465–478. Stajkovic, A. D., & Luthans, F. (1998). Self-efficacy and
work-related performance: A meta-analysis. Psy-
Organ, D. W., & Lingl, A. (1995). Personality, satisfac- chological Bulletin, 124(2), 240–261.
tion, and organizational citizenship behavior. Jour-
Van Dyne, L. & Ang, S. (1998). Organizational citizen-
nal of Social Psychology, 135(3), 339–350.
ship behavior of contingent workers in Singapore.
Owens, W. A., & Schoenfeldt, L. F. (1979). Toward a Academy of Management Journal, 41(6), 692–703.
classification of persons. Journal of Applied Psy-
Van Maanen, J., & Schein, E. H. (1979). Toward a theo-
chology, 63(5), 569–607.
ry of organizational socialization. In B. Staw (Ed.),
Phillips, J. M. (1998). Effects of realistic job previews Research in organizational behavior (pp. 209–264).
on multiple organizational outcomes: A meta- Greenwich, CT: JAI Press.
analysis. Academy of Management Journal, 41(6),
Wanous, J. P. (1980). Organizational entry: Recruit-
673–691.
ment, selection, and socialization of newcomers.
Premack, S. L. & Wanous, J. P. (1985). A meta-analysis Reading, MA: Addison-Wesley.
of realistic job preview experiments. Journal of Ap-
Wanous, J. P., & Colella, A. (1989). Organizational entry
plied Psychology, 70(4), 706–719.
research: Current status and future directions. In G.
Price, J. L., & Mueller, C. W. (1986). Absenteeism and Ferris & K. Rowland (Eds.), Research in personnel
turnover of hospital employees. Greenwich, CT: JAI and human resources management (pp. 59–120).
Press. Greenwich, CT: JAI Press.

Watson, D., & Clark, L. A. (1984). Negative affectivity: Wernimont, P. F., & Campbell, J. P. (1968). Signs, sam-
The disposition to experience negative emotional ples, and criteria. Journal of Applied Psychology,
states. Psychological Bulletin, 96(3), 465–490. 52(5), 372–376.
Watson, D., Clark, L. A., & Tellegen, A. (1988). De- Wright, T. A. (1997). Time revisited in organizational
velopment and validation of brief measures of behavior. Journal of Organizational Behavior, 18(3),
positive and negative affect: The PANAS scales. 201–204.
Journal of Personality and Social Psychology, Wright, T. A. (2002). The importance of time in
54(6), 1063–1070. organizational research. Academy of Management
Weiss, H. M., & Cropanzano, R. (1996). Affective Review, 27(3), 343–345.
events theory: A theoretical discussion of the Zimmerman, R. D., & Darnold, T. C. (2009). The impact
structure, causes, and consequences of affective of job performance on employee turnover inten-
experiences at work. In B. M. Staw & L. L. Cum- tions and the voluntary turnover process:
mings (Eds.), Research in organizational behavior A meta-analysis and path model. Personnel
(Vol. 19, pp. 1–74). Greenwich, CT: JAI Press. Review, 38(2), 142–158.

Hiring For Retention and

Încărcat de

Informații document

Descriere originală:

Titlu original

Drepturi de autor

Formate disponibile

Partajați acest document

Partajați sau inserați document

Opțiuni de partajare

Vi se pare util acest document?

Este necorespunzător acest conținut?

Drepturi de autor:

Formate disponibile

Hiring For Retention and

Încărcat de

Drepturi de autor:

Formate disponibile

HIRING FOR RETENTION AND

This study evaluated the usefulness of several pre-hire variables to predict

Keywords: turnover, job performance, conscientiousness, emotional

Human Resource Management DOI: 10.1002/hrm

Human Resource Management DOI: 10.1002/hrm

Human Resource Management DOI: 10.1002/hrm

Human Resource Management DOI: 10.1002/hrm

posited that individuals with religious or conscientiousness and emotional stability

Human Resource Management DOI: 10.1002/hrm

Human Resource Management DOI: 10.1002/hrm

Human Resource Management DOI: 10.1002/hrm

Human Resource Management DOI: 10.1002/hrm

TABLE I Correlations Among Variables

Human Resource Management DOI: 10.1002/hrm

Human Resource Management DOI: 10.1002/hrm

Human Resource Management DOI: 10.1002/hrm

Up to Six Months After Hire

Human Resource Management DOI: 10.1002/hrm

Entered After Entered After

Human Resource Management DOI: 10.1002/hrm

Human Resource Management DOI: 10.1002/hrm

Model F-value, df .949; 1, 86 1.917; 2, 85 .499; 2, 85 1.448; 3, 84

Human Resource Management DOI: 10.1002/hrm

Human Resource Management DOI: 10.1002/hrm

Human Resource Management DOI: 10.1002/hrm

Human Resource Management DOI: 10.1002/hrm

RYAN D. ZIMMERMAN is an assistant professor of human resource management in the

Human Resource Management DOI: 10.1002/hrm

References value of theory-based biodata for predicting

Human Resource Management DOI: 10.1002/hrm

Human Resource Management DOI: 10.1002/hrm

Human Resource Management DOI: 10.1002/hrm

Human Resource Management DOI: 10.1002/hrm

S-ar putea să vă placă și