contingent workers' impact on standard employee withdrawal behaviors: does what you use them...

30
Human Resource Management, Human Resource Management, January/February 2010, Vol. 49, No. 1, Pp. 109– 138 © 2010 Wiley Periodicals, Inc. Published online in Wiley InterScience (www.interscience.wiley.com). DOI: 10.1002/hrm.20336 CONTINGENT WORKERS’ IMPACT ON STANDARD EMPLOYEE WITHDRAWAL BEHAVIORS: DOES WHAT YOU USE THEM FOR MATTER? SEAN A. WAY, DAVID P. LEPAK, CHARLES H. FAY, AND JAMES W. THACKER Previous research has suggested that workforce mixing—simultaneously using contingent workers and standard employees—can negatively affect standard employee attitudes and behaviors. In this study, we consider the impact of two reasons employers choose to use contingent workers (to enhance standard em- ployee employment stability and to reduce labor costs) on standard employee withdrawal behaviors (absenteeism and turnover). We posit that when the aim of using contingent labor is to enhance standard employee employment stabil- ity (employment stability contingent labor strategy or ESCLS), the effects on standard employee withdrawal behaviors will differ from when the aim is to reduce labor costs (labor cost contingent labor strategy, or LCCLS). Using a sample of 90 firms that employ a mixed workforce, we examine the influence of ESCLS, LCCLS, and high investment HR systems (HIHRS) on standard em- ployee withdrawal behaviors at the firm level. In addition to supporting the hy- pothesized direct (positive) effect of LCCLS on standard employee withdrawal behaviors, this study’s results support the hypothesized moderating effects of HIHRS on the negative relationship between ESCLS and standard employee withdrawal behaviors and the positive relationship between LCCLS and stand- ard employee withdrawal behaviors. Implications for research and practice and suggestions for further research are discussed. © 2010 Wiley Periodicals, Inc. Keywords: contingent workers, standard employee withdrawal behaviors, high investment human resource systems (HIHRS) R esearch has suggested that using contingent labor 1 provides firms with flexibility in deploying human capital (Connelly & Gallagher, 2004; Lepak & Snell, 1999). Al- though scholars have given different labels to this form of flexibility (e.g., numeric flexibility, Atkinson, 1984; coordination flex- ibility, Wright & Snell, 1998), the underlying theme is to enhance the firm’s overall flex- ibility (Connelly & Gallagher, 2004; Lepak & Snell, 2002; Lepak, Takeuchi, & Snell, 2003). Correspondence to: Sean A. Way, Cornell University, School of Hotel Administration, 541 Statler Hall, Ithaca, NY 14853, Phone: 607-255-9017, Fax: 607-254-2971, E-mail: [email protected].

Upload: sean-a-way

Post on 11-Jun-2016

213 views

Category:

Documents


0 download

TRANSCRIPT

Human Resource Management,Human Resource Management, January/February 2010, Vol. 49, No. 1, Pp. 109– 138

© 2010 Wiley Periodicals, Inc.

Published online in Wiley InterScience (www.interscience.wiley.com).

DOI: 10.1002/hrm.20336

CONTINGENT WORKERS’ IMPACT

ON STANDARD EMPLOYEE

WITHDRAWAL BEHAVIORS: DOES

WHAT YOU USE THEM FOR MATTER?

S E A N A . W AY, D A V I D P. L E PA K , C H A R L E S H . F AY, A N D J A M E S W. T H A C K E R

Previous research has suggested that workforce mixing—simultaneously using contingent workers and standard employees—can negatively affect standard employee attitudes and behaviors. In this study, we consider the impact of two reasons employers choose to use contingent workers (to enhance standard em-ployee employment stability and to reduce labor costs) on standard employee withdrawal behaviors (absenteeism and turnover). We posit that when the aim of using contingent labor is to enhance standard employee employment stabil-ity (employment stability contingent labor strategy or ESCLS), the effects on standard employee withdrawal behaviors will differ from when the aim is to reduce labor costs (labor cost contingent labor strategy, or LCCLS). Using a sample of 90 fi rms that employ a mixed workforce, we examine the infl uence of ESCLS, LCCLS, and high investment HR systems (HIHRS) on standard em-ployee withdrawal behaviors at the fi rm level. In addition to supporting the hy-pothesized direct (positive) effect of LCCLS on standard employee withdrawal behaviors, this study’s results support the hypothesized moderating effects of HIHRS on the negative relationship between ESCLS and standard employee withdrawal behaviors and the positive relationship between LCCLS and stand-ard employee withdrawal behaviors. Implications for research and practice and suggestions for further research are discussed. © 2010 Wiley Periodicals, Inc.

Keywords: contingent workers, standard employee withdrawal behaviors, high investment human resource systems (HIHRS)

Research has suggested that using contingent labor1 provides firms with flexibility in deploying human capital (Connelly & Gallagher, 2004; Lepak & Snell, 1999). Al-

though scholars have given different labels

to this form of flexibility (e.g., numeric flexibility, Atkinson, 1984; coordination flex-ibility, Wright & Snell, 1998), the underlying theme is to enhance the firm’s overall flex-ibility (Connelly & Gallagher, 2004; Lepak & Snell, 2002; Lepak, Takeuchi, & Snell, 2003).

Correspondence to: Sean A. Way, Cornell University, School of Hotel Administration, 541 Statler Hall, Ithaca, NY 14853, Phone: 607-255-9017, Fax: 607-254-2971, E-mail: [email protected].

110 HUMAN RESOURCE MANAGEMENT, JANUARY/FEBRUARY 2010

Human Resource Management DOI: 10.1002/hrm

Using contingent labor may allow firms 1) to modify the number of people used to perform tasks according to variations in en-vironmental demand (Cardon, 2003; Harri-son & Kelley, 1993) and 2) to access external skills as circumstances require terminating these relationships when certain skills are no longer needed (Lepak & Snell, 1999; Wright & Snell, 1998). Given these potential ben-

efits, researchers such as Cardon (2003), Lepak and Snell (1999, 2002), Lepak et al. (2003), Matusik and Hill (1998), and Wright and Snell (1998), have suggested that firms may benefit from workforce mixing: that is, simultaneously using contingent workers and standard2 employees.

While there may be organiza-tional benefits associated with workforce mixing, several research-ers have cautioned that workforce mixing may negatively affect stan-dard employee attitudes and be-haviors (see, e.g., David, 2005; Davis-Blake, Broschak, & George, 2003; Pearce, 1993). For instance, Pearce (1993) assessed the impact of contingent labor on standard

employees and found that using contract labor alongside standard employees was as-sociated with diminished trust among stan-dard employees. Similarly, Davis-Blake and colleagues (2003) found that simultaneously using contingent workers and standard em-ployees was negatively related to standard employee loyalty and positively related to standard employee intentions to leave.

Concern surrounding the impact of work-force mixing is not simply theoretical. Esti-mates of the size of the contingent labor workforce range from 11% to 34% of the total employed Canadian workforce (see Vosko, Zukewich, & Cranford, 2003) and from 11% to 31% of the total employed U.S. workforce (see U.S. Bureau of Labor Statistics, 2005; U.S. Government Accountability Office, 2006), de-pending on how contingent workers are de-fined. Using a broad definition of contingent workers, including part-time workers, tempo-rary workers, own-account self-employment

(a self-employed person with no paid employ-ees), and multiple job-holders (two or more concurrent jobs), Vosko et al. (2003) estimated that contingent workers compose 34% of the employed Canadian workforce. Temporary workers (including term or contract, seasonal, casual, temporary agency, and all other em-ployment with a specific predetermined end date) are estimated to compose 11% of the employed Canadian workforce (see Vosko et al., 2003). Again, using a broad definition of contingent workers, including contract com-pany workers, agency temps, on-call workers/day laborers, direct-hire temps, self-employed workers, independent contractors, and stan-dard part-time workers, the General Account-ability Office estimated that contingency workers in the United States made up 30.6% of the employed workforce (U.S. General Ac-countability Office, 2006). The U.S. Bureau of Labor Statistics identified 7.4% of the U.S. workforce as independent contractors, inde-pendent consultants, or freelance workers (regardless of whether they were self-employed or wage and salary workers); 1.8% of the U.S. workforce as on-call workers; 0.9% of the U.S. workforce as temporary help agency workers; and 0.6% of the U.S. workforce as employees of contract firms.

Sufficient numbers of contingent workers are therefore employed in the Canadian and U.S. workforces that the positive and nega-tive effects of having contingent workers working alongside standard employees pres-ent the possibility of real and significant problems to many firms. Although contin-gent workers are more likely to work in professional, construction, and extraction oc-cupations, contingent workers are concen-trated in ways (i.e., they are distributed throughout the major occupational groups) that make it likely for most standard employ-ees to be in regular contact with contingent workers (U.S. Bureau of Labor Statistics, 2005). Workforce mixing is thus likely to be a fact for many firms, and understanding when workforce mixing may have positive or negative effects has practical and theoretical implications.

Although prior research has indicated that workforce mixing may have some potential

The positive and

negative effects of

having contingent

workers working

alongside standard

employees present

the possibility of

real and significant

problems to many

firms.

CONTINGENT WORKERS’ IMPACT ON STANDARD EMPLOYEE WITHDRAWAL BEHAVIORS 111

Human Resource Management DOI: 10.1002/hrm

negative effects on standard employees, re-searchers to date have not addressed the pos-sibility that firms may use contingent labor for different reasons, differences that may in-fluence how standard employees respond to workforce mixing (cf. Connelly & Gallagher, 2004). In this study, we consider the impact of two reasons employers choose to use con-tingent workers (to enhance standard em-ployee employment stability and to reduce labor costs) on standard employee withdrawal behaviors (absenteeism and turnover). We propose that when the aim of using contin-gent labor is to enhance standard employee employment stability (employment stability contingent labor strategy, or ESCLS), the managerial values conveyed to standard em-ployees and the effects of workforce mixing on standard employee withdrawal behaviors will be very different from when the aim of using contingent labor is to reduce labor costs (labor cost contingent labor strategy, or LCCLS) (cf. Bishop, Goldsby, & Neck, 2002; Connelly & Gallagher, 2004). Specifically, we posit that 1) ESCLS will be negatively related to firm-level standard employee withdrawal behaviors, and 2) LCCLS will be positively related to firm-level standard employee with-drawal behaviors.

In addition, it is important to consider organizational context when examining the effects of workforce mixing on standard em-ployees. In particular, the firm’s system of HR practices for its standard employees is likely to influence how these employees re-spond to workforce mixing (cf. David, 2005; Davis-Blake et al., 2003). While prior organi-zation-level research has shown that a negative relationship exists between high investment HR systems3 (HIHRS) and turn-over (e.g., Guthrie, 2001; Huselid, 1995; Shaw, Delery, Jenkins, & Gupta, 1998; Way, 2002), the extent to which firms with a mixed workforce invest in their standard em-ployees via HIHRS may also provide an im-portant contextual lens for how standard employees view and react to workforce mixing.

The premise of this study is that among firms with a mixed workforce, standard em-ployee employment stability enhancement

as a contingent labor strategy (ESCLS) and labor cost reduction as a contingent labor strategy (LCCLS) convey different managerial values to standard employees. These values, in turn, influence how standard employees respond to workforce mixing. Using a sample of 90 firms with a mixed workforce, we em-pirically examine the effects of ESCLS and LCCLS on firm-level standard em-ployee withdrawal behaviors (ab-senteeism and turnover). We also examine whether using HIHRS moderates the performance ef-fects of these contingent labor (workforce mixing) strategies.

Conceptual Model and Hypotheses

Using contingent labor with stan-dard employees (workforce mix-ing) may allow a firm to balance the demands placed on standard employees while also gaining ac-cess to skills and capabilities that exist in the external labor market (Lepak et al., 2003). By doing so, firms in-crease their ability to respond to fluctuations in environmental demand (Bishop et al., 2002; Cappelli & Neumark, 2004; Houseman & Polivka, 2000; Lepak & Snell, 1999; Tsui, Pearce, Porter, & Tripoli, 1997; Wright & Snell, 1998). Yet, while workforce mixing may en-hance the firm’s overall flexibility, it is impor-tant to remain cognizant of how standard employees respond to the use of contingent labor (Connelly & Gallagher, 2004). While we expect, therefore, that firm flexibility may increase regardless of strategic rationale, we posit that the influence of workforce mixing on firm-level standard employee withdrawal behaviors is affected by the firm’s strategic rationale for using contingent workers.

Approaches to Workforce Mixing

Perhaps the most dominant perspective with regard to contingent labor use is that firms rely on contingent labor to reduce labor costs (Davis-Blake & Uzzi, 1993; Kalleberg et al.,

While workforce

mixing may enhance

the firm’s overall

flexibility, it is

important to remain

cognizant of how

standard employees

respond to the use

of contingent labor.

112 HUMAN RESOURCE MANAGEMENT, JANUARY/FEBRUARY 2010

Human Resource Management DOI: 10.1002/hrm

2000; Matusik & Hill, 1998; Polivka & Nar-done, 1989). Because contingent workers are not typically given all of the elements associ-ated with a standard employment relation-ship, such as benefits or training (Houseman,

2001; Kalleberg et al., 2000; Ma-tusik & Hill, 1998), the costs associated with their use are ex-pected to be lower than the costs associated with using standard employees. For example, House-man (2001) demonstrated that saving on benefit costs is an im-portant determinant of contin-gent labor use. Consistent with Houseman’s (2001) findings, other studies have reported results that indicate that training needs are negatively associated with tempo-rary labor use (Davis-Blake & Uzzi, 1993), and the wages an organiza-

tion must pay are positively associated with contracting out labor (Harrison & Kel-ley, 1993). In addition to direct cost savings, Matusik and Hill (1998) noted that using contingent labor can reduce firms’ recruit-ment, socialization, and severance package expenses. And, while there may actually be a wage premium paid to contingent workers (David, 2005; Hipple & Stewart, 1996), the costs of these premiums are often offset by the fact that these workers are paid only for time worked. Thus, by using contingent workers, a firm is more likely to satisfy fluctu-ating staffing needs that result from changes in environmental demand without increas-ing its long-term, fixed labor costs.

In addition to lowering labor costs, firms may use contingent workers to enhance em-ployment stability for their standard employ-ees (Bishop et al., 2002; Lepak & Snell, 1999, 2002; Polivka & Nardone, 1989). As Davis-Blake and Uzzi (1993) noted, when contin-gent workers and standard employees are used together, the firm has a mechanism for developing stable, yet adaptable work ar-rangements. For example, firms may use con-tingent labor to modify the number and type of people who perform tasks to cope with variations in environmental demand. More-over, they may do so without committing

resources toward hiring standard employees or disrupting the productivity of, and de-mands on, their standard employees (Bishop et al., 2002; Davis-Blake & Uzzi, 1993; Lepak & Snell, 1999). Contingent workers can there-fore be used to buffer standard employees from environmental uncertainty (Cappelli & Neumark, 2004; Connelly & Gallagher, 2004; Lepak et al., 2003). Thus, contingent labor becomes the secondary labor market, with standard employees constituting the primary labor market of the firm (Doeringer & Piore, 1971).

When contingent workers work alongside standard employees, it is likely to heighten standard employees’ awareness of the nature of their relationship with their employer (Pearce, 1993). Scholars have suggested that workforce mixing can have a negative impact on standard employee job security (see Davis-Blake et al., 2003). Furthermore, standard employee employment stability enhance-ment as a contingent labor strategy (ESCLS) and labor cost reduction as a contingent labor strategy (LCCLS) convey different man-agerial values to standard employees and are expected to affect standard employees differently. Specifically, although prior re-search has indicated that LCCLS is expected to have a negative effect on standard em-ployee employment security, ESCLS is an approach to workforce mixing that uses con-tingent labor to buffer the standard employee workforce from fluctuations in environmen-tal demand and provide the standard em-ployee workforce with employment security. In other words, ESCLS is expected to have a positive relationship with standard em-ployee employment security. Among firms with a mixed workforce, therefore, ESCLS is also expected to have a different relationship than LCCLS with standard employee with-drawal behaviors such as absenteeism and turnover.4

Davis-Blake et al. (2003) argued that workforce mixing causes some deterioration in the standard employee-employer relation-ship. The social exchange perspective (Blau, 1964) provides a framework for understand-ing employee-employer relationships at both the individual and firm levels (Evans & Davis,

In addition to

lowering labor

costs, firms may

use contingent

workers to enhance

employment stability

for their standard

employees.

CONTINGENT WORKERS’ IMPACT ON STANDARD EMPLOYEE WITHDRAWAL BEHAVIORS 113

Human Resource Management DOI: 10.1002/hrm

2005). Building on the social exchange per-spective (Blau, 1964) and the norm of reciprocity (Gouldner, 1960), Eisenberger, Huntington, Hutchison, and Sowa (1986) developed perceived organizational support (POS) as a central construct to explain the employee-employer relationship. POS refers to employee beliefs concerning the extent to which the firm both values their contribu-tions and cares about their well-being (Eisen-berger et al., 1986; Rhoades & Eisenberger, 2002). POS is also enhanced by positive and discretionary treatment by the firm, which leads employees to perceive the firm is com-mitted to them (Loi, Hang-yue, & Foley, 2006). Based on the norm of reciprocity, em-ployees with high POS are obliged to respond favorably to the firm in the form of positive employee attitudes and behaviors (Loi et al., 2006). In summary, theory suggests that standard employees’ attitudes and behaviors reflect their perceptions and expectations, reciprocating the treatment they receive from their employing firm (Whitener, 2001).

Among firms with a mixed workforce, using contingent labor to enhance employ-ment stability for standard employees (ESCLS) signals to these standard employees that their firm values their contributions and cares about their well-being (cf. Allen, Shore, & Griffeth, 2003). As this contingent labor strategy is likely to be viewed by standard employees as positive and discretionary treat-ment by their firm, ESCLS is expected to enhance standard employee POS (cf. Bishop et al., 2002). Moreover, among firms with a mixed workforce, when firms use contingent labor to enhance standard employee employ-ment stability (ESCLS), they signal to their standard employees that they are seeking to establish or continue a social exchange rela-tionship with them, which prior POS re-search5 rooted in social exchange theory (e.g., Allen et al., 2003; Bishop et al., 2002; Eisenberger, Armeli, Rexwinkel, Lynch, & Rhodes, 2001; Eisenberger et al., 1986; Wayne, Shore, & Liden, 1997) has suggested will lead to greater standard employee POS and lower standard employee absenteeism and turnover (i.e., standard employee withdrawal behav-iors). Thus, among firms with a mixed work-

force, we posit that ESCLS will be negatively related to firm-level standard employee with-drawal behaviors.

In contrast, contingent labor to reduce labor costs (LCCLS) does not signal to the firm’s standard employee work-force that the firm values their contributions or cares about their well-being. The focus of LCCLS is on reducing labor costs and improving efficiency. Although LCCLS may be related to other measures of firm performance (e.g., labor productivity), it is not expected to enhance firm-level standard employee attendance and retention. LCCLS is a contin-gent labor (mixed workforce) strategy that prior research has indicated may negatively impact standard employee employment security, POS, commitment, loy-alty, and retention (see Bishop et al., 2002; Davis-Blake et al., 2003). We posit, therefore, that among firms with a mixed work-force, LCCLS will be positively related to firm-level standard employee withdrawal behaviors.

Hypothesis 1: Among fi rms with a mixed work-force, there will be a negative relationship be-tween employment stability contingent labor strategy (ESCLS) and standard employee with-drawal behaviors.

Hypothesis 2: Among fi rms with a mixed work-force, there will be a positive relationship between labor cost contingent labor strategy (LCCLS) and standard employee withdrawal behaviors.

High Investment HR Systems

Understanding how workforce mixing im-pacts a firm’s standard employees requires considering the HR system the firm has in place for standard employees (Davis-Blake et al., 2003). Whether or not firms use high investment HR systems (HIHRS) has been shown to be an important predictor of valu-able individual and organizational outcomes

By using contingent

workers, a firm

is more likely to

satisfy fluctuating

staffing needs that

result from changes

in environmental

demand without

increasing its long-

term, fixed labor

costs.

114 HUMAN RESOURCE MANAGEMENT, JANUARY/FEBRUARY 2010

Human Resource Management DOI: 10.1002/hrm

across different industries, countries, and business sectors (e.g., Combs, Liu, Hall, & Ketchen, 2006; Guthrie, 2001; Huselid, 1995; Takeuchi, Chen, & Lepak, 2009; Way, 2002). Consistent with extant research, we expect that HIHRS will be negatively related to firm-level standard employee withdrawal be-haviors.

High investment HR systems entail high investments in standard employees

that may facilitate superior em-ployee attendance and retention. HIHRS include the following six parameters: 1) selective staff-ing, 2) above-market compensa-tion, 3) continuous training, 4) frequent experience-enhancing (developmental) opportunities, 5) participation/involvement in de-cision making, 6) performance-based pay. These six parameters, described in Table I, represent those typically found in previous HR systems research (see Combs et al., 2006; Evans & Davis, 2005; Way, 2002).

The rationale for a positive benefit from HIHRS stems from the logic that these systems in-volve investing in standard em-ployees (e.g., selective staffing, Takeuchi et al., 2009; above-mar-ket compensation, Way, 2002; continuous training, Wayne et al., 1997; frequent experience-enhancing (developmental) op-portunities, Allen et al., 2003; Eisenberger et al., 1986) and recognize standard employee con-tributions (e.g., participation/in-volvement in decision making,

Eisenberger et al., 1986; performance-based pay, Whitener, 2001). HR practices that sug-gest investing in standard employees and recognizing their contributions signal to stan-dard employees that the firm is supportive and seeks to establish or continue a social exchange relationship with them (Allen et al., 2003). Perceptions that one’s firm offers these HR practices should enhance standard employees’ perceptions of their firm’s com-

mitment to them (POS) and create a sense of obligation in standard employees to repay the firm. Lower absenteeism (e.g., Eisenberger et al., 2001, 1986) and lower turnover (e.g., Allen et al., 2003; Wayne et al., 1997) have been viewed as means by which employees can repay firms for obligations created by treating them well (Rhoades & Eisenberger, 2002).

Together, the HR parameters included in HIHRS provide a medium through which standard employees and firms form exchange relationships. HIHRS are expected to shape the nature of a firm’s exchange with standard employees, which may induce standard em-ployees to reciprocate in the form of high attendance and a high desire to remain with the firm (e.g., Wayne et al., 1997). Extending this logic, we would anticipate that using HIHRS for standard employees would have a direct negative effect on firm-level standard employee withdrawal behaviors.

Beyond this direct effect, however, we anticipate that using HIHRS may also provide a contextual lens for how standard employ-ees view their exposure to working with contingent labor in a workforce mixing situ-ation. On one hand, ESCLS prioritizes stan-dard employee employment stability and signals to these employees that the firm is seeking to establish or continue a social ex-change relationship (cf. Pfeffer, 1998). This priority on standard employee employment stability is consistent with HIHRS logic and its associated investments in long-term em-ployee well-being and skill development (cf. Cappelli & Neumark, 2004; Pfeffer, 1998; Tsui et al., 1997). Given this reinforcing logic, we anticipate that when used together, ESCLS and HIHRS would have a positive influence on standard employees’ perceptions of their firm’s commitment, feeling of obligation to stay with the firm, and decisions regarding work attendance and maintaining member-ship in the firm (cf. Allen et al., 2003; Davis-Blake et al., 2003; Evans & Davis, 2005; Shaw, Gupta, & Delery, 2005; Tsui et al., 1997; Wayne et al., 1997). In turn, we posit that the interaction between ESCLS and HIHRS will be negatively related to firm-level standard employee withdrawal behaviors.

HIHRS include

the following six

parameters: 1)

selective staffing,

2) above-market

compensation,

3) continuous

training, 4) frequent

experience-

enhancing

(developmental)

opportunities,

5) participation/

involvement in

decision making,

6) performance-

based pay.

CONTINGENT WORKERS’ IMPACT ON STANDARD EMPLOYEE WITHDRAWAL BEHAVIORS 115

Human Resource Management DOI: 10.1002/hrm

TA

BL

E

I

The

Uni

tary

HIH

RS In

dex:

The

Six

HR

Prac

tice

Para

met

ers

and

16 It

ems

Ext

ant

HR

sys

tem

s st

ud

ies

typ

ical

ly in

clu

de

the

follo

win

g p

aram

eter

s (s

ee C

om

bs

et a

l., 2

006;

Eva

ns

& D

avis

, 200

5; W

ay, 2

002)

:1.

Sel

ecti

ve s

taffi

ng

(e.

g.,

Bec

ker

& H

use

lid, 1

998;

Dat

ta e

t al

., 20

05; L

epak

et

al.,

2007

; Tak

euch

i et

al.,

2009

; Way

, 200

2)2.

Ab

ove-

mar

ket

com

pen

sati

on

(e.

g.,

Co

mb

s et

al.,

200

6; S

haw

et

al.,

2005

; Way

, 200

2; Y

ou

nd

t et

al.,

199

6)3.

Co

nti

nu

ou

s tr

ain

ing

(e.

g.,

Co

mb

s et

al.,

200

6; D

eler

y &

Do

ty, 1

996;

Sh

aw e

t al

., 20

05; Z

ach

arat

os

et a

l., 2

005)

4.Fr

equ

ent

exp

erie

nce

-en

han

cin

g o

pp

ort

un

itie

s (e

.g.,

Eva

ns

& D

avis

, 200

5; L

epak

& S

nel

l, 19

99; P

feff

er, 1

998;

Way

, 200

2)5.

Part

icip

atio

n/in

volv

emen

t in

dec

isio

n m

akin

g (

e.g

., B

att,

200

2; C

om

bs

et a

l., 2

006;

Gu

thri

e et

al.,

200

2; P

feff

er, 1

998)

6.Pe

rfo

rman

ce-b

ased

pay

(e.

g.,

Bec

ker

& H

use

lid, 1

998;

Co

mb

s et

al.,

200

6; D

elan

ey &

Hu

selid

, 199

6; H

use

lid, 1

995;

Zac

har

ato

s et

al.,

200

5)

Su

rvey

item

s an

d d

escr

ipti

on

of

pro

ced

ure

s u

sed

to

cre

ate

the

six

HR

pra

ctic

e p

aram

eter

s1

Res

po

nd

ents

wer

e as

ked

to

ind

icat

e (o

n a

5-p

oin

t sc

ale

ran

gin

g f

rom

1 =

nev

er t

o 5

= a

lway

s), o

ver

the

last

2 y

ears

, th

e ex

ten

t to

wh

ich

th

eir

fi rm

has

use

d (

a) a

pti

tud

e, s

kill,

or

gen

eral

inte

llig

ence

tes

ts o

r (b

) p

erso

nal

ity

test

s to

eva

luat

e st

and

ard

em

plo

yee

job

can

did

ates

. S

elec

tive

sta

ffi n

g w

as c

alcu

late

d a

s th

e av

erag

e ra

tin

g f

or

the

mea

sure

’s t

wo

item

s (�

= 0

.72)

.2

Res

po

nd

ents

wer

e as

ked

to

ind

icat

e (o

n a

5-p

oin

t sc

ale

ran

gin

g f

rom

1 =

wel

l bel

ow

ave

rag

e to

5 =

wel

l ab

ove

aver

age)

ove

r th

e la

st

2 ye

ars

in c

om

par

iso

n t

o t

he

mar

ket

the

leve

l of

tota

l co

mp

ensa

tio

n r

ecei

ved

by

thei

r fi

rm’s

typ

ical

sta

nd

ard

em

plo

yee;

ab

ove-

mar

ket

com

pen

sati

on

.3

Res

po

nd

ents

wer

e as

ked

to

ind

icat

e (o

n a

5-p

oin

t sc

ale

ran

gin

g f

rom

1 =

no

ne

to 5

= a

t le

ast

48 h

ou

rs)

over

th

e la

st 2

yea

rs t

he

aver

age

tota

l len

gth

of

(a)

form

al o

rien

tati

on

th

at t

hei

r fi

rm h

as p

rovi

ded

to

its

new

ly h

ired

sta

nd

ard

em

plo

yees

, (b

) fo

rmal

tra

inin

g

that

th

eir

fi rm

has

pro

vid

ed t

o it

s st

and

ard

em

plo

yees

in t

hei

r fi

rst

year

of

emp

loym

ent,

(c)

fo

rmal

tra

inin

g t

hat

th

eir

fi rm

has

pro

vid

ed

to it

s st

and

ard

em

plo

yees

aft

er t

hei

r fi

rst

year

of

emp

loym

ent.

Co

nti

nu

ou

s tr

ain

ing

was

cal

cula

ted

as

the

aver

age

rati

ng

fo

r th

e m

easu

re’s

th

ree

item

s (�

= 0

.79)

.4

Res

po

nd

ents

wer

e as

ked

to

ind

icat

e (o

n a

5-p

oin

t sc

ale

ran

gin

g f

rom

1 =

no

ne

to 5

= a

ll) t

he

pro

po

rtio

n o

f th

eir

fi rm

’s s

tan

dar

d e

mp

loye

es

invo

lved

in (

a) t

emp

ora

ry–

(b)

cro

ss-f

un

ctio

nal

–wo

rk g

rou

ps/

team

s in

th

e la

st 2

yea

rs.

Freq

uen

t ex

per

ien

ce-e

nh

anci

ng

(d

evel

op

men

tal)

o

pp

ort

un

itie

s w

as c

alcu

late

d a

s th

e av

erag

e ra

tin

g f

or

the

mea

sure

’s t

wo

item

s (�

= 0

.72)

.5

Res

po

nd

ents

wer

e as

ked

to

ind

icat

e (o

n a

5-p

oin

t sc

ale

ran

gin

g f

rom

1 =

no

infl

uen

ce t

o 5

= a

ver

y g

reat

dea

l of

infl

uen

ce)

over

th

e la

st 2

yea

rs t

he

leve

l to

wh

ich

th

eir

fi rm

’s s

tan

dar

d e

mp

loye

es h

ave

bee

n a

ble

to

infl

uen

ce d

ecis

ion

s re

gard

ing

(a)

sta

nd

ard

em

plo

yee

hea

lth

an

d s

afet

y is

sues

, (b

) th

e ti

mef

ram

e in

wh

ich

sta

nd

ard

wo

rk w

as s

ched

ule

d (

e.g

., ov

erti

me,

fl ex

tim

e, a

nd

sh

ort

-tim

e), (

c) h

ow

to

imp

rove

th

e p

hysi

cal e

nvir

on

men

t o

f st

and

ard

em

plo

yees

, (d

) th

e im

ple

men

tati

on

rat

e o

f te

chn

ical

ch

ang

es

that

imp

act

stan

dar

d e

mp

loye

e w

ork

, (e)

layo

ffs

and

th

e h

irin

g o

f st

and

ard

em

plo

yees

, (f)

ho

w t

o im

pro

ve s

ervi

ce t

o c

ust

om

ers.

Pa

rtic

ipat

ion

/invo

lvem

ent

in d

ecis

ion

mak

ing

was

cal

cula

ted

as

the

aver

age

rati

ng

fo

r th

e m

easu

re’s

six

item

s (�

= 0

.71)

.6

Res

po

nd

ents

wer

e as

ked

to

ind

icat

e (o

n a

5-p

oin

t sc

ale

ran

gin

g f

rom

1 =

no

infl

uen

ce t

o 5

= a

ver

y g

reat

dea

l of

infl

uen

ce)

over

th

e la

st 2

yea

rs t

he

infl

uen

ce t

hat

per

form

ance

ap

pra

isal

s h

ave

had

wh

en d

eter

min

ing

(a)

bo

nu

s o

r in

cen

tive

pay

, (b

) w

age

incr

ease

s fo

r st

and

ard

em

plo

yees

. Pe

rfo

rman

ce-b

ased

pay

was

cal

cula

ted

as

the

aver

age

rati

ng

fo

r th

e m

easu

re’s

tw

o it

ems

(� =

0.7

8).

116 HUMAN RESOURCE MANAGEMENT, JANUARY/FEBRUARY 2010

Human Resource Management DOI: 10.1002/hrm

While ESCLS may be viewed as consis-tent with HIHRS, among firms that exten-sively use HIHRS, using contingent labor to reduce labor costs (LCCLS) may be inter-preted as inconsistent with HIHRS principles. Based on the belief that employees are an asset to be developed rather than a cost to be minimized, using HIHRS is posited to pro-vide a medium through which standard employees and firms form exchange rela-tionships. Exposing standard employees to these types of HR systems in conjunction with LCCLS is likely to convey mixed mes-sages to standard employees regarding their role in the firm and their relationship with the firm. As Rousseau (1995) noted, inconsis-tent messages or conflicting signals regard-ing the nature of the employee-employer exchange can confuse employees regarding how they are to perform on the job to fulfill their end of the relationship. These conflict-ing signals would logically diminish stan-dard employees’ commitment to the firm and enhance standard employee withdrawal behavior. Thus, we posit that the interaction between LCCLS and HIHRS will be positively related to firm-level standard employee with-drawal behaviors.

Hypothesis 3: Among fi rms with a mixed work-force, using HIHRS will negatively moderate the relationship between employment stability contingent labor strategy (ESCLS) and standard employee withdrawal behaviors, such that the relationship will be stronger when HIHRS use is high.

Hypothesis 4: Among fi rms with a mixed work-force, using HIHRS will positively moderate the relationship between labor cost contingent labor strategy (LCCLS) and standard employee with-drawal behaviors, such that the relationship will be stronger when HIHRS use is high.

Methods

Data and Sample

From a listing of 3,000 Canadian for-profit firms that employed 100 or more employees purchased from Dun & Bradstreet (D&B)

Canada6 we identified the senior HR manager from 1,950 firms. From this list of 1,950 se-nior HR managers, the Environment, Behav-ior, and Risk Research Laboratory (EBRRL) at the University of Arizona randomly selected 649 senior HR managers and asked them to participate in this study; 142 agreed. Next, EBRRL used computer-assisted telephone in-terviewing to administer a survey to the 142 senior HR managers who agreed to partici-pate in the study. This response rate (21.9%) is consistent with prior organization-level HR systems studies (e.g., Becker & Huselid, 1998; Datta, Guthrie, & Wright, 2005; Huselid, 1995). Analyzing the respondents and nonre-spondents on firm differences based on industry membership and firm size (data in-cluded in the D&B listing) showed that the two groups were not significantly different on these two dimensions.

All cases with missing values on variables (13 cases) and cases in which the firm’s re-spondents reported that their firm had not used contingent labor7 in the last two years (39 cases) were deleted.8 This reduced our final sample to 90 firms with a mixed work-force. Within our final mixed workforce sample, 32.2% of firms were unionized (i.e., had employees who were covered by a collective agreement), 42.2% of firms were from the manufacturing sector, 13.3% were from the transportation/utilities sector, 5.6% were from the retail trade sector, 21.1% were from the services sector, 10.0% were from the banking/real estate/insurance sector, and 7.8% were from the wholesale trade sector. In terms of size, 53.3% of firms employed between 100 and 249 employees, 28.9% em-ployed between 250 and 499 employees, and 17.8% employed 500 or more employees (mean = 346.48, s.d. = 373.56, minimum = 100, maximum = 2,000).

In Appendix A, we present descriptive sta-tistics for our study’s entire sample (n = 129), the final mixed workforce sample (n = 90; the firms that reported they had used contingent workers in the last two years), and the ex-cluded cases sample (n = 39; the firms that reported they had not used contingent work-ers in the last two years). The firms that had and had not (mixed workforce sample versus

CONTINGENT WORKERS’ IMPACT ON STANDARD EMPLOYEE WITHDRAWAL BEHAVIORS 117

Human Resource Management DOI: 10.1002/hrm

excluded cases sample) used contingent labor in the last two years were then analyzed based on differences in the following variables: in-dustry membership (six different industry membership dummy codes), firm size, firm age, union presence (dummy code), grievance procedure (dummy code), HIHRS, selective staffing, above-market compensation, con-tinuous training, frequent experience-enhancing (developmental) opportunities, participation/involvement in decision mak-ing performance-based pay, standard em-ployee withdrawal behaviors, standard em-ployee absenteeism, and standard employee turnover. The results showed that the two groups were only statistically different on one dimension: above-market compensation. That is, the firms from the mixed workforce sample made greater use of above-market compensa-tion than firms from our excluded cases sam-ple (see Appendix A).

Variables and Measures

Standard Employee Withdrawal Behaviors

Studies examining the relationship between HR systems and firm-level outcomes (perfor-mance) have typically conceptualized firm-level outcomes in terms of HR outcomes (e.g., absenteeism, turnover), organizational outcomes (e.g., productivity, quality), finan-cial/accounting outcomes (e.g., return of in-vestment, profitability), or, for publicly traded firms, capital market outcomes (e.g., shareholder return, Tobin’s q). Prior research has suggested that HR outcomes are most directly impacted by HR systems (e.g., Combs et al., 2006; Guthrie, 2001; Way & Johnson, 2005). Given the proximal nature of these outcomes, this paper focused on firm-level standard (full-time, nonmanagement) em-ployee absenteeism and turnover: that is, firm-level standard employee withdrawal behaviors.

Although perceptual measures of firm-level outcomes may increase measurement error and the potential for common method bias (Gerhart, Wright, & McMahan, 2000), comparing absenteeism and turnover rates (which are likely to be affected by industry-

related factors) across diverse industry set-tings may be misleading. Moreover, prior HR systems studies have used perceptual mea-sures of performance (e.g., Bae & Lawler, 2000; Delaney & Huselid, 1996; Youndt, Snell, Dean, & Lepak, 1996; Way, 2002), and studies have shown that perceptual and ob-jective measures of firm performance are highly correlated (Robinson & Pearce, 1988; Venkatraman & Ramanujam, 1986; Zahra, 1993). Because the contextual pressures (com-petitive market, strategy, structure, etc.) that firms face are likely to lead firms to focus on and prioritize generating some outcomes over others (Gupta & Govindarajan, 1984; Rogers & Wright, 1998; Way & Johnson, 2005), scholars have suggested that firm-level measures of performance should reflect the degree to which a firm is generating out-comes that have strategic value to the firm (e.g., Covin & Slevin, 1989; Gupta & Govin-darajan, 1984; Way & Johnson, 2005).

We measure firm-level standard employee withdrawal behaviors with a modified ver-sion of an instrument that has been used in the strategy, marketing, and entrepreneur-ship literatures (see Covin & Slevin, 1989; Govindarajan & Fisher, 1990; Gupta & Gov-indarajan, 1984; Miles, Covin, & Heeley, 2000; Zahra, 1993). The basic idea is that items contained in the measure, in this case standard employee absenteeism and turn-over, have different importance for firms; therefore, in capturing the top management team’s perception, feeling, or evaluation of these items, information on the items’ im-portance should be incorporated in the scor-ing procedure and reflected in the final score (Wu & Yao, 2006, pp. 485–486). Accordingly, “importance weighting” is proposed to serve this purpose, and the common procedure is to weight satisfaction score by importance score at the item level (Wu & Yao, 2006, p. 486). This measure’s scoring scheme reflects the assumption that overall managerial per-ception of standard employee withdrawal behaviors is a composite of the top HR man-agement team’s satisfactions in particular areas of standard employee withdrawal be-haviors (e.g., standard employee absentee-ism, standard employee turnover) weighted

118 HUMAN RESOURCE MANAGEMENT, JANUARY/FEBRUARY 2010

Human Resource Management DOI: 10.1002/hrm

by their relative importance to the top HR management team. The product of the satis-faction and importance ratings for each area of withdrawal behaviors is computed.

Our standard employee withdrawal be-haviors measure reflects the degree to which a firm is generating outcomes that have stra-tegic value to the firm’s top HR management team (see Appendix B). Following Zahra (1993, p. 56), each criterion (standard em-ployee absenteeism and standard employee workforce turnover) was rated twice by re-spondents (each firm’s most knowledgeable person in HR): once for its importance to the firm’s top HR management team (rated as of little importance, somewhat important, mod-erately important, quite important, or very important) and once to gauge the firm’s top HR management team’s satisfaction with per-formance on the criterion (rated as very dis-satisfied, dissatisfied, neither dissatisfied nor satisfied, satisfied, or very satisfied). Tradi-tionally, prior research (e.g., Zahra, 1993) has, for each criterion, multiplied the importance rating (rated from 1 to 5 with 1 = of little im-portance and 5 = very important) by the sat-isfaction score for each criterion (rated from 1 to 5 with 1 = very dissatisfied and 5 = very satisfied). In other words, prior research has used the data on criterion importance as weights. For each firm, a weighted average performance index is obtained (Gupta & Go-vindarajan, 1984). Following this approach, we next created two measures (see Appendix B): standard employee attendance (mean = 14.31; s.d. = 5.90; minimum = 2.00; maxi-mum = 25.00); standard employee retention (mean = 13.83; s.d. = 7.13; minimum = 1.00; maximum = 25.00). Then, a single, standard employee HR outcome measure was calcu-lated as the sum of the standard employee attendance and retention scores (mean = 28.14; s.d. = 11.58; minimum = 3.00; maxi-mum = 50.00).

Given the self-report nature of this perfor-mance measure, it is worth noting that Zahra (1993) reported that a high correlation existed between perceptual measures of sales growth and ROA created using the technique outlined earlier and objective measures of sales growth (r = .88; p < .001) and ROA (r = .83; p < .001).

If the procedure outlined previously was fol-lowed, a firm would have an employee reten-tion score of 5 if standard employee retention was of “little importance” to the firm’s top HR management team and the firm’s top HR management team was “highly satisfied” with standard employee retention (1 * 5 = 5). This would be the same employee retention score (5) that the firm would be assigned if standard employee retention was “extremely impor-tant” to the firm’s top HR management team and the firm’s top HR management team was “highly dissatisfied” with standard employee retention (5 * 1 = 5).9 Furthermore, the depen-dent variable of interest in this study is firm-level standard employee withdrawal behavior (absenteeism and turnover), not standard em-ployee HR outcome performance (attendance and retention).

To overcome the problems described pre-viously, we rescored the satisfaction anchors (rated from –2 to 2 with –2 = very satisfied and 2 = very dissatisfied). Then, for each cri-terion, we multiplied the satisfaction score (rated from –2 to 2 with –2 = very satisfied and 2 = very dissatisfied) by its importance rating (rated from 1 to 5 with 1 = of little im-portance and 5 = very important). See Appen-dix B. This resulted in standard employee absenteeism (mean = –2.14; s.d. = 4.51; mini-mum = –10.00; maximum = 10.00) and stan-dard employee turnover (mean = –2.30; s.d. = 4.90; minimum = –10.00; maximum = 10.00). Finally, a single firm-level standard employee withdrawal behaviors measure was calculated as the sum of the standard employee absen-teeism and turnover scores (Cronbach’s alpha = 0.71). This variable was used in the subse-quent analysis (mean = –4.44; s.d. = 8.28; minimum = –20.00; maximum = 20.00). As we report in Appendix B, the pattern of re-sults was the same whether the firm-level standard employee withdrawal behaviors variable was created following the procedure that we followed in this study or that re-searchers have traditionally followed.

Contingent Labor Strategies

Employment stability contingent labor strat-egy (ESCLS) was assessed using a single

CONTINGENT WORKERS’ IMPACT ON STANDARD EMPLOYEE WITHDRAWAL BEHAVIORS 119

Human Resource Management DOI: 10.1002/hrm

survey item in which the respondent from each firm (i.e., the most knowledgeable person in HR) was asked to indicate how often (1 = never to 5 = always) the firm’s con-tingent workers were used in the last two years to provide employment stability for the firm’s standard employees (mean = 2.03; s.d. = 1.04; minimum = 1; maximum = 5). Labor cost contingent labor strategy (LCCLS) was assessed using a single survey item in which the respondent from each firm was asked to indicate how often (1 = never to 5 = always) the firm’s contingent workers were used to reduce its labor costs in the last two years (mean = 1.79; s.d. = 0.98; minimum = 1; maximum = 5). We standardized (mean-cen-tered) both of these contingent workforce strategy variables because they were used to create multiplicative interaction terms (to test Hypotheses 3 and 4).

In this article we posit that, with all else being equal, pursuing each contingent labor strategy (ESCLS and LCCLS) will have a dis-tinct relationship with firm-level standard employee withdrawal behaviors. Because ESCLS and LCCLS were not proposed to be mutually exclusive (see note 4), both the ESCLS item and the LCCLS item were scored separately: that is, respondents scored each item separately on a 5-point scale ranging from 1 = never to 5 = always. Within our sample of 90 firms with a mixed workforce, the correlation between ESCLS and LCCLS was low (r = .03) and not statistically signifi-cant (see Tables II and III). A negative and statistically significant correlation between ESCLS and LCCLS emerged, however, when firms that had not used either ESCLS or LCCLS in the last two years were filtered from our data set (r = –.35, p < .01). In addition, the ESCLS and LCCLS frequency data presented in Appendix C indicate that firms that use one of these contingent labor strategies tend not to use the other.

HIHRS Index

Although studies have used a variety of ap-proaches to measure HIHRS, there appears to be a general consensus among scholars that a system of HR practices (rather than single,

isolated practices) is the appropriate level of analysis (Becker & Huselid, 1998; Delery & Shaw, 2001). This study’s HIHRS index in-cludes six HR practice parameters (see Table I, Table II, and Appendix D), which reflect dis-tinct but interrelated investments in standard employee development and long-term mutu-ally beneficial and cooperative standard em-ployee-employer relationships. Together these are expected to have a negative effect on standard employee absenteeism and turn-over (cf. Allen et al., 2003; Cappelli & Neu-mark, 2004; Evans & Davis, 2005; Guthrie, 2001; Tsui et al., 1997; Way, 2002). In Appen-dix D, we review an exploratory examination of the relationship between the dependent variable (firm-level standard employee with-drawal behaviors) and the six HR practice parameters used to construct our HIHRS index.

We measured HIHRS as an additive index,10 which is appropriate for systems composed of distinct components that in combination are expected to influence be-havior (see Edwards, 2001). Consistent with the existing literature (e.g., Becker & Huselid, 1998; Combs et al., 2006; Datta et al., 2005; Guthrie, Spell, & Nyamori, 2002; Way, 2002; Youndt et al., 1996; Zacharatos, Barling & Iverson, 2005) and the goals of our research,11 information concerning using selective staffing, above-market compensation, con-tinuous training, frequent experience-enhancing (developmental) opportunities, participation/involvement in decision-making, and performance-based pay for stan-dard employees (Table I) is combined in a unitary index that captures the extent to which a firm invests in its standard employ-ees (Becker & Huselid, 1998; Combs et al., 2006; Evans & Davis, 2005; Way, 2002). Spe-cifically, the value of the unitary HIHRS index for a firm is the sum of the system’s six HR practice components described in Table I (mean = 16.07; s.d. = 3.00; minimum = 8.33; maximum = 24.00). Because we used the HIHRS index to create multiplicative interac-tion terms (to test Hypotheses 3 and 4) we transformed the HIHRS index to a z-score (mean = 0.00; s.d. = 1.00; minimum = –2.58; maximum = 2.64).

120 HUMAN RESOURCE MANAGEMENT, JANUARY/FEBRUARY 2010

Human Resource Management DOI: 10.1002/hrm

TA

BL

E

II

Des

crip

tive

Stat

istic

s an

d Co

rrel

atio

ns: C

ontin

gent

Wor

kfor

ce S

trat

egy,

HIH

RS, a

nd H

R Pr

actic

e Va

riab

lesª

Vari

ab

les

12

34

56

78

9

Co

rrel

atio

ns

1E

SC

LS2

LCC

LS .0

33

HIH

RS

ind

ex .2

0† .0

54

Sel

ecti

ve s

taffi

ng

.08

.12

.58*

*5

Ab

ove-

mar

ket

com

pen

sati

on

–.06

.12

.32*

* .0

06

Co

nti

nu

ou

s tr

ain

ing

.05

–.06

.57*

* .1

0 .2

0†

7Fr

equ

ent

exp

erie

nce

-en

han

cin

g o

pp

ort

un

itie

s .1

7 .0

1 .4

6**

.06

.12

.18†

8Pa

rtic

ipat

ion

/invo

lvem

ent

in d

ecis

ion

mak

ing

.02

.14

.48*

* .1

0 .0

7 .1

8† .2

6*9

Perf

orm

ance

-bas

ed p

ay .2

4*–.

12 .4

9**

.00

–.04

.16

.00

.13

Des

crip

tive

sta

tist

ics

Min

imu

m–1

.03

–0.7

9–2

.58

1.00

2.00

1.00

1.00

1.00

1.00

Max

imu

m 2

.97

3.2

1 2

.64

5.00

5.00

5.00

5.00

4.33

5.00

Mea

n 0

.00

0.0

0 0

.00

2.94

3.32

2.79

1.94

2.53

2.54

Sta

nd

ard

dev

iati

on

1.0

4 0

.98

1.0

01.

550.

680.

930.

880.

661.

28

ª n

= 9

0.

† p

< .1

0, t

wo

-tai

led

tes

t; *

p <

.05,

tw

o-t

aile

d t

est;

**

p <

.01,

tw

o-t

aile

d t

est.

CONTINGENT WORKERS’ IMPACT ON STANDARD EMPLOYEE WITHDRAWAL BEHAVIORS 121

Human Resource Management DOI: 10.1002/hrm

TA

BL

E

II

I

D

escr

iptiv

e St

atis

tics

and

Corr

elat

ions

for S

tudy

’s Ke

y Va

riab

les

Vari

ab

les

12

34

56

78

910

11

12

13

14

15

16

Co

rrel

atio

ns

1M

anu

fact

uri

ng

2Tr

ansp

ort

atio

n/u

tilit

ies

–.34

**

3R

etai

l–.

21*

–.10

4S

ervi

ces

–.44

**–.

20–.

13

5B

anki

ng

/rea

l est

ate/

fi n

ance

–.28

**–.

13–.

08–.

17

6Fi

rm s

ize

.21*

–.02

.05

–.11

–.02

7Fi

rm a

ge

–.04

–.10

.01

.01

.26*

.23*

8U

nio

n .3

3**

.01

–.06

–.30

**–.

15 .2

1* .0

0

9G

riev

ance

.10

–.04

.05

–.03

–.11

–.02

–.09

.12

10C

on

tin

gen

t w

ork

forc

e si

ze .0

6 .2

0†–.

02–.

08–.

11 .5

4**

–.03

.03

.02

11E

SC

LS .0

6 .0

2–.

10–.

12 .0

2–.

22*

–.09

–.16

–.07

–.06

12LC

CLS

–.02

.25*

.10

–.17

–.12

–.03

–.15

.10

–.19

†.2

2* .0

3

13H

IHR

S–.

14 .0

9–.

17.0

4 .1

5 .0

6 .0

0–.

14 .1

3–.

04 .2

0†.0

5

14S

tan

dar

d e

mp

loye

e w

ith

dra

wal

b

ehav

iors

.0

9–.

08 .0

2–.

07–.

06–.

10 .0

8–.

01 .2

6*–.

15–.

09.0

8–.

19†

15S

tan

dar

d e

mp

loye

e ab

sen

teei

sm .1

1–.

15–.

08–.

07–.

05–.

04 .1

4 .0

3 .2

0*–.

18†

–.07

.03

–.09

.87*

*

16S

tan

dar

d e

mp

loye

e tu

rnov

er .0

5 .0

0 .1

0–.

05–.

06–.

14 .0

1–.

06 .2

5*–.

10–.

09.1

1–.

23*

.89*

* .5

5**

Des

crip

tive

sta

tist

ics

Min

imu

m0.

000.

000.

000.

000.

004.

611.

000.

000.

00–3

8.56

–1.0

3–0

.79

–2.5

8–2

0.00

–10.

00–1

0.00

Max

imu

m1.

001.

001.

00

1.00

1.00

8.01

100.

001.

001.

00 4

85.4

4 2

.97

3.2

1 2

.64

20.

00 1

0.00

10.

00

Mea

n0.

420.

130.

060.

210.

105.

5024

.85

0.32

0.71

0.0

0 0

.00

0.0

0 0

.00

–4.4

4–2

.14

–2.3

0

Sta

nd

ard

dev

iati

on

0.50

0.34

0.23

0.41

0.30

0.78

25.2

90.

470.

46 8

3.18

1.0

4 0

.98

1.0

0 8

.28

4.5

1 4

.90

ª n

= 9

0.

† p

< .1

0, t

wo

-tai

led

tes

t; *

p <

.05,

tw

o-t

aile

d t

est;

**

p <

.01,

tw

o-t

aile

d t

est.

122 HUMAN RESOURCE MANAGEMENT, JANUARY/FEBRUARY 2010

Human Resource Management DOI: 10.1002/hrm

Table I displays the survey items used to measure each of the six HR practice parame-ters. As noted, the six HR practice parameters represent those typically found in previous research and cover the major areas of HR practice: selection, training, pay for perfor-mance, and participation (Wright, Gardner, & Moynihan, 2003, p. 28). Descriptive statis-tics and the correlations among the contin-gent worker variables (ESCLS and LCCLS), the HIHRS index, and the six HR practice pa-rameters used to construct the HIHRS index are shown in Table II.

Control Variables

Consistent with previous studies (e.g., Guth-rie et al., 2002; Way, 2002), we used dummy variables to control for industry membership (information extracted from the D&B listing). The industry groups represented include man-ufacturing, transportation/utilities, retail trade, service, banking/real estate/insurance, and wholesale trade (the omitted category). We controlled for firm size as it was expected to be positively associated with using HIHRS (e.g., Datta et al., 2005; Way, 2002). Firm size was measured as the natural logarithm of the average number of employees the firm em-ployed in the last year (information from the D&B listing). We also control for the effects of firm age with information extracted from the D&B listing. Specifically, firm age was mea-sured as the number of years a firm has been under the control of its current management. Prior research indicates that unions may influence HIHRS use and the performance ef-fects associated with using these systems (see Becker & Huselid, 1998; Cooke, 1994; Lepak et al., 2003). A dummy code for union pres-ence was created for each firm by assigning a 1 if the firm’s respondent reported that the firm employed standard employees who were covered by a collective agreement; otherwise, a 0 was assigned. To control for the effects of a formal grievance procedure (see Becker & Huselid, 1998), a dummy code was created for each firm by assigning a 1 if the respondent reported that the firm had a formal process through which standard employees could re-solve work-related disputes in place; other-

wise, a 0 was assigned. Finally, prior research implies that the size of the contingent work-force (the number of contingent workers present) may negatively influence standard employee morale and perceived organiza-tional trust (e.g., Chattopadhyay & George, 2001). Contingent workforce size12 was calcu-lated by multiplying the number of employ-ees each firm employed (information extracted from the D&B listing) by the degree to which each firm used contingent workers (derived from a single item in which the firm’s respondent was asked “Over the last 2 years, what percentage of your firm’s total workforce, on average, consisted of contin-gent workers?”)

Analyses

Moderated regression analyses were used to test our hypotheses (Baron & Kenny, 1986; Cohen & Cohen, 1983). In step 1, all controls were entered into the regression equation. To test the direct relationships, the independent variables (ESCLS, LCCLS, and HIHRS) were entered into the regression equation in step 2. To test for moderation, we created two interac-tion terms, which were entered into the equation in step 3 to determine the amount of additional variance explained by the interac-tions: 1) ESCLS × HIHRS (mean = 0.21; s.d. = 0.94; minimum = –2.46; maximum = 3.23) and 2) LCCLS × HIHRS (mean = 0.05; s.d. = 1.08; minimum = –5.70; maximum = 4.49). To reduce multicollinearity, the data used to cre-ate interaction terms were standardized (Cohen & Cohen, 1983, p. 325). The HPWS index was transformed into a z-score, and the ESCLS and LCCLS variables were mean-centered.

Results

Descriptive statistics and the correlations among the contingent worker variables (ESCLS and LCCLS), the HIHRS index, and the six HR practice parameters used to construct the HIHRS index are presented in Table II. As we report, ESCLS was positively correlated with HIHRS (p < .10) and performance-based pay (p < .05). LCCLS was not correlated with ESCLS, HIHRS, or any of the six HR practice

CONTINGENT WORKERS’ IMPACT ON STANDARD EMPLOYEE WITHDRAWAL BEHAVIORS 123

Human Resource Management DOI: 10.1002/hrm

parameters used to construct the HIHRS index. Examining the correlations among the vari-ables included in Table II and all of our study’s control variables13 revealed the following:

Selective staffi ng was positively correlated with the transportation/utilities industry (p < .10) and negatively correlated with the retail trade industry (p < .05).

Continuous training was negatively corre-lated with the manufacturing industry (p < .05).

Frequent experience-enhancing (develop-mental) opportunities were positively correlated with the manufacturing in-dustry (p < .10).

Performance-based pay was positively correlated with the service industry (p < .10) and the banking/real estate/insurance industry (p < .05) and negatively correlated with the manufacturing industry (p < .01) and union presence (p < .01).

Descriptive statistics and the correlations among our study’s key variables are presented in Table III. As we report, ESCLS was nega-tively correlated with firm size (p < .05) and positively correlated with HIHRS (p < .10). HIHRS use was negatively correlated with standard employee withdrawal behaviors (p < .10) and turnover (p < .05). LCCLS was positively correlated with the transportation/utilities industry (p < .05) and contingent workforce size (p < .05) and negatively corre-lated with the formal grievance procedure dummy code (p < .10).

Table IV reports the results of the moder-ated regression analyses. Among the control variables, only the grievance procedure dummy variable (� = –4.85, p < .05) reached statistical significance in step 1 of the regres-sion equation (see Table IV, Model 1).

Step 2 (see Table IV, Model 2), as a set, explained a significant amount of variance in firm-level standard employee withdrawal be-haviors (� R ² = .11, p < .05). In step 2, among firms with a mixed workforce, standard em-ployee withdrawal behaviors was positively related to LCCLS (� = 2.32, p < .05) and nega-

1.

2.

3.

4.

tively related to HIHRS (� = –2.35, p < .05). The relationship between standard employee withdrawal behaviors and ESCLS (� = –0.55) was not statistically significant. Thus, we found support for Hypothesis 2, but not for Hypothesis 1.

Step 3 (see Table IV, Model 3), as a set, explained a significant amount of variance in firm-level standard employee withdrawal be-haviors (� R ² = .06, p < .05). In step 3, results provide support for Hypothesis 3 and Hy-pothesis 4. As we posited in Hypothesis 3, the interaction between ESCLS and HIHRS (i.e., the ESCLS × HIHRS interaction term) was negatively related to standard employee with-drawal behaviors (� = –1.81, p < .05). Figure 1 illustrates this interaction effect, indicating further support for Hypothesis 3. As we pos-ited in Hypothesis 4, the interaction between LCCLS and HIHRS (i.e., the LCCLS × HIHRS interaction term) was positively related to standard employee withdrawal behaviors (� = 1.38, p < .10). Figure 2 illustrates this interac-tion effect, indicating further support for Hypothesis 4.

Supplemental Moderated Regression Analyses

In Table V, we report the results of supple-mental moderated regression analysis that we conducted to explore the interactive effects of ESCLS and HIHRS and LCCLS and HIHRS on standard employee turnover (Table V, Model 1) and absenteeism (Table V, Model 2). Consistent with the results reported in Table IV, Model 3, the results presented in Table V, Model 1, show that the relationship between standard employee turnover and 1) ESCLS was not statistically significant (� = –0.21, p > 0.10), 2) LCCLS was positive (� = 1.21, p < 0.05), 3) HIHRS was negative (� = –1.52, p < .01), and 4) the ESCLS × HIHRS interaction term was negative (� = –1.17, p < .05). Note that the results reported in Table V, Model 1, show that the positive relationship between standard employee turnover and the LCCLS × HIHRS interaction term was not sta-tistically significant (� = 0.54, p > 0.10).

Consistent with the results reported in Table IV, Model 3, the results presented in

124 HUMAN RESOURCE MANAGEMENT, JANUARY/FEBRUARY 2010

Human Resource Management DOI: 10.1002/hrm

Table V, Model 2, show that the relationship between standard employee absenteeism and 1) ESCLS was not statistically significant (� = -0.21, p > 0.10), 2) LCCLS was positive (� = 1.00, p < .10), 3) HIHRS was negative (� = –0.85, p < .10), and 4) the LCCLS × HIHRS interaction term was positive (� = 0.85, p < .10). Note that the results reported in Table V, Model 2, show that the negative relationship between standard employee absenteeism and the ESCLS × HIHRS interaction term was not statistically significant (� = –0.64, p > 0.10).

Assessing the Robustness of Our Findings

Supplemental moderated regression analyses were conducted to demonstrate that the rela-tionships reported in this study are robust

and not a function of multicollinearity and to show that contingent workforce size does not influence our study’s reported findings (see Appendix E). The results of these supple-mental moderated regression analyses (see Appendix E) parallel those presented in Table IV and confirm our results.

Discussion

Scholars have suggested that within today’s competitive environment, simultaneously using both contingent workers and standard employees (workforce mixing) may enable firms to gain access to people who possess the skills necessary to respond to fluctuations in environmental demand (e.g., Lepak et al., 2003). At the same time, however, other re-searchers have cautioned that workforce mix-

T A B L E I V Moderated Regression Results: Predicting Firm–Level Standard Employee Withdrawal Behavior

Standard Employee Withdrawal Behaviors

Model 1 Model 2 Model 3

Variables B t-Score B t-Score B t-Score

Step 1: Control variables

Manufacturing –1.24 (–0.36) –2.57 (–0.77) –2.34 (–0.71)Transportation/utilities –3.10 (–0.78) –4.32 (–1.13) –3.90 (–1.04)Retail –2.53 (–0.51) –7.15 (–1.47) –5.25 (–1.07)Services –4.16 (–1.11) –4.85 (–1.34) –4.87 (–1.37)Banking/real estate/insurance –4.94 (–1.14) –4.48 (–1.08) –2.94 (–0.70)Firm size –0.56 (–0.38) 0.76 (0.50) 0.79 (0.53)Firm age 0.05 (1.28) 0.05 (1.33) 0.04 (1.08)Union presence –1.78 (–0.86) –3.60† (–1.76) –3.40† (–1.70)Grievance procedure 4.85* (2.50) 6.90*** (3.50) 7.26*** (3.73)Contingent workforce size –0.01 (–0.99) –0.03† (–1.98) –0.02† (–1.86)Step 2: Independent variables

ESCLS –0.55 (–0.63) –0.43 (–0.50)LCCLS 2.32* (2.39) 2.21* (2.33)HIHRS –2.35* (–2.51) –2.37** (–2.60)Step 3: Two–way multiplicative

interactions

ESCLS*HIHRS –1.81* (–2.07)LCCLS*HIHRS 1.38† (1.69) � R ² .11** .06* Model R ² .14 .25 .31 Model F 1.24 1.96* 2.21*

n = 90. † p < .10, two-tailed test; * p < .05, two-tailed test; ** p < .01, two-tailed test.

CONTINGENT WORKERS’ IMPACT ON STANDARD EMPLOYEE WITHDRAWAL BEHAVIORS 125

Human Resource Management DOI: 10.1002/hrm

ing may have a negative impact on standard employee attitudes and behaviors (e.g., Con-nelly & Gallagher, 2004; Davis-Blake et al., 2003). To date, research examining the influ-ence of workforce mixing on standard em-ployees has typically focused on the presence of contingent workers in the workplace. In this study, we argued that the impact of workforce mixing on standard employees de-pends on the firm’s strategic rationale for using contingent workers. Specifically, we ar-gued that among firms with a mixed work-force, standard employee employment stabil-ity enhancement as a contingent labor strategy (ESCLS) and labor cost reduction as a contin-gent labor strategy (LCCLS) convey different managerial values to standard employees and have different effects on standard employee withdrawal behaviors.

This study’s findings indicate that among firms with a mixed workforce, an association exists between the strategic rationale for using contingent workers and firm-level standard employee withdrawal behaviors (absenteeism and turnover). Although this study focused on firm-level standard em-ployee absenteeism and turnover—or what Dyer and Reeves (1995) and others (e.g., Way & Johnson, 2005) have referred to as HR out-comes—previous macro (strategic) HR man-agement research has indicated that these outcomes are antecedent to organizational and financial/accounting outcomes and an important determinant of firm performance (e.g., Dyer & Reeves, 1995; Rogers & Wright, 1998; Way & Johnson, 2005). We believe, therefore, that this study extends existing research and provides some insight into the

FIGURE 1. Employment Stability Contingent Labor Strategy

-12.72

-4.44

3.84

Stan

dard

Em

ploy

ee W

orkf

orce

Wit

hdra

wal

B

ehav

ior

- 1 s.d +1 s.d.

Employment Stability Contingent Labor Strategy

-1 s.d. HIHRS

+1 s.d. HIHRS

FIGURE 2. Labor Cost Contingent Labor Strategy

-12.72

-4.44

3.84

Stan

dard

Em

ploy

ee W

orkf

orce

With

draw

alB

ehav

ior

-1 s.d +1 s.d.

Labor Cost Contingent Labor Strategy

-1 s.d.HIHRS

+1 s.d.HIHRS

126 HUMAN RESOURCE MANAGEMENT, JANUARY/FEBRUARY 2010

Human Resource Management DOI: 10.1002/hrm

link between workforce mixing and firm performance.

Although the hypothesized negative rela-tionship between ESCLS and firm-level standard employee withdrawal behaviors was not supported (Hypothesis 1), LCCLS was positively related with firm-level standard employee withdrawal behaviors as we hy-pothesized (Hypothesis 2). We believe that these findings provide support for our argu-ment that among firms with a mixed work-force, the effects of workforce mixing on standard employees and standard employee withdrawal behaviors depend on the strategic rationale underlying contingent labor use.

Results provide support for a direct, nega-tive association between using HIHRS and firm-level standard employee withdrawal be-haviors. This finding is consistent with the growing body of research that indicates that HIHRS are beneficial to firms (e.g., Combs et al., 2006; Evans & Davis, 2005). Moreover,

as we hypothesized (Hypothesis 3), among firms with a mixed workforce, using HIHRS negatively moderated the relationship be-tween ESCLS and standard employee with-drawal behaviors. That is, as we expected, the interaction between ESCLS and HIHRS (i.e., the ESCLS × HIHRS interaction term) was negatively related to firm-level standard em-ployee withdrawal behaviors (see Table IV, Model 3, and Figure 1). When HIHRS use was low (–1 s.d. HIHRS), however, a positive rela-tionship emerged between ESCLS and firm-level standard employee withdrawal behav-iors (see Figure 1). In addition, as we hypothesized (Hypothesis 4), among firms with a mixed workforce, the interaction be-tween LCCLS and HIHRS (i.e., the LCCLS × HIHRS interaction term) was positively re-lated to firm-level standard employee with-drawal behaviors (see Table IV, Model 3, and Figure 2). As we expected, these findings indicate that using HIHRS is an important

T A B L E V Supplemental Moderated Regression Results: Predicting Firm–Level Standard Employee

Turnover and Absenteeism

Standard Employee

Turnover Absenteeism

Model 1 Model 2

Variables B t-score B t-score

Manufacturing 0.53 (0.27) –2.87 (–1.56)Transportation/utilities 0.57 (0.25) –4.47* (–2.15)Retail 0.61 (0.21) –5.86* (–2.14)Services –0.58 (–0.27) –4.29* (–2.17)Banking/real estate/insurance 0.58 (0.23) –3.51 (–1.52)Firm size –0.01 (–0.01) 0.80 (0.97)Firm age 0.01 (0.44) 0.03 (–1.48)Union presence –2.04† (–1.69) –1.36 (–1.22)Grievance procedure 4.03** (3.45) 3.23** (2.98)Contingent workforce size –0.01 (–1.27) –0.01† (–1.97)ESCLS –0.21 (–0.42) –0.21 (–0.45)LCCLS 1.21* (2.12) 1.00† (1.90)HIHRS –1.52** (–2.77) –0.85† (–1.68)ESCLS*HIHRS –1.17* (–2.22) –0.64 (–1.32)LCCLS*HIHRS 0.54 (1.09) 0.85† (1.86) Model R ² .29 .28 Model F 1.99* 1.94*

n = 90. † p < .10, two-tailed test; * p < .05, two-tailed test; ** p < .01, two-tailed test.

CONTINGENT WORKERS’ IMPACT ON STANDARD EMPLOYEE WITHDRAWAL BEHAVIORS 127

Human Resource Management DOI: 10.1002/hrm

contextual factor that influences how stan-dard employees interpret and respond to workforce mixing.

Managerial Implications

From a practical perspective, this research sug-gests that firms should seriously question using contingent workers simply to cut costs and should analyze the trade-off in saved labor costs with the additional costs associ-ated with standard employee absenteeism and turnover. It may be that communicating the reasons for using contingent workers and assurances that even though cost saving is the purpose, such savings allow the continued employment of standard employees, would ameliorate standard employee concerns and lessen standard employee withdrawal behav-iors. It seems particularly important that firms using contingent workers for standard em-ployee employment stability purposes de-velop communication programs that stress that workforce mixing is not being pursued as a labor cost-saving strategy.

The practical implications of our findings concerning the relationship between HIHRS and ESCLS are that consistent messages to standard employees are important. Using HIHRS appears to be a factor that makes the ESCLS basis for using contingent workers believable to standard employees. Certainly the messages sent by using both HIHRS and ESCLS are consistent. Similarly, the inconsis-tent messages of using ESCLS while not using HIHRS appear to make the ESCLS less believ-able to standard employees. These finding suggest that ESCLS is a viable strategy for HIHRS firms, and that firms who would like to use ESCLS should consider using HIHRS as well.

Our findings regarding the LCCLS-HIHRS interaction reinforce the link between using contingent workers and using HIHRS. The inconsistency of LCCLS in an HIHRS setting only worsens the impact of the LCCLS. Firms would be well advised to avoid this inconsis-tency. Overall, our research suggests that firms can successfully mix standard employ-ees and contingent workers when the purpose of doing so is to enhance standard

employee employment stability, and particu-larly when they combine this workforce mix-ing strategy with the use of high investment HR systems.

Limitations and Future Research

The findings this study reports suggest a need for some caution in using contingent work-ers. Specifically, the results show that the impact of workforce mixing on standard em-ployees and important firm-level standard employee behaviors/outcomes may depend upon the strategic rationale for using contin-gent labor and the HR system in place. We would encourage future research to build upon our study’s results and examine the im-pact of other potential contextual factors such as leader member exchange and organi-zational climate, which may influence how standard employees respond to workforce mixing. We also encourage future research to examine the influence of workforce mixing on additional firm-level outcomes. Examin-ing how workforce mixing affects more distal measures of firm performance such as organi-zational outcomes (e.g., productivity and quality), market outcomes (e.g., customer satisfaction, sales growth, and growth in mar-ket share; see Becker & Homburg, 1999), and financial/accounting outcomes (e.g., return of investment and profitability) would be particularly beneficial.

As with all studies, the results of this re-search must be interpreted in light of its limitations. First, an underlying assumption of this study is that standard employees’ in-ferences regarding their firm’s rationale for using contingent labor are congruent with the firm’s actual rationale for using contin-gent labor. It is possible that firms may influ-ence these perceptions by communicating to standard employees the strategic rationale for using contingent labor, even if the informa-tion communicated to standard employees regarding the firm’s strategic rationale for using contingent labor is not completely accurate. Moreover, it is possible that the presence or absence of HIHRS influences the employee perceptions regarding their firm’s strategic rationale for using contingent labor.

128 HUMAN RESOURCE MANAGEMENT, JANUARY/FEBRUARY 2010

Human Resource Management DOI: 10.1002/hrm

Appendix B), the two items we used to create our ESCLS and LCCLS variables (see Table II), and the six HR practice parameters that we used to create our HIHRS index (see Tables I and II) were entered into a principal compo-nents factor analysis and the results of the unrotated solution were examined. Four fac-tors with an eigenvalue greater than 1 emerged from this analysis and no single fac-tor accounted for the majority of the vari-ance. Podsakoff and Organ (1986, p. 536) stated that if a substantial amount of com-mon method variance were present, either a single factor would emerge or one general factor would explain a majority of the cova-riance in the independent and criterion vari-ables. Although the results of the Harman one-factor test indicate that common method variance is not a problem in our study, we would encourage additional research that draws from multiple respondents within each firm to bolster the reliability of these measures. We also encourage additional re-search that examines these relationships in different contexts (such as countries) to as-sess the generalizability of our study’s find-ings further.

In conclusion, firms continue to rely on workforce mixing as a way to increase their flexibility in deploying human capital. And, while workforce mixing may facilitate the firm’s ability to respond to fluctuations in environmental demand (i.e., foster firm flex-ibility), the results of this study indicate that among firms with a mixed workforce the stra-tegic rationale for using contingent labor is a critical factor when examining the effects of workforce mixing on firm-level standard em-ployee withdrawal behaviors.

Notes

1. Contingent labor “consists of independent contrac-

tors; individuals brought in through employment

agencies; on call or day labor; and workers on site

whose services are provided by contract firms”

(Matusik & Hill, 1998, p. 680).

2. Standard employees work at the employer’s place

of business on a full-time basis, under the employ-

er’s supervision, and with the mutual expectation

that employment will continue indefinitely (Con-

Using HIHRS in conjunction with using con-tingent labor to reduce labor costs may send a mixed message, for example. While we cer-tainly encourage future research that explic-itly examines the attributions that standard employees make regarding their firm’s deci-sions to use contingent workers, the results of this study provide support for the assertion that among firms with a mixed workforce, a firm’s strategic rationale for using contingent labor is a critical factor associated with im-portant firm-level standard employee with-drawal behaviors.

Common method bias may be a poten-tial limitation of this study. Gerhart and

colleagues (2000) presented re-sults that call into question the reliability and validity of HR re-search that obtains information from a single source. These schol-ars state, however, that their results are based on the data col-lected from large firms (mean = 40,000 employees) where HR systems and practices may vary considerably within the firm (Gerhart et al., 2000). Both Ger-hart et al. (2000) and Huselid and Becker (2000) conclude that in smaller firms, within-firm variation in HR systems and prac-tices is likely to be smaller. Firms in our sample employed an aver-age of 346.48 employees (see Ap-pendix A); thus, it is unlikely that within-firm variation is a

problem. Furthermore, we focused on a single group of employees (standard em-ployees), which Wilk and Cappelli (2003) suggest may improve reliability.

In this study, dependent and indepen-dent measures were derived from informa-tion obtained from a single respondent (each firm’s most knowledgeable person in HR). To address the common method variance issue empirically, we used the Harman one-factor test (Podsakoff & Organ, 1986). Specifically, the two parameters that we used to create our standard employee withdrawal behaviors measure (the standard employee absentee-ism and turnover measures; see

An underlying

assumption of

this study is that

standard employees’

inferences regarding

their firm’s rationale

for using contingent

labor are congruent

with the firm’s actual

rationale for using

contingent labor.

CONTINGENT WORKERS’ IMPACT ON STANDARD EMPLOYEE WITHDRAWAL BEHAVIORS 129

Human Resource Management DOI: 10.1002/hrm

nelly & Gallagher, 2004; Kalleberg, 2000; Kalleberg,

Reskin, & Hudson, 2000).

3. Within the existing strategic (macro) HR manage-

ment literature, high investment HR systems (Lepak,

Taylor, Tekleab, Marrone, & Cohen, 2007) have also

been referred to as high-performance (Huselid,

1995), high-involvement (Benson, Young, & Lawler,

2006), or high-commitment (Arthur, 1994) HR sys-

tems. Although these systems may have conceptual

distinctions, scholars have typically used these

terms interchangeably in the strategic (macro) HR

management literature (Takeuchi, Lepak, Wang, &

Takeuchi, 2007; Wood, de Menezes, & Lasaosa,

2003). In this article, we adopt the term high invest-

ment HR systems (HIHRS) to denote a system of HR

practices that reflect investments in standard em-

ployee development and long-term mutually benefi-

cial and cooperative standard employee-employer

relationships that together are expected to have a

negative impact on standard employee withdrawal

behaviors.

4 . That is, we posit that with all else being equal, pur-

suing each contingent labor strategy (ESCLS; LCCLS)

will have a distinct relationship with firm-level stan-

dard employee withdrawal behaviors. In this article,

we present empirical evidence that indicates that

firms that use one of these contingent labor strate-

gies tend not to use the other (see Appendix C).

Conceptually, however, using both strategies simul-

taneously could occur (especially in different areas

of the firm). As we have defined them, however,

each strategy is designed to achieve a different set

of outcomes, and standard employees could be ex-

pected to recognize the different aims of each strat-

egy. If nothing else, a firm pursuing LCCLS may well

be eliminating standard employees and replacing

them with contingent workers, while a firm pursu-

ing ESCLS might maintain standard employee em-

ployment levels.

5. Conducted at the individual level and the job level

of analysis.

6. Dun & Bradstreet (D&B) Canada randomly selected

these firms from D&B Canada’s marketing data-

base.

7. Prior to asking respondents questions regarding

contingent labor use, respondents were told,

“Contingent workers refer to independent contrac-

tors, individuals brought in through employment

agencies, on-call or day labor, and workers on site

whose services are provided by contract firms”

(Matusik & Hill, 1998, p. 680).

8. We thank James Guthrie and an anonymous re-

viewer for suggesting that our analyses should

only be conducted within our sample of 90 firms

that had used contingent workers in the last two

years (i.e., firms with a mixed workforce).

9. We would like to thank an anonymous reviewer for

drawing our attention to this problem.

10. Refer to Becker and Huselid (1998), Delery (1998),

and Delery and Shaw (2001) for comprehensive

discussions regarding using an additive approach

to create a unitary index in HR systems research.

11. We expect that the HR system a firm has in place

for its standard employees is an important contex-

tual factor that influences how standard employ-

ees interpret and respond to workforce mixing.

Although the effects of the individual practices in-

cluded in the system may vary, it is the impact of

the system as whole that is of interest in this

study.

12 . As suggested by an anonymous reviewer, we used

this variable (contingent workforce size) to control

for the number of contingent workers used by

each firm.

13. These results are available from the first author

upon request.

Acknowledgments

The authors wish to thank James P. Guthrie, Theresa M. Welbourne, and three anonymous re-viewers for their insightful feedback and suggestions on earlier versions of this article. This research was supported by a grant awarded to the fi rst author by R. P. Scherer Inc., Canada. The interpretations, con-clusions, and recommendations, however, are those of the authors and do not necessarily represent those of R. P. Scherer Inc., Canada.

130 HUMAN RESOURCE MANAGEMENT, JANUARY/FEBRUARY 2010

Human Resource Management DOI: 10.1002/hrm

SEAN A. WAY is an assistant professor of human resource management at Cornell University’s School of Hotel Administration. He earned a Ph.D. from the School of Man-agement and Labor Relations at Rutgers University. His dissertation, titled “A fi rm-level analysis of HR fl exibility,” received the 2006 Ralph Alexander Best Dissertation Award by the Academy of Management HR division in the fi eld of human resource management. His current research focuses on a number of strategic human resource management topics, including the effects of HRM systems, HR fl exibility, workforce mixing, climate, and market orientation on the performance and effectiveness of organizations and their employees.

DAVID P. LEPAK is professor and department chair of the Human Resource Management Department in the School of Management and Labor Relations at Rutgers University. He earned his Ph.D. in management from the Pennsylvania State University. His current research focuses on strategic human resource management with an emphasis on em-ployment subsystems and the HR architecture, contingent labor, intellectual capital, and linking HRM systems to important company outcomes.

CHARLES H. FAY is professor of human resource management in the School of Manage-ment and Labor Relations at Rutgers University, where he teaches managing rewards systems and economics and demographics of labor markets. He completed his Ph.D. in HRM and organization theory at the University of Washington. He has testifi ed before congressional committees on performance management and rewards several times. His research currently focuses on incentive rewards, public-sector performance manage-ment, and the valuation of human capital. He is currently an associate editor with Human Resource Management.

JAMES W. THACKER is professor emeritus, Odette School of Business, at the University of Windsor (Canada). His research has appeared in both academic (Journal of Applied Psychology, Academy of Management) and practitioner (Journal of Managerial Psychol-ogy, HR Professional) journals. He is the coauthor of Effective Training: Systems Strat-egies and Practices (4th ed.). While at the University of Windsor, he was head of the Human Resource area and the director of the MBA program. He is also involved in cor-porate training and consulting; some of his clients are Honda Canada, Ford, HJ Heinz, Hiram Walker, and Revenue Canada.

References

Allen, D. G., Shore, L. M., & Griffeth, R. W. (2003). The role of perceived organizational support and supportive human resource practices in the turnover process. Journal of Management, 29(1), 99–118.

Arthur, J. B. (1994). Effects of human resource systems on manufacturing performance and turnover. Academy of Management, 37(3), 670–687.

Atkinson, A. (1984). Manpower strategies for fl ex-ible organizations. Personnel Management, 16(8), 28–31.

Bae, J., & Lawler, J. J. (2000). Organizational and HRN strategies in Korea: Impact on organizational

performance in an emerging economy. Academy of Management Journal, 43(3), 502–517.

Baron, R. M., & Kenny, D. A. (1986). The moderator-mediator variable distinction in social psychologi-cal research: Conceptual, strategic, and statistical considerations. Journal of Personality and Social Psychology, 51(6), 1173–1182.

Batt, R. (2002). Managing customer services: Human resource practices, quit rates, and sales growth. Academy of Management Journal, 45(3), 587–597.

Becker, B. E., & Huselid, M. A. (1998). High perform-ance work systems and fi rm performance: A synthesis of research and managerial implications. In G. R. Ferris (Ed.), Research in personnel and hu-man resources management (Vol. 16, pp. 53–101). Stamford, CT: JAI Press.

CONTINGENT WORKERS’ IMPACT ON STANDARD EMPLOYEE WITHDRAWAL BEHAVIORS 131

Human Resource Management DOI: 10.1002/hrm

Becker, J., & Homburg, C. (1999). Market-oriented management: A systems-based perspective. Journal of Market-Focused Management, 4(1), 17–41.

Benson, G. S., Young, S. M., & Lawler, E. E., III. (2006). High-involvement work practices and analysts’ forecasts of corporate earnings. Human Resource Management, 45(4), 519–537.

Bishop, J. W., Goldsby, M. G., & Neck, C. P. (2002). Who goes? Who cares? Who stays? Who wants to? The role of contingent workers and corporate layoff practices. Journal of Managerial Psychology, 17(4), 298–315.

Blau, P. (1964). Exchange and power in social life. New York: Wiley.

Cappelli, P., & Neumark, D. (2004). External churning and internal fl exibility: Evidence on the functional fl exibility and core-periphery hypotheses. Industrial Relations, 43(1), 148–182.

Cardon, M. S. (2003). Contingent labor as an enabler of entrepreneurial growth. Human Resource Management, 42(4), 357–373.

Chattopadhyay, P., & George, E. (2001). Examining work externalization through the lens of social identity theory. Journal of Applied Psychology, 86(4), 781–788.

Cohen, J., & Cohen, P. (1983). Applied multiple regres-sion/correlation analysis in behavioral sciences (2nd ed.). Hillsdale, NJ: Erlbaum.

Combs, J., Liu, Y., Hall, A., & Ketchen, D. (2006). How much do high-performance work practices matter? A meta-analysis of their effects on organi-zational performance. Personnel Psychology, 59(3), 501–528.

Connelly, C. E., & Gallagher, D. G. (2004). Emerging trends in contingent work research. Journal of Management, 30(6), 959–983.

Cooke, W. N. (1994). Employee participation programs, group-based incentives and company perform-ance: A union-nonunion comparison. Industrial and Labor Relations Review, 47(4), 594–609.

Covin, J., & Slevin, D. (1989). Strategic management of small fi rms in hostile and benign environments. Strategic Management Journal, 10(1), 75–87.

Datta, D. K., Guthrie, J. P., & Wright, P. M. (2005). Hu-man resource management and labor productivity: Does industry matter? Academy of Management Journal, 48(1), 135–145.

David, J. (2005). The unexpected employee and organizational costs of skilled contingent workers. Human Resource Planning, 28(2), 32–40.

Davis-Blake, A., Broschak, J. P., & George, E. (2003). Happy together? How using nonstandard workers affects exit, voice, and loyalty among standard employees. Academy of Management Journal, 46(4), 475–485.

Davis-Blake, A., & Uzzi, B. (1993). Determinants of non-standard work: A study of temporary workers and independent contractors. Administrative Science Quarterly, 38(2), 195–223.

Delaney, J. T., & Huselid, M. A. (1996). The impact of human resource management practices on percep-tions of organizational performance. Academy of Management Journal, 39(4), 949–969.

Delery, J. E. (1998). Issues of fi t in strategic human resource management: Implications for research. Human Resource Management Review, 8(3), 289–309.

Delery, J. E., & Doty, D. H. (1996). Modes of theorizing in strategic human resource management: Tests of universalistic, contingency, and confi gurational performance predictions. Academy of Management Journal, 39(4), 802–835.

Delery, J. E., & Shaw, J. D. (2001). The strategic management of people in work organizations: Review, synthesis, and extension. In G. R. Ferris (Ed.), Research in personnel and human resources management (Vol. 20, pp. 165–197). Stamford, CT: JAI Press.

Doeringer, P., & Piore, M. (1971). Internal labor mar-kets and manpower analysis. Lexington, MA: Heath Lexington Books.

Dyer, L., & Reeves, T. (1995). Human resource strate-gies and fi rm performance: What do we know and where do we need to go? International Journal of Human Resource Management, 6(3), 656–670.

Edwards, J. R. (2001). Multidimensional constructs in organizational behavior research: An integra-tive analytical framework. Organizational Research Methods, 4(2), 144–192.

Eisenberger, R., Armeli, S., Rexwinkel, B., Lynch, P. D., & Rhodes, L. (2001). Reciprocation of perceived organizational support. Journal of Applied Psychol-ogy, 86(1), 42–51.

Eisenberger, R., Huntington, R., Hutchison, S., & Sowa, D. (1986). Perceived organizational support. Journal of Applied Psychology, 71(3), 500–507.

Evans, W. R., & Davis, W. D. (2005). High-performance work systems and organizational performance: The mediating role of internal social structure. Journal of Management, 31(5), 758–775.

Gerhart, B., Wright, P., & McMahan, G. (2000). Meas-urement error in research on the human resources

132 HUMAN RESOURCE MANAGEMENT, JANUARY/FEBRUARY 2010

Human Resource Management DOI: 10.1002/hrm

and fi rm performance relationship: Further evi-dence and analysis. Personnel Psychology, 53(4), 855–872.

Gouldner, A. W. (1960). The norm of reciprocity. Ameri-can Sociological Review, 25(2), 161–178.

Govindarajan, V., & Fisher, J. (1990). Strategy, control systems, and resource sharing: Effects on busi-ness-unit performance. Academy of Management Journal, 33(2), 259–285.

Gupta, A., & Govindarajan, V. (1984). Business unit strategy, managerial characteristics, and business unit effectiveness at strategy implementation. Academy of Management Journal, 27(1), 25–41.

Guthrie, J. (2001). High-involvement work practices, turnover, and productivity: Evidence from New Zealand. Academy of Management Journal, 44(1), 180–190.

Guthrie, J. P., Spell, C. S., & Nyamori, R. O. (2002). Correlates and consequences of high involvement work practices: The role of competitive strategy. International Journal of Human Resource Manage-ment, 13(1), 183–197.

Harrison, B., & Kelley, M. (1993). Outsourcing and the search for fl exibility. Work, Employment and Society, 7(2), 213–235.

Hipple, S., & Stewart, J. (1996). Earnings and ben-efi ts of workers in alternative work arrangements. Monthly Labor Review, 119(10), 46–54.

Houseman, S. (2001). Why employers use fl exible staffi ng arrangements: Evidence from an estab-lishment survey. Industrial and Labor Relations Review, 55(1), 149–170.

Houseman, S., & Polivka, A. E. (2000). The implications of fl exible staffi ng arrangements for job security. In D. Neumark (Ed.), On the job: Is long-term employ-ment a thing of the past? (pp. 427–462). New York: Russell Sage Foundation.

Huselid, M. A. (1995). The impact of human resource management practices on turnover, productivity, and corporate fi nancial performance. Academy of Management Journal, 38(3), 635–672.

Huselid, M. A., & Becker, B. E. (2000). Comment on “Measurement error in research on human resources and fi rm performance: How much error is there and how does it infl uence effect size estimates?” by Gerhart, Wright, McMahan, and Snell. Personnel Psychology, 53(4), 835–854.

Kalleberg, A. L. (2000). Nonstandard employment relations: Part-time, temporary and contract work. Annual Review of Sociology, 26(1), 341–365.

Kalleberg, A. L., Reskin, B. F., & Hudson, K. (2000). Bad jobs in America: Standard and nonstandard employ-

ment relations and job quality in the United States. American Sociological Review, 65(2), 256–278.

Lepak, D. P., & Snell, S. A. (1999). The human resource architecture: Toward a theory of human capital allo-cation and development. Academy of Management Review, 24(1), 31–48.

Lepak, D. P., & Snell, S. A. (2002). Examining the human resource architecture: The relationships among human capital, employment, and human resource confi gurations. Journal of Management, 28(4), 517–543.

Lepak, D. P., Takeuchi, R., & Snell, S. A. (2003). Employ-ment fl exibility and fi rm performance: Examining the interactive effects of employment mode, envi-ronmental dynamism and technological intensity. Journal of Management, 29(5), 681–703.

Lepak, D. P., Taylor, M. S., Tekleab, A., Marrone, J. A., & Cohen, D. J. (2007). An examination of the use of high-investment human resource systems for core and support employees. Human Resource Manage-ment, 46(2), 223–246.

Loi, R., Hang-yue, N., & Foley, S. (2006). Linking employees’ justice perceptions to organizational commitment and intention to leave: The mediating role of perceived organizational support. Journal of Occupational and Organizational Psychology, 79(1), 101–120.

Matusik, S. F., & Hill, C. W. L. (1998). The utilization of contingent work, knowledge creation, and competi-tive advantage. Academy of Management Review, 23(4), 680–697.

Miles, M. P., Covin, J. G., & Heeley, M. B. (2000). The relationship between environmental dynamism and small fi rm structure, strategy, and perform-ance. Journal of Marketing Theory and Practice, 8(2), 63–74.

Pearce, J. L. (1993). Toward an organizational behavior of contract laborers: Their psychological involve-ment and effects on employee co-workers. Acad-emy of Management Journal, 36(5), 1082–1096.

Pfeffer, J. (1998). The human equation: Building profi ts by putting people fi rst. Boston: Harvard Business School Press.

Podsakoff, P. M., & Organ, D. W. (1986). Self-reports in organizational research: Problems and prospects. Journal of Management, 12(4), 531–544.

Polivka, A. E., & Nardone, T. (1989). The quality of jobs: On the defi nition of “contingent work.” Monthly Labor Review, 112(12), 9–16.

Rhoades, L., & Eisenberger, R. (2002). Perceived organizational support: A review of the literature. Journal of Applied Psychology, 87(4), 698–714.

CONTINGENT WORKERS’ IMPACT ON STANDARD EMPLOYEE WITHDRAWAL BEHAVIORS 133

Human Resource Management DOI: 10.1002/hrm

Robinson, R. B., Jr., & Pearce, J. A. (1988). Product life-cycle considerations and the nature of strategic activities in entrepreneurial fi rms. Journal of Business Venturing, 1(2), 207–224.

Rogers, E., & Wright, P. M. (1998). Measuring organi-zational performance in strategic human resource management: Problems, prospects, and perform-ance information markets. Human Resource Management Review, 8(3), 311–331.

Rousseau, D. M. (1995). Psychological contracts in or-ganizations: Understanding written and unwritten agreements. Thousand Oaks, CA: Sage.

Shaw, J., Delery, J., Jenkins, C., & Gupta, N. (1998). An organizational-level analysis of voluntary and involuntary turnover. Academy of Management Journal, 41(5), 511–525.

Shaw, J. D., Gupta, N., & Delery, J. E. (2005). Alterna-tive conceptualizations of the relationship between voluntary turnover and organizational performance. Academy of Management Journal, 48(1), 50–68.

Takeuchi, R., Chen, G., & Lepak, D. P. (2009). Through the looking glass of a social system: Cross-level ef-fects of high performance work systems on employ-ees’ attitudes. Personnel Psychology, 62(1), 1–29.

Takeuchi, R., Lepak, D. P., Wang, H., & Takeuchi, K. (2007). An empirical examination of the mecha-nisms mediating between high-performance work systems and the performance of Japanese or-ganizations. Journal of Applied Psychology, 92(4), 1069–1083.

Tsui, A. S., Pearce, J. L., Porter, L. W., & Tripoli, A. M. (1997). Alternative approaches to the employee-organization relationship: Does investment in employees pay off? Academy of Management Journal, 40(5), 1089–1121.

U.S. Bureau of Labor Statistics. (2005). Contingent and alternative employment arrangements, February 2005. Washington, DC: U.S. Department of Labor News Release No. USDL 05-1433.

U.S. Government Accountability Offi ce. (2006). Em-ployment arrangements: Improved outreach could help ensure proper worker classifi cation, July 2006. Washington, DC: U.S. Government Accountabil-ity Offi ce Report No. GAO-06-656 (United States Government Accountability Offi ce Report to the Ranking Minority Member, Committee on Health, Education, Labor and Pensions, U.S. Senate).

Venkatraman, N., & Ramanujam, V. (1986). Measure-ment of business performance in strategy research: A comparison of approaches. Academy of Manage-ment Review, 11(4), 801–814.

Vosko, L. F., Zukewich, N., & Cranford, C. (2003). Precarious jobs: A new typology of employment. Perspectives on Labour and Income, 4(10), 16–26.

Way, S. A. (2002). High performance work systems and intermediate indicators of fi rm performance within the U.S. small business sector. Journal of Management, 28(6), 765–785.

Way, S. A., & Johnson, D. E. (2005). Theorizing about the impact of strategic human resource manage-ment. Human Resource Management Review, 15(1), 1–19.

Wayne, S. J., Shore, L., & Liden, R. C. (1997). Perceived organizational support and leader-member exchange: A social exchange perspective. Academy of Management Journal, 40(1), 82–111.

Whitener, E. M. (2001). Do “high commitment” human resource practices affect employee commitment? A cross-level analysis using hierarchical linear modeling. Journal of Management, 27(5), 515–535.

Wilk, S. L., & Cappelli, P. (2003). Understanding the de-terminants of employer use of selection methods. Personnel Psychology, 56(1), 103–124.

Wood, S. J., de Menezes, L. M., & Lasaosa, A. (2003). Family-friendly management in Great Britain: Test-ing various perspectives. Industrial Relations, 42(2), 221–250.

Wright, P. M., Gardner, T. M., & Moynihan, L. M. (2003). The impact of HR practices on the performance of business units. Human Resource Management Journal, 13(3), 21–36.

Wright, P. M., & Snell, S. A. (1998). Toward a unify-ing framework for exploring fi t and fl exibility in strategic human resource management. Academy of Management Review, 23(4), 756–772.

Wu, C., & Yao, G. (2006). Do we need to weight item satisfaction by item importance? A perspective from Locke’s range-of-affect hypothesis. Social Indicators Research, 79(3), 485–502.

Youndt, M. A., Snell, S. A., Dean, J., Jr., & Lepak, D. P. (1996). Human resource management, manufactur-ing strategy, and fi rm performance. Academy of Management Journal, 39(4), 836–866.

Zacharatos, A., Barling, J., & Iverson, R. D. (2005). High-performance work systems and occupational safety. Journal of Applied Psychology, 90(1), 77–93.

Zahra, S. A. (1993). New product innovation in estab-lished companies: Associations with industry and strategy variables. Entrepreneurship Theory and Practice, 18(2), 47–69.

134 HUMAN RESOURCE MANAGEMENT, JANUARY/FEBRUARY 2010

Human Resource Management DOI: 10.1002/hrm

A P P E N D I X A Descriptive Statistics: Complete Data, Mixed Workforce, and Excluded Cases Samplesª

Complete Data

Sample Mean

Mixed Workforce

Sample

Excluded

Cases Sample

Variables Mean s.d. Mean s.d. Mean s.d.

Manufacturing 0.40 0.49 0.42 0.50 0.33 0.48

Transportation/utilities 0.13 0.34 0.13 0.34 0.13 0.34

Retail 0.05 0.21 0.06 0.23 0.03 0.16

Services 0.24 0.43 0.21 0.41 0.31 0.47

Banking/real estate/fi nance 0.11 0.31 0.10 0.30 0.13 0.34

Wholesale 0.08 0.27 0.08 0.27 0.08 0.27

Firm sizeb 343.12 379.00 346.48 373.56 335.37 396.15

Firm age 23.59 23.60 24.85 25.29 20.68 19.09

Union 0.33 0.47 0.32 0.47 0.36 0.49

Grievance 0.67 0.47 0.71 0.46 0.59 0.50

HIHRS indexc 15.94 3.16 16.07 3.00 15.64 3.52

Selective hiring 2.94 1.57 2.94 1.55 2.92 1.64

Above-market compensationd 3.24 0.68 3.32 0.68 3.05 0.63

Continuous training 2.80 0.89 2.79 0.93 2.82 0.80

Frequent experience-enhancing opportunities 1.92 0.89 1.94 0.88 1.88 0.92

Participation/involvement in decision making 2.52 0.64 2.53 0.66 2.50 0.60

Performance-based pay 2.52 1.25 2.54 1.28 2.46 1.21

Standard employee workforce withdrawal behavior –4.31 8.13 –4.44 8.28 –4.00 7.85

Standard employee workforce turnover –2.29 4.75 –2.30 4.90 –2.26 4.45

Standard employee workforce absenteeism –2.02 4.55 –2.14 4.51 –1.74 4.69a Our entire (complete data) sample consists of all fi rms from which we received complete data (n = 129). Our mixed workforce sample (n = 90) consists of fi rms with a mixed workforce (i.e., that reported they had used contingent workers in the last two years). Our ex-cluded cases sample (n = 39) consists of fi rms that reported they had not used contingent workers in the last two years.

b Raw fi rm size scores (versus the natural logarithm of the number of employees employed—see Table III) are reported in Appendix A.c Raw HIHRS index scores (versus z-scores—see Tables II and III) are reported in Appendix A.d An analysis of fi rms that had and had not used contingent labor in the last two years (mixed workforce sample versus excluded cases

sample) on differences based on all of the variables listed showed that the two groups were only statistically different on one dimen-sion: above-market compensation (i.e., fi rms from our mixed workforce sample made greater use of the HR practice).

CONTINGENT WORKERS’ IMPACT ON STANDARD EMPLOYEE WITHDRAWAL BEHAVIORS 135

Human Resource Management DOI: 10.1002/hrm

Traditional Procedure

Standard employee workforce withdrawal behaviors were measured with a modified version of an instrument that has been used in the strategy, marketing, and entrepreneurship literatures. For each firm, researchers using this instrument have traditionally adopted the procedure de-scribed below to create a weighted performance score for each item (see Covin & Slevin, 1989; Govindarajan & Fisher, 1990; Gupta & Govindarajan, 1984; Miles et al., 2000; Zahra, 1993). For each firm, researchers have typically created weighted performance scores for each item (e.g., standard employee absenteeism and turnover):

Next, for each firm, researchers have typically used these weighted scores to create a single subjective measure of performance. Although we present the traditional procedure above, because of problems stated in the text (see note 9) we adopted a different procedure to create a single firm-level standard employee withdrawal behaviors variable.

Procedure Used in Current Study

For each firm in this study we adopted the following procedure first to create weighted stan-dard employee absenteeism and turnover scores.

Consistent with prior research, for each firm, we create a single firm-level standard em-ployee withdrawal behaviors variable by adding the firm’s weighted standard employee absen-teeism and turnover scores together. Note that the pattern of results was consistent whether the dependent variable was calculated following the traditional procedure or the procedure that we used in this study.

Top HR Management Team’s

Level of Dissatisfaction/

Satisfaction With Standard

Employee (a) Absenteeism

(b) Turnover

Level of Importance That Firm’s Top HR Management Team

Attaches to Standard Employee (a) Absenteeism (b) Turnover

5

Very

Important

4

Quite

Important

3

Moderately

Important

2

Somewhat

Important

1

Of Little

Importance

5, Very satisfi ed 25 20 15 10 5 4, Satisfi ed 20 16 12 8 4 3, Neutralª 15 12 9 6 3 2, Dissatisfi ed 10 8 6 4 2 1, Very dissatisfi ed 5 4 3 2 1

Top HR Management Team’s

Level of Dissatisfaction/

Satisfaction With Standard

Employee (a) Absenteeism

(b) Turnover

Level of Importance That Firm’s Top HR Management Team Attaches

to Standard Employee (a) Absenteeism (b) Turnover

5

Very

Important

4

Quite

Important

3

Moderately

Important

2

Somewhat

Important

1

Of Little

Importance

–2, Very satisfi ed –10 –8 –6 –4 –2–1, Satisfi ed –5 –4 –3 –2 –10, Neutralª 0 0 0 0 01, Dissatisfi ed 5 4 3 2 12, Very dissatisfi ed 10 8 6 4 2

ª That is, neither dissatisfi ed nor satisfi ed (neutral).

A P P E N D I X B Standard Employee Workforce Withdrawal Behaviors

136 HUMAN RESOURCE MANAGEMENT, JANUARY/FEBRUARY 2010

Human Resource Management DOI: 10.1002/hrm

Single-Item ESCLS Measure: Response (Score) Frequencies

Single-Item LCCLS Measure: Response (Score) Frequencies

Thirty-two percent (32.2%) of our final sample’s 90 firms with a mixed workforce reported that they used contingent labor to provide their standard employee workforce with employment stability (ESCLS) more often than they used contingent workers to reduce labor costs (LCCLS). Twenty percent (20.0%) of our final sample reported that they used contingent labor to reduce labor costs (LCCLS) more often than they used contingent labor to provide their standard em-ployee workforce with employment stability (ESCLS).

As we reported in Table II and Table III, the correlation between ECSLS and LCCLS is not sta-tistically significant. If firms whose respondent (the firm’s most knowledgeable HR person) indicated that the firm did not use either ESCLS or LCCLS are filtered from our data set, how-ever, the correlation between ESCLS and LCCLS is negative and statistically significant (r = –.35, p < .01). The results presented in this appendix indicate that firms that use one of these con-tingent labor strategies tend not to use the other.

Response Score Frequency (%)

Always 5 1 (1.1%)Often 4 11 (12.2%)Sometimes 3 11 (12.2%)Rarely 2 34 (37.8%)Never 1 33 (36.7%)

Response Score Frequency (%)

Always 5 1 (1.1%)Often 4 7 (7.8%)Sometimes 3 8 (8.9%)Rarely 2 30 (33.3%)Never 1 44 (48.9%)

A P P E N D I X C ESCLS and LCCLS Frequency Data

CONTINGENT WORKERS’ IMPACT ON STANDARD EMPLOYEE WITHDRAWAL BEHAVIORS 137

Human Resource Management DOI: 10.1002/hrm

We posited that HIHRS included HR practices that would reflect distinct but interrelated invest-ments in standard employee development and long-term mutually beneficial and cooperative standard employee-employer relationships. Together, these are expected to reduce standard employee absenteeism and turnover (standard employee withdrawal behaviors). Thus, each of the six parameters we used to create our HIHRS index (see Table I) is expected to foster lower standard employee withdrawal behaviors. Using our entire (complete data) sample (n = 129; see Appendix A) we examined the Pearson correlations between firm-level standard employee withdrawal behaviors (absenteeism and turnover) and each of the six parameters that we used to construct our HIHRS index:

Pearson Correlations

The findings reported above provide some support for the six parameters we used to construct our HIHRS index. Above-market compensation was positively (but not significantly) correlated with standard employee absenteeism and withdrawal behaviors; therefore, we eliminated this parameter and constructed a revised HIHRS index (composed of the five parameters that foster lower standard employee withdrawal behaviors). Although the effects of the individual prac-tices included in the system may vary, it is the impact of the system as a whole that is of inter-est in this study. Within our sample of firms with a mixed workforce (n = 90; see Appendix A), therefore, we reexamined the relationship between HIHRS and standard employee withdrawal behaviors (see Table IV; Model 2) using the revised HIHRS index. The relationship between standard employee withdrawal behaviors and HIHRS was negative whether our original (� = –2.35, t = –2.51, p < .05) or revised (� = –2.37, t = –2.49, p < .05) HIHRS index was used to mea-sure HIHRS use.a Although our results may indicate that the revised index is a better measure of HIHRS use, the results are exploratory and should be interpreted as such.

a Results were similar when we examined the relationship between standard employee withdrawal behaviors and

the two HIHRS indices (the [original] HIHRS index and the revised HIHRS index) within our complete data sample

(n = 129).

A P P E N D I X D Standard Employee Withdrawal Behaviors and the Six HR Practice Parameters

Firm-Level Standard Employee

Withdrawal

Behaviors

Absenteeism Turnover

Selective staffi ng –.09ns –.14† –.02ns

Above-market compensation .01ns .06ns –.04ns

Continuous training –.24** –.15* –.26**Frequent experience-enhancing opportunities

–.15* –.06ns –.19*

Participation/involvement in decision making

–.14† –.12† –.13†

Performance-based pay –.05ns –.05ns –.04ns

ns not (statistically) signifi cant (p > .10, two-tailed test).† p < .10, two-tailed test; * p < .05, two-tailed test; ** p < .01, two-tailed test.

138 HUMAN RESOURCE MANAGEMENT, JANUARY/FEBRUARY 2010

Human Resource Management DOI: 10.1002/hrm

Overview of Analyses

To demonstrate that this study’s findings were robust and not a function of multicollinearity and to show that the number of contingent workers used (contingent workforce size) did not influence our study’s reported findings, four supplemental moderated regression models were tested. Steps 1 and 2 were unchanged (see Table IV). In Step 3, however, Model A included the following multiplicative interactions: (a) ESCLS × HIHRS (mean = 0.20, s.d. = 0.94); (b) LCCLS × HIHRS (mean = 0.05, s.d. = 1.08); (c) ESCLS × contingent workforce size (mean = –4.95, s.d. = 62.37); (d) LCCLS × contingent workforce size (mean = 17.90, s.d. = 134.81); (e) HIHRS × contingent workforce size (mean = –3.65, s.d. = 44.35); and (f) ESCLS × LCCLS (mean = 0.03, s.d. = 0.83). Model B included the following multiplicative interactions: (a) ESCLS × HIHRS; (b) LCCLS × HIHRS; (c) ESCLS × contingent workforce size; (d) LCCLS × contingent workforce size; and (e) HIHRS × Contingent workforce size. Model C included the following multiplica-tive interactions: (a) ESCLS × HIHRS; (b) LCCLS × HIHRS; (c) ESCLS × contingent workforce size; and (d) LCCLS × contingent workforce size. Model D included the following multiplica-tive interactions: (a) ESCLS × HIHRS; (b) LCCLS × HIHRS; and (c) ESCLS × LCCLS.

Overview of Results

In each of these four models (Models A, B, C, and D), only the ESCLS × HIHRS and LCCLS × HIHRS interaction terms were significant; that is, none of the other multiplicative interaction terms was significant in any of these four models. These results indicate that the findings re-ported in this study are 1) robust and not a function of multicollinearity and 2) not influenced by the number of contingent workers used (contingent workforce size). The results of these analyses are available from the first author upon request.

A P P E N D I X E Supplemental Moderated Regression Analyses: Demonstrating that Study Results are

Robust and not Infl uenced by Contingent Workforce Size