Review of Public Personnel

Administration
Volume 28 Number 2
June 2008 120-143

Explaining Turnover Intention in © 2008 Sage Publications

10.1177/0734371X08315771
http://roppa.sagepub.com
State Government hosted at
http://online.sagepub.com
Examining the Roles of Gender,
Life Cycle, and Loyalty
Donald P. Moynihan
University of Wisconsin–Madison
Noel Landuyt
The University of Texas at Austin

This article tests a model of turnover intention on a large sample of Texas state employ-
ees focusing on four issues. First, the findings support a life cycle stability hypothesis,
which suggests that age, experience, and geographic preferences reduce turnover inten-
tion, an effect compounded by economic/familial constraints for primary wage earners
and members of large households. Second, contrary to previous research, the results show
that females are significantly less likely to state an intention to quit. This finding reflects
changing patterns of labor force participation, as well as the particular advantages that the
public sector offers female employees. Third, the results distinguish between the relative
contributions of three overlapping concepts: organizational loyalty, voice, and empower-
ment. Organizational loyalty and empowerment reduce turnover intention, but voice is
not a significant factor. Finally, the article provides a detailed test of different personnel
policies, providing particular support for diversity policies.

Keywords: turnover; loyalty; gender; human resource management; diversity

W hy do public employees decide to leave their organization? This is a basic ques-

tion for human resource managers and organization theorists. Organizations use
expensive human resource management (HRM) strategies not only in the hope of
recruiting staff, but also of keeping them. The underlying bias of turnover research is
that voluntary turnover is a negative outcome for an organization. When an organiza-
tion loses the human capital it wants to retain, it incurs process separation costs,
recruitment costs, training costs, and lost productivity costs. Research has sought to
understand why employees leave, with the hope of providing employers with an
informed understanding of the strategies they can use to limit turnover.
This article has two particular goals. First, we identify and test a detailed model of
turnover based on previous research. Although there is an extensive empirical litera-
ture on turnover in the private sector, there is surprisingly little in the public sector.

120

Downloaded from rop.sagepub.com at COLUMBIA UNIV on February 4, 2015

 Moynihan, Landuyt / Turnover Intention in State Government 121

Some older exceptions (Huang, Chuang, & Lin, 2003; Ito, 2003; Kellough & Osuna,
1995; Lewis, 1991; Lewis & Park, 1989; Mor Barak, Nissly, & Levin, 2001; Selden &
Moynihan, 2000) have been joined by recent work (Bertelli, 2007; Kim, 2005; Lee &
Whitford, in press; Meier & Hicklin, in press; Moynihan & Pandey, in press), but such
research rarely focuses on state government (Smith, 1979). This article seeks to pro-
vide better empirical understanding of the causes of turnover in the public sector, by
testing some frequently hypothesized predictors on a large sample of employees from
the state of Texas.
The second goal of the article is to advance turnover theory in three particular
ways. We propose what we call the life cycle stability hypothesis, which suggests
that individuals who are older and have considerable experience with an organiza-
tion will be reluctant to change jobs, a reluctance compounded by economic and/or
familial constraints for primary wage earners and members of large households. The
article also proposes that the traditional hypotheses that females are more likely to
plan to quit are not only wrong, but that the reverse is true because of changing pat-
terns of gender labor force participation and the particular advantages that the public
sector can offer to female employees. We also seek to advance the understanding of
turnover by distinguishing between the relative contributions of three sometimes
overlapping concepts: organizational loyalty, voice, and empowerment.

Literature Review

The factors associated with turnover generally fall within three broad categories:
environmental or economic, individual, and organizational (Mobley, Griffeth, Hand,
& Meglino, 1979; Selden & Moynihan, 2000). This article employs a cross-sectional
model for a fixed point in time in a specific state that allows us to examine individ-
ual characteristics and individual perceptions of work characteristics while control-
ling for agency factors, but we do not control for geographical variation or variation
across time for economic factors such as unemployment.
This section develops a detailed list of individual and job-related hypotheses of
turnover. Our approach derives from the assumption of March and Simon (1958) that
employees choose to stay or leave depending on the factors that influence the desir-
ability and ease of movement. The empirical literature that has followed this broad
proposal has produced a long list of claims as to why an employee decides to leave
or stay.

Individual Characteristics
Standard individual characteristics associated with turnover include controls for
gender and minority status; familial constraints on job movement, such as primary
household earner and household size; age; length of time in position and length of

Downloaded from rop.sagepub.com at COLUMBIA UNIV on February 4, 2015

122 Review of Public Personnel Administration

time living in a state; and the greater job opportunities provided by higher levels of
education.

The life cycle stability hypothesis. Individual characteristics are typically consid-
ered separately as controls for other variables and without consideration of the con-
ceptual relationship with one another. However, some individual characteristics (age,
tenure, familial commitments, and geographic preferences) commonly indicate that
an employee is at a point in life where he or she has a strong preference for the sta-
tus quo, including his or her job. Collectively, we characterize this as a life cycle sta-
bility hypothesis: Older and more settled employees with familial obligations are
less likely to quit.
Previous research suggests that older and longer serving employees are less likely
to quit (Cotton & Tuttle, 1986; Iverson & Currivan, 2003; Kellough & Osuna, 1995;
Lazear, 1999; Lewis, 1991; Lewis & Park, 1989; Mor Barak et al., 2001; Muchinsky
& Morrow, 1980). We therefore include a variable that measures age as well as
a measure of length of time employees have worked for the state. In many studies of
turnover, age is often used without controlling for length of service. Although
the two are connected, they point to different theoretical explanations for turnover
intentions, and for this reason it is important to incorporate measures of both age and
tenure. Another indicator of a preference for stability is the length of time an
employee has lived in the state. We hypothesize that the longer an employee remains
within a job or a particular geographical area, the more likely that he or she has com-
mitted to that job or area and the less likely that he or she will quit.
Labor economics offers an additional theoretical explanation for why employees
with greater service continue in a job. Becker (1962) proposed that workers and
firms invest jointly in firm-specific human capital. Both enjoy the benefits of this
relationship, making it relatively costly to end the relationship. The longer an
employee is with a company, the greater the mutual investments and benefits, and
the greater the disincentives to quit. In the initial years of employment, as employ-
ees seek to ascertain whether their job meets their preferences, the limited firm-
specific knowledge they have developed makes it relatively easier for them to switch
to another firm. This human capital perspective therefore provides additional expla-
nation for the reason that time on the job (or its frequent correlate, age) is generally
found to be a strong and negative predictor of turnover. Another reason that employ-
ees with a good deal of service may be reluctant to quit is that governments often
structure significant pension penalties (Ippolito, 1987).
Other individual characteristics show how economic imperatives and familial
constraints reinforce the desire for the status quo. Individuals who are a household’s
primary earner are less likely to exit a stable job and face the risks associated with
seeking a new job. For the same reasons, we expect that employees from larger
households will be less likely to leave the workplace (Blau & Kahn, 1981), although
Mor Barak et al. (2001) suggest evidence to the contrary.

Downloaded from rop.sagepub.com at COLUMBIA UNIV on February 4, 2015

 Moynihan, Landuyt / Turnover Intention in State Government 123

Among the individual-level variables, the role of education has been widely
tested. A consistent finding is that employees who are more educated are more likely
to exit (Cotton & Tuttle, 1986), although Curry, McCarragher, and Dellmann-
Jenkins (2005) found the opposite to be true among social service employees. The
basic logic for these findings is that education provides an externally legitimated
measure of human capital that is accepted in other organizations, and it therefore
increases the ease of movement of the employee.

Changing perspectives on gender and race. Gender and minority status have been
included as standard controls in most studies of turnover. The traditional hypothesis
was that both females and members of minority groups were more likely to quit.
However, this traditional hypothesis came under challenge, particularly with regard
to gender. The most careful investigation of the relationship between gender and
race, and turnover found that factors such as age, education, promotion opportunity,
experience, and salary mediated the effect of gender (Blau & Kahn, 1981; Kellough
& Osuna, 1995; Lewis & Park, 1989; Smith, 1979).
With regard to race, recent studies at the federal level show no relationship between
minority status and turnover intention (Bertelli, 2007; Lee & Whitford, in press),
although such research follows the practice of most turnover scholarship by organizing
racial categories broadly into Whites and minorities. Because 44% of our sample of
Texas state employees fall into categories traditionally defined as minority, we divided
this group further into four separate dummy variables: African American/Black,
Hispanic/Mexican American, Asian American/Pacific Islander/Native American Indian,
and Multiracial/Other.
For gender, the traditional hypothesis has been largely set aside, and treatments
of gender and contemporary research often now assume no relationship with
turnover. However, we argue that both the traditional hypothesis and the contempo-
rary failure to consider gender overlook a number of reasons that females in the
public sector may be less likely to consider quitting than their male counterparts.
Changing patterns of workforce participation, and the particular attraction of the
public sector, provide the basis for proposing a negative relationship between
females and turnover intention.
The traditional hypothesis may have been more accurate once, but it fails to
reflect changing patterns of labor participation. The hypothesis that women were
more likely to leave are primarily based on outdated assumptions of a traditional
family structure with a single breadwinner. Women have increased their participa-
tion in the labor force (from 43% in 1970 to 59% in 2004), increased their earnings
as a percentage of men’s (from 62% in 1970 to 80% in 2004), are more likely to par-
ticipate when they have a child under 18 (47% in 1975 to 71% in 2004), and con-
tributed a larger share of household earnings (26% in 1973 to 35% in 2003, Bureau
of Labor Statistics, 2005). Women are less likely to exit the workforce because of
familial commitments (Stier & Lewin-Epstein, 2001, p. 1731). This pattern holds

Downloaded from rop.sagepub.com at COLUMBIA UNIV on February 4, 2015

124 Review of Public Personnel Administration

true across cross-national studies in large part because of policies supportive of

working mothers (Stier & Lewin-Epstein, 2001), especially family leave policies
(Waldfogel, Higuchi, & Abe, 1999). We find some support for the suggestion that
the traditional hypothesis is simply outdated by the fact that more recent research
rarely finds an impact of gender on turnover even when failing to include many of
the gender-relevant controls whose absence was used to critique older studies (e.g.,
Huang et al., 2003; Iverson & Currivan, 2003; Kim, 2005). Changing patterns of
labor force participation have made it less likely that women will quit.
The relationship between gender and turnover may also be distinct for public sec-
tor employment, which we propose will be more attractive to women. Most of the lit-
erature on gender and turnover come from the private sector. Women have traditionally
been more likely to work in the public sector, a trend that has increased of late
(Bernhardt & Dresser, 2002, p. 4). At the federal level, the percentage of women in the
Senior Executive Service increased from 5.1% in 1979 to 26.7% in 2004 (Stivers,
2002, p. 27; U.S. Office of Personnel Management [OPM], 2004, p. 72), and women
made up 45% of the civilian workforce in 2003, up from 41% in 1985 (U.S. OPM,
1997, p. 10; 2004, p. 10). Among military personnel, patterns of participation have also
been increasing, albeit from a much lower base. In 1990, women made up 10.9% of
enlisted personnel and 11.5% of officers. By 2003, they made up 15.0% of enlisted
personnel and 15.3% of officers (U.S. Department of Defense, 2003). Women are less
likely to encounter a glass ceiling in the public sector, with Stivers (2002) reporting
that there are about twice as many senior managers who are females in the federal gov-
ernment than there are in the private sector. At the state and local levels, we also find
progress. In 2001, women made up 45.3% of state and local employees (not including
employees in school systems), up from 41.0% in 1980. Between 1987 and 2001,
women increased their share of official and administrative positions in state and local
government from 29.1% to 36.9%, as well as their share of professional jobs from
48.1% to 55.6% (Stivers, 2002; U.S. Census Bureau, 2007).
The substantial progress in gender representation achieved in the public sector
suggests that it might provide attractive setting for females. There is empirical evi-
dence to support the claim that differences in public–private employment conditions
lead women to prefer working in the public sector. Most pertinent for this article is
recent work by Llorens, Wenger, and Kellough (in press). They cite data that show
that in 2003, women held 50.7% of all state government positions, relative to 46.5%
of positions in the private sector. They also demonstrate that the ratio of female par-
ticipation in state government to that in the state civilian labor force is positive in all
states, with a U.S. average of 1.19.
Why would women prefer the public sector? A primary reason is that women face
a lower wage penalty than they would in the private sector. The existence of such
wage penalties is significantly associated with the overrepresentation of women in
state governments (Llorens et al., in press). Cross-national studies that include the
United States, Canada, and European countries provide additional evidence that

Downloaded from rop.sagepub.com at COLUMBIA UNIV on February 4, 2015

 Moynihan, Landuyt / Turnover Intention in State Government 125

women disproportionately choose to join the public sector, and in doing so they tend
to earn more than in the private sector and also have a greater potential to close the
earning gap with male colleagues (Gornick & Jacobs, 1998). Bernhardt and Dresser
(2002, p. 1) also argue that women will be more attracted to the public sector
because of the stability and due process protections provided by stronger unions as
well as the better benefits (see also Kolberg, 1991). In the public sector, women have
greater potential to find balance between work and family life with classification
systems that offer a predictable career path and less potential for discrimination.
Women are also more likely to enjoy better leave policies than in the private sector,
if for no other reason than the fact that public employers are subject to the require-
ments of the Family and Medical Leave Act, whereas only private employers with
more than 50 people are subject to these provisions. Beyond the ability to access
more attractive employment policies, women may be more attracted to stay in the
public sector because of a desire to serve the public interest. Women are more likely
than men to express high levels of public service motivation (DeHart-Davis,
Marlowe, & Pandey, 2006; Naff & Crum, 1999).
In fact, previous public sector studies provide no empirical evidence for the belief
that women in the public sector are significantly more likely to depart (Huang et al.,
2003; Kellough & Osuna, 1995; Kim, 2005; Lewis & Park, 1989; Mor Barak et al.,
2001; Smith, 1979). Some recent work has found women significantly less likely to
quit public sector jobs in social services (Curry et al., 2005) and federal positions
(Bertelli, 2007; Lee & Whitford, in press), although the implications of these find-
ings were not explored.
For the above reasons, we propose that females are less likely to express an inten-
tion to quit than males.

Organizational Factors
Organizational factors can be divided into three categories (Kim, 2005)—job
characteristics, HRM practices, and work environment—although it should be
acknowledged that many variables placed into one of these categories overlap with
others. Job characteristics include workload, job satisfaction, and supervisory status.
HRM practices include pay, promotion, benefits, efforts to foster diversity, and
family-friendly job attributes. The aspects of work environment include empower-
ment, voice, and loyalty.

Job characteristics. Employees with favorable job characteristics are expected to be

more satisfied and less likely to leave (Cotton & Tuttle, 1986; Iverson & Currivan, 2003;
Mobley, 1977; Mor Barak et al., 2001; Munn, Barber, & Fritz, 1996). We measure job
satisfaction in the context of whether employees feel that their expectations have been
met, an approach inspired by Porter and Steers’ (1973) argument on the role of met
expectations and turnover (see also Cotton & Tuttle, 1986; Michaels & Spector, 1982).
This reflects the understanding that the job satisfaction variable represents in a single

Downloaded from rop.sagepub.com at COLUMBIA UNIV on February 4, 2015

126 Review of Public Personnel Administration

indicator a range of job characteristic factors, such as role clarity, goal congruency with
employers, recognition of work, and relationships with supervisors (Wright & Davis,
2002). Job characteristics that lead to employee burnout are expected to increase inten-
tion to leave (Huang et al., 2003; Kim, 2005). We therefore test the effects of workload,
assuming that employees who perceive that they are asked to maintain an unreasonable
workload will be more likely to consider leaving.
Evidence on the relationship between occupational level and turnover is mixed.
The limited public sector literature suggests that higher level employees are less
likely to quit, but the more extensive private literature suggests the opposite, perhaps
because of greater opportunities to move. Kellough and Osuna (1995) find that fed-
eral agencies with more clerical employees have higher quit rates, and Lewis (1991)
finds more senior federal employees less likely to leave. However, once the related
effects of age and length of experience are taken into account, individuals at higher
levels may be more likely to leave because of access to alternative positions (Cotton
& Tuttle, 1986).

HRM policies. A number of authors have made the case that HRM policies can
and should be designed to reduce turnover (Arthur, 1994; Gould Williams, 2004;
Mobley, 1977; Osterman, 1987; Selden & Moynihan, 2000; Shaw, Delery, Jenkins,
& Gupta, 1998). There are a number of specific policies that have been hypothesized
to reduce turnover. We test the role of promotion, pay, benefits, family-friendly poli-
cies, employee development, and employee diversity policies.
Employees who have received a merit promotion are expected to be less likely to
leave. This is consistent with research on the effects of promotions on individuals
(Lazear, 1999; Lee & Whitford, in press) and with findings that organizations with
greater room for promotion tend to enjoy lower turnover (Kellough & Osuna, 1995;
Selden & Moynihan, 2000; Smith, 1979). Another straightforward hypothesis is that
individuals who enjoy better compensation are less likely to quit (Blau & Kahn,
1981; Cotton & Tuttle, 1986; Kim, 1999; Lazear, 1999; Leonard, 1987; Park, Ofori-
Dankwa, & Bishop, 1994; Shaw et al., 1998). Public sector research supports this
hypothesis based on comparisons of state pay differentials (Selden & Moynihan,
2000), perception of pay satisfactoriness (Lee & Whitford, in press), and actual mea-
sures of salary (Lewis & Park, 1989). A related hypothesis is that employees who
perceive that they enjoy relatively generous benefits are less likely to leave (Shaw
et al., 1998). There is some preliminary empirical support that these hypotheses
should be true for this sample: A survey of voluntary separations from Texas state
government found that 24% cited a desire for better pay or benefits as a primary rea-
son for their exit (State Auditor’s Office [SAO], 2007).
HRM policies that facilitate employee development are expected to reduce
turnover, but this hypothesis has a mixed record in terms of empirical support. For
instance, Curry et al. (2005) find that meeting employee development needs encour-
ages retention whereas others do not (Kim, 2005; Shaw et al., 1998). Indeed, Ito
(2003) argues that the reverse outcome is likely and offers evidence that investments

Downloaded from rop.sagepub.com at COLUMBIA UNIV on February 4, 2015

 Moynihan, Landuyt / Turnover Intention in State Government 127

in staff training increase employee marketability and make it more likely that the
employee will leave.
HRM policies that make it easier for employees to balance work and family com-
mitments have been hypothesized to reduce turnover rates (Durst, 1999; Ezra &
Deckman, 1996; Newman & Mathews, 1999). At the state level, there is evidence
that the existence of family-friendly child care policies helps to explain variation in
turnover rates (Selden & Moynihan, 2000). HRM policies that aim to increase diver-
sity in the work environment may also reduce turnover. Wise and Tschirhart (2000)
identify a number of empirical studies that show a negative relationship between
diversity, intent to leave, and actual turnover.

Work environment: Loyalty, empowerment, and exit. The work environment

reflects how organizational rules and culture interact with individual employee per-
spectives. Three concepts that have been hypothesized to reduce turnover are loyalty,
voice, and empowerment.
A widely tested and supported hypothesis is that the sense of attachment an indi-
vidual feels toward job or fellow employees reduces turnover intent. This hypothe-
sis has been tested using a variety of measures but is usually conceptualized as
organizational commitment (see Cohen, 1993, for a meta-analysis of this research;
see also Arthur, 1994; Hochwarter, Perrewé, Ferris, & Guercio, 1999; Ito, 2003;
Michaels & Spector, 1982; Mor Barak et al., 2001; Stremmel, 1991). A recent vari-
ation on the commitment concept is Lee and Whitford’s (in press) elegant applica-
tion of Hirschman (1970) to argue that the choices of exit, voice, and loyalty are
interactive influences on turnover, with loyalty reducing the intent to quit.
Another aspect of the work environment is the sense of control that employees
have in the workplace. Some turnover research proposes that employees who per-
ceive greater control are less likely to leave. This general proposition has found form
in two different concepts. The first is Hirschman’s (1970) voice concept. Voice
options have been found to reduce turnover (Iverson & Currivan, 2003; Lee &
Whitford, in press). The second concept that ties individual control to turnover is the
concept of empowerment, in the form of participation (Kim, 2005) or autonomy.
Private sector research provides evidence that employees who perceive greater
autonomy are less likely to quit (Muchinsky & Morrow, 1980). Among public sec-
tor studies, a study of teacher turnover found that inability to provide input into deci-
sion making contributed to turnover (Ingersoll, 2001), and a study of social work
employees found job autonomy negatively related to turnover (Lee & Ashworth,
1993). We test the effects of both voice and empowerment separately.

Data and Method

Our data come from a survey of state employees in Texas. In 2007, 15,769 of the
state’s 145,721 employees voluntarily separated from their agencies (not including

Downloaded from rop.sagepub.com at COLUMBIA UNIV on February 4, 2015

128 Review of Public Personnel Administration

retirements), providing a voluntary turnover rate of 10.8% and suggesting that vol-
untary turnover is a significant enough problem in the state to merit empirical inves-
tigation (SAO, 2007).1 The data were obtained from a 2004 survey of Texas state
agencies (listed in Appendix A), the Survey of Organizational Excellence. The sur-
vey was administered by the Organizational Excellence Group at the University of
Texas at Austin. Although the Office of Personnel Management and the Merit
Systems Protection Board regularly conduct large-scale surveys of federal person-
nel, there is not, to our knowledge, a similar-scale effort to survey state government
employees. A total of 62,628 employees were surveyed in 53 different state agen-
cies, resulting in 34,668 usable responses, a response rate of more than 55%. This
confidential survey was administered primarily via e-mail from which the respon-
dent was provided an Internet link to an online survey. Data collection was open for
up to 3 weeks and reminder e-mails were sent to potential respondents. Employees
without easy access to the Internet were identified in advance and given a paper ver-
sion of the survey. Most organizations used a census strategy in which all full-time
benefit eligible employees were invited to participate. However, two organizations
used a 20% random sample of all employees.
The survey is designed to capture common organizational and employment con-
cepts. The measurement of all of our variables, Cronbach’s alpha scores for
indexes, and descriptive statistics are found in Appendix B. The measurement of
most of the variables is relatively straightforward and requires little comment. One
measurement issue that is of note, however, is clarifying the relative contribution of
between voice and empowerment. These concepts are sometimes used interchange-
ably and measured in different ways, muddying our understanding of causal rela-
tionships. We therefore test each concept, employing measures we consider to have
strong validity. Iverson and Currivan (2003) measure voice as participation in union
activities, whereas Lee and Whitford (in press) consider voice in terms of the sense
of empowerment employees enjoy, as well as control over their work processes.
Kim’s (2005) operationalization of participation is similar to Lee and Whitford’s
treatment of voice, in that both reflect employee input on decisions. We test the
voice variable with an index of responses to the following statements: “The work
atmosphere encourages honest and open communication,” “My ideas and opinions
count at work,” and “People who challenge the status quo are valued.” This index
conceptualizes voice as the degree to which members feel that they can represent
their views within the hierarchy, even if those views run counter to the dominant
view. The index is therefore consistent with Hirschman’s (1970) original definition
of voice: “The firm’s customers or the organization’s members express their dissat-
isfaction directly to management or to some other authority to which management
is subordinate or through general protest addressed to anyone who cares to listen:
This is the voice option” (p. 4). We measure empowerment based on the degree to
which employees agree that decision making and control are given to employees
doing the actual work.

Downloaded from rop.sagepub.com at COLUMBIA UNIV on February 4, 2015

 Moynihan, Landuyt / Turnover Intention in State Government 129

Respondents could choose from a five-point Likert scale response set ranging
from strongly agree to strongly disagree. Respondents could also mark don’t
know/not applicable or leave the item blank, and these responses were not used in
our analysis. Various demographic items were also collected. Our respondents
closely mirror state employees (SAO, 2002).2 To control for the effect of retirements,
we excluded any respondent aged 60 or over.3
Because our dependent variable is based on a yes or no response, we used a logis-
tic regression, including a dummy variable for each of the 53 agencies included in
the analysis. One advantage of an agency fixed-effects approach is that it controls for
unobserved agency-level differences that we do not otherwise specify in our model.
A good example would be how differences in organizational culture shape turnover
intention of employees. Although a fixed-effects approach does not tell us the size
or significance of such agency-level variables, it does at least ensure that such fac-
tors do not otherwise bias the variables that are tested in the model. None of the
agency dummy variables proved to be significant or alter the nature of the relation-
ship between the independent and dependent variables. Although the agency con-
trols are not reported in the results section, they were retained in the model.

Measurement of the Dependent Variable

Our dependent variable is a reversed measure of turnover intention: “I plan to be

working for this organization in two years.” To ease interpretation, we coded a no
response, signaling intent to leave the organization as 1, and yes as 0. Of course,
turnover intention is not actually turnover and this raises the question of using inten-
tion as a surrogate for actual quits. There are research advantages to examining intent
to quit rather than actual quits. Researchers can use cross-sectional models, can more
easily access the perceptions of potential quits and relate them to their organizational
context, can examine a larger sample of employees, and can identify differences
between those who wish to stay in the organization and those intent on leaving. The
use of turnover intention is relatively common. For example, a majority of the
turnover research examined in the meta-analyses by Wise and Tschirhart (2000) and
Mor Barak et al. (2001) used intent to leave rather than actual turnover.
Validity concerns about using turnover intention is reduced by evidence of its
strong relationship with actual quits, and evidence that suggests that there is not
marked variation between variables that explain intention to quit and actual turnover
(Cotton & Tuttle, 1986, p. 65; Mobley et al., 1979). A survey of five meta analyses
of turnover have shown correlations between intent to quit and actual turnover rang-
ing between .31 and .52 (Dalton, Johnson, & Daily, 1999, p. 1342). The relationship
between intention to quit and actual quits has been found to vary considerably across
studies, but turnover intention remains the best affective measure of turnover
(Vandenberg & Nelson 1999, p. 1315).

Downloaded from rop.sagepub.com at COLUMBIA UNIV on February 4, 2015

130 Review of Public Personnel Administration

Table 1
Logistic Regression on Intent to Leave
Variable (Hypothesized Relationship) Coefficient Odds Ratio Predicted Probabilities

Individual
Primary earner (–) –.189**** .828 –.020
Household size (–) –.149**** .861 –.016
Age (–) –.512**** .599 –.053
Years in state of Texas (–) –.103* .902 –.011
Agency experience (–) –.366**** .693 –.038
Education (+) .484**** 1.623 .050
Female (–) –.261 .770 –.027
African American/Black .182*** 1.200 .019
Hispanic/Mexican American .063 1.065 .007
Asian American/Pacific Islander/ –.233 .792 –.024
Native American
Multiracial/Other .269** 1.308 .028
Job characteristics
Job satisfaction (–) –.480**** .619 –.050
Workload (+) .029 1.030 .003
Supervisor (–) –.971* .907 –.010
HRM practices
Salary (–) .269**** 1.308 .028
Perception of fair pay (–) –.081**** .922 –.008
Benefits (–) .023 1.024 .002
Merit promotion (–) –.383**** .682 –.040
Family-friendly work practices (–) –.033* .967 –.004
Diverse workforce practices (–) –.140**** .869 –.015
Employee development (–) –.009 .991 –.001
Work environment
Commitment/loyalty (–) –.115**** .891 –.012
Empowerment (–) –.042** 1.042 .004
Voice (–) –.002 1.002 .000

Note: Agency-specific effects included; N = 24074; log-likelihood = –8913.97; Nagelkerke R-square =

.223; McFadden’s adjusted R-square = .0152; percentage correctly predicted = 85.2; HRM = human
resource management.
*significant at .10 (two-tailed tests). **significant at .05. ***significant at .01. ****significant at .001.

Results and Discussion

The results of the analysis are presented in Table 1. As with standard regressions,
we provide coefficients and measures of significance. In addition, because we use a
logit approach, we provide the odds ratio and predicted probabilities (Long & Freese,
2006). The predicted probabilities tell us the probability differences between aspects
of the independent variable. The odds ratio is a relative measure of the probability of

Downloaded from rop.sagepub.com at COLUMBIA UNIV on February 4, 2015

 Moynihan, Landuyt / Turnover Intention in State Government 131

the dependent variable to occur given the event of the independent variable under
study. The higher the score is above one, the more likely it is that employees will state
intention to quit; the further the score is below one, the less likely it is that employees
will declare their intent to quit. The difference between the two can be illustrated with
the following example. The predicted probability for supervisors is –.01. This means
that supervisors are 1 percentage point less likely to state an intention to quit than non-
supervisors. The odds ratio for supervisors is .907. This means that supervisors are
9.3% less likely than non-supervisors to state an intention to quit.
The results provide varying degrees of support for most of the variables hypoth-
esized to have a relationship with turnover, and largely in directions consistent with
previous literature. Most of the standard hypotheses about turnover appear to apply
to this particular public setting. To the extent that results run counter to previous
research, we consider them in greater detail below. As an aside, it is worth noting
that the large sample size helps to move more variables into the realm of statistical
significance than would be the case for a smaller sample.

Findings on Life Cycle Stability

The findings on individuals support the life cycle stability hypothesis. Employees
who have reached a certain measure of stability in their life, and who have pressing
economic and familial concerns, are less likely to seek the changes brought about by
seeking a new job. Although much previous work sometimes conflates the effects of
age and experience, this article shows that each has a separate negative effect on
turnover intention. Employees who are over the age of 30 are 5.3 percentage points
less likely to state an intention to quit than employees under the age of 30. Similarly,
employees with 6 or more years of service are 3.8 percentage points less likely to
state an intent to leave than employees with less experience. The finding on length
of service is not only consistent with a human capital view that employees develop
firm-specific capabilities that make it difficult for them to switch firms, but it is also
consistent with a general reluctance to pursue change, as suggested by the signifi-
cant results for the other life cycle stability variables.
Unlike previous studies, we also incorporate geographical predictors of employ-
ment stability. The tendency to have lived in the same area for a considerable period
of time also indicates reluctance to change, which tends to limit the search for new
employment opportunities and the intention to exit the workplace. It is worth noting
that this factor is significant only at .1 levels, and so the finding should remain
subject to further questioning. Economic constraints posed by being the primary
wage earners or being part of a larger household also reinforce a preference for the
status quo. Our confidence in these findings is increased by the fact that we measure
multiple aspects of the life cycle stability hypothesis (as well as other possible

Downloaded from rop.sagepub.com at COLUMBIA UNIV on February 4, 2015

132 Review of Public Personnel Administration

correlates such as education, pay, and supervisory status). The results therefore
reveal the separate effects of each variable.

Findings on Gender and Race

We find that women are significantly less likely to state intent to quit than their
male counterparts. Women were 2.7 percentage points less likely than males to state
an intention to quit, holding other factors constant. The size of the odds ratio is rel-
atively large. Females were 23% less likely than males to state an intention to quit.
The finding is consistent with evidence presented on the changing nature of female
participation in the labor force and on the particular attraction of public work to
females. This result proved robust to different specifications of the model we
employed.4
One qualification to the finding is that our dependent variable may have limitations
in predicting gender effects for actual turnover. This is true if many females who
would prefer to stay in their current organization, and imagine themselves doing so,
find themselves forced to drop out of the workforce for unexpected reasons. In our
sample, even before the inclusion of controls, women were significantly less likely to
state an intention to quit than men. The state of Texas reports little difference between
all turnover (including retirements and firing) by gender with the gender breakdown
of the workforce, but this does not control for factors that reduce the likelihood that
women will quit, and it therefore underestimates women’s willingness to stay (SAO,
n.d.). Another qualification is that our findings may be of limited generalizability if a
state government job in Texas is unusually attractive relative to other work opportu-
nities for women. Some evidence for this qualification comes from Llorens et al. (in
press), who report that, along with Oklahoma, Texas is the state with the highest over-
representation of women relative to their participation in the state labor force.
However, they also note that women are overrepresented in all state governments sug-
gesting that although Texas is something of an outlier, the basic relationship between
gender and attractiveness of public service holds in other states.
The results on race and turnover intention were mixed. Both the African
American/Black and the Multiracial/Other categories are positively and significantly
related to turnover intention. However, the other minority categories, Hispanic/
Mexican American and Asian/Pacific Islander/Native American, are not. One clear
caution to our findings is the disproportionate size of the Hispanic category, which
includes one-quarter of the sample. Although members of this category are techni-
cally a minority, they represent a large enough group that they may not face the per-
ceived disadvantages of minority status in the workplace. Given the increasing
trend toward greater diversity in the workplace and the growing representation of
Hispanics in particular, the results suggest that grouping minorities into one single

Downloaded from rop.sagepub.com at COLUMBIA UNIV on February 4, 2015

 Moynihan, Landuyt / Turnover Intention in State Government 133

category may miss distinct patterns across groups, although it is beyond the scope of
this article to explain such differences.

Findings on Job Characteristics

The findings on job-related factors confirm the powerful role that previous
research has detailed for job satisfaction, even as we measure job satisfaction in
terms of met expectations, rather than a more direct indicator of satisfaction typically
found in previous research. Job satisfaction, along with education, has the largest
probability effect of any of the variables tested. A one-unit increase in our job satis-
faction scale was associated with a 5-percentage-point reduction in the probability
that an individual would state an intention to quit. The results for workload suggest
that greater workload is negatively associated with turnover, but the result is slightly
above standard levels of statistical significance.

Findings on HRM Practices

As predicted, we find that perceptions of fair pay were negatively and significantly
related to turnover. This is consistent with other findings, including those from the fed-
eral level (Lee & Whitford, in press). However, actual salary is positively related to
turnover. The probability that employees earning more than $45,000 per year will state
an intention to quit is 2.8 percentage points higher than employees earning less than
that amount. This finding is at odds with previous evidence on the effect of salaries at
the federal level (Lewis & Park, 1989), but it is robust to running alternate specifica-
tions of the model, such as dropping frequent correlates of pay (age, education, and
supervisory status). The results also remain if we use a categorical measure of pay
rather than a dummy variable. So what explains this seemingly counterintuitive result?
If we accept the claims that public–private pay disparities are greatest among more
senior staff and that public employees are not likely to measure pay satisfaction with
others in the organization (because classification controls limit variation among work
peers), but with their perceived sense of fair pay relative to opportunities outside the
organization, this raises a possible explanation. Actual salary may indicate the senior-
ity that the employee has reached. Therefore, the higher the salary the more likely the
employee perceives a gap in salary relative to private sector counterparts, the greater
the sense that pay is not adequate, and the greater the likelihood of turnover.
Another possible explanation is suggested by the findings on education, which,
consistent with previous research, show a positive relationship with turnover inten-
tion. Employees with a college degree or higher were 5 percentage points more
likely to state an intention to leave than those with lower educational attainment. The
logic behind the role of education—that employees with greater human capital have

Downloaded from rop.sagepub.com at COLUMBIA UNIV on February 4, 2015

134 Review of Public Personnel Administration

a greater ability to win alternative employment—may also apply to those who are
paid better. Salary might be considered as an indicator of the skill set or capacity of
employees in the workplace, beyond educational attainment and supervisory status
(which is also controlled for and is negatively related to turnover intention), and
these skills may provide the employee with greater ability to switch to another job.
This interpretation is somewhat at odds with a human capital view, as it suggests that
organizations reward nonfirm specific skills that employees can use to find other
jobs. In many instances, organizations may be forced to do so precisely because the
potential threat of exit pushes them to provide greater rewards to employees whose
skills are more marketable.
The receipt of a merit promotion in the past 2 years had a powerful effect on
turnover intention. Those who had received a promotion were 4 percentage points
less likely to state an intention to quit than those who had not received a promotion.
This suggests the importance of individual recognition and concrete career improve-
ment provided by promotions and is consistent with findings from both the private
sector (Lazear, 1999) and the federal level (Lee & Whitford, in press). The existence
of family-friendly policies was negatively associated with turnover, but only at mar-
ginal levels of statistical significance.5 HRM efforts to foster employee development
through the provision of training and information about job opportunities was not
significantly related to reducing turnover intention. One reason for this may be that
employee development has two conflicting effects that cancel each other out. On one
hand, such investment may signal that the organization values an employee, which
in turn fosters a greater sense of employee loyalty to the organization (Curry et al.,
2005). On the other hand, providing greater training to the employee may make them
more skilled and capable of winning a job in other organizations (Ito, 2003). A
research challenge that suggests itself is the need for a measurement strategy that can
sort out the relative influence of these conflicting effects.
Our findings on diversity are more encouraging for HRM managers. To the extent
that employees feel that they are in a work environment where there is a deliberate
effort to foster diversity, they are less likely to consider leaving. Employees who
were one point higher on the five-point diversity practices scale were 1.5 percentage
points less likely to state an intention to quit. Lee and Whitford (in press) also find
such a relationship at the federal level, but only for the most senior managers.6 The
findings offer empirical support for the claims of Wise and Tschirhart (2000) on the
benefits of diversity for public organizations.

Findings on Loyalty, Empowerment, and Voice

Our analysis of work environment factors focused on clarifying the relative con-
tributions of loyalty, the provision of employee autonomy over work processes and

Downloaded from rop.sagepub.com at COLUMBIA UNIV on February 4, 2015

 Moynihan, Landuyt / Turnover Intention in State Government 135

decisions, and the abilities of employees to use voice to critique the organization. We
find that our measure of loyalty is a significant negative predictor of turnover inten-
tion. Every one-unit increase on the 3-15-point scale for loyalty is associated with a
1.2 percentage point reduction in the probability of stating an intention to quit. This
is consistent with previous evidence on the related concept of commitment and with
Lee and Whitford’s (in press) findings, even though we use a different measure of
the concept.
From a research perspective, our measurement approach offers a way of distin-
guishing between empowerment and voice. Although both are negatively associated
with turnover intention, only the empowerment variable falls into an acceptable range
of statistical significance, and even this variable is just significant at the .05 level.
Every one-unit increase on the 1-5 empowerment scale leads to a .4 percentage point
reduction in the likelihood of stating an intention to quit. This supports previous
findings. However, the voice variable proved not to be significant, in contrast to
other research (Iverson & Currivan, 2003; Lee & Whitford, in press). From a prac-
tical perspective, the results from our sample suggest that organizations seeking to
reduce turnover by shaping the work environment will have the greatest effect if they
focus on fostering organizational loyalty rather than on empowerment or voice
mechanisms.

Conclusion

This article identified a model of employee turnover based on previous research,

and it tested this model on a large sample of state employees. Our focus was on indi-
vidual and organizational factors, and the results provide information on whether
well-established claims about turnover apply to this population. One important
caveat is that our cross-sectional test of a single state limits our ability to include
other variables that have been previously found to affect turnover, such as local eco-
nomic conditions or union presence. Another caveat is that we examined only a sin-
gle state, and so generalizations to other governments should be made with caution.
However, most of the variables found predictable results and therefore they do not
suggest that the factors that cause turnover in Texas are dramatically different from
the factors that cause turnover elsewhere.
One area where our results differ from much of previous research is the relation-
ship between gender and turnover. Our most interesting finding is that females are
less likely to plan to quit, contradicting previous research that suggested females
were more likely to quit, or (more convincingly) that gender did not matter to
turnover once relevant controls were accounted for. As with any finding that chal-
lenges previous research, we present our results cautiously and urge replication in

Downloaded from rop.sagepub.com at COLUMBIA UNIV on February 4, 2015

136 Review of Public Personnel Administration

other settings. However, this article presented logical reasons to expect such a result,
tied to changing levels of female workforce participation and to the attractive fea-
tures that public employment offers to females. At the very least, researchers inves-
tigating the role of gender in public sector turnover in the future should be more
willing to entertain the hypothesis that females are more likely to desire to stay
rather than to leave.
Our findings on individual factors provide support for the life cycle stability
hypothesis. Rather than consider age, experience, and geographical stability as unre-
lated control variables, they should be conceptualized as closely connected indica-
tors of why some employees have reached a point in their life where they want to
maintain lifestyle status quo rather than embrace change. Economic and familial
constraints also enforce a measure of lifestyle stability inconsistent with turnover.
Such employees should be recognized among HR managers as having a tangible
value in terms of their likelihood to stay with the organization.
From an HRM perspective, the results are somewhat sobering. HR managers have
the least control over the factors that seem to matter the most in predicting turnover.
Individual factors are clearly important (as are economic variables not tested here),
but there is little HRM managers can do to control these. Of the organizational
factors that matter a great deal—job satisfaction, merit promotions, and to a lesser
extent pay and loyalty—these are the factors that HRM managers have the least con-
trol over. They are determined by the qualities and capacity of the individual, by
externally imposed pay constraints, relationships with supervisors, organizational
culture, and only partly by HRM policies. The factors that HRM managers have
most direct control over—employee development through training and provision of
information and creating flexible work environments—are not critical factors in
reducing turnover when these other variables are accounted for. The one exception to
this finding is diversity policies, which have a negative effect on turnover intention.

Appendix A
Agencies Surveyed
Board of Law Examiners
Board of Nurse Examiners
Board of Plumbing Examiners
Employees Retirement System
Executive Council on PT & OT Examiners
General Land Office
Interagency Council on Early Childhood Intervention
Office of the Attorney General of Texas
Public Utility Commission of Texas
State Pension Review Board

(continued)

Downloaded from rop.sagepub.com at COLUMBIA UNIV on February 4, 2015

 Moynihan, Landuyt / Turnover Intention in State Government 137

Appendix A (continued)
Teacher Retirement System of Texas
Texas Alcoholic Beverage Commission
Texas Board of Architectural Examiners
Texas Board of Medical Examiners
Texas Commission on Fire Protection
Texas Commission on State Emergency Communications
Texas Credit Union Department
Texas Department of Banking
Texas Department of Criminal Justice–Administration
Texas Department of Health
Texas Department of Housing and Community Affairs
Texas Department of Human Services
Texas Department of Information Resources
Texas Department of Insurance
Texas Department of Licensing and Regulation
Texas Department of Mental Health and Mental Retardation
Texas Department of Protective and Regulatory Services
Texas Department of Transportation
Texas Department on Aging
Texas Education Agency
Texas Forest Services
Texas Funeral Service Commission
Texas Higher Education Coordinating Board
Texas Historical Commission
Texas Juvenile Probation Commission
Texas Lottery Commission
Texas Military Facilities Commission
Texas Office of Consumer Credit Commissioner
Texas Parks and Wildlife Commission
Texas Real Estate Commission
Texas Savings and Loan Department
Texas State Auditor’s Office
Texas State Board for Educator Certification
Texas State Board of Barber Examiners
Texas State Board of Examiners of Psychologists
Texas State Board of Pharmacy
Texas State Board of Veterinary Medical Examiners
Texas State Library and Archives Commission
Texas Structural Pest Control Board
Texas Transportation Institute
Texas Veteran’s Commission
Texas Water Development Board
Texas Workforce Network

Downloaded from rop.sagepub.com at COLUMBIA UNIV on February 4, 2015

 Appendix B

138
Variables Measured and Descriptive Statistics
Standard
Variables Survey Items Minimum Maximum Mean Deviation

Intent to leave I plan to be working for this organization in two 0 1 .16 .37
years (1 = No; 0 = Yes)
Supervisor I am currently in a supervisory role (1 = Yes; 0 = No) 0 1 .23 .42
Primary earner I am the primary wage earner in the household 0 1 .66 .48
(1 = Yes; 0 = No)
Household size Persons in my household, including myself 1 5 2.75 1.23
(1 = 1; 2 = 2; 3 = 3; 4 = 4; 5 = 5 or more)
Age If age is 30-59 = 1; less than 30 = 0 0 1 .63 .48
Years in state of Texas Years that I have lived in this state: 1 = less than 1 3 2.91 .32
2; 2 = 2-10; 3 = more than 10
Agency experience Years of service with this organization: 1 = 6 + years; 0 1 .60 .49
0 = 5 years or less
Education Bachelor’s degree or higher = 1; less than a 0 1 .49 .50
bachelor’s degree = 0
Female Male = 0; female = 1 0 1 .67 .47
African American/Black African American/Black = 1; all others = 0 0 1 .14 .34
Hispanic/Mexican American Hispanic/Mexican American = 1; all others = 0 0 1 .25 .43
Asian American/Pacific Islander/ Asian American/Pacific Islander/Native 0 1 .02 .14
Native American American = 1; all others = 0

Downloaded from rop.sagepub.com at COLUMBIA UNIV on February 4, 2015

Multiracial/Other Multiracial/Other = 1; all others = 0 0 1 .04 .20
Job satisfaction My job meets my expectations 1 5 3.44 1.07
Workload The amount of work I am asked to do is reasonable 1 5 3.30 1.19
Salary 1 = $45,001 or more; 0 = $45,000 or less 0 1 .19 .39
Fair pay Salaries are competitive with similar jobs in the community 1 5 2.40 1.17
Benefits (Cronbach’s I am satisfied with my medical insurance/sick leave/ 8 44 26.20 5.61
alpha = .823) vacation/retirement/dental insurance/vision insurance/
holiday benefit/Employee Assistance Program

(continued)
 Appendix B (continued)
Standard
Variables Survey Items Minimum Maximum Mean Deviation

Merit promotion I received a merit promotion in the last two 0 1 .23 .42
years (1 = Yes; 0 = No)
Family-friendly work When possible, alternative work schedules (flextime, 1 5 3.50 1.24
practices compressed work weeks, job sharing, telecommuting)
are offered to employees
Diverse workforce We work to attract, develop, and retain people 1 5 3.35 1.16
practices with diverse backgrounds
Employee development Training is made available to us for personal growth 2 10 6.92 1.94
(Cronbach’s alpha = .761) and development. We have access to information
about job opportunities, conferences, workshops, and
training
Commitment (Cronbach’s There is a basic trust between employees and 3 15 10.08 2.90
alpha = .787) supervisors. We feel a sense of pride when we
tell people that we work for this organization.
Within my workplace, there is a feeling of community
Empowerment Decision making and control are given to employees 1 5 3.03 1.20
doing the actual work
Voice (Cronbach’s alpha = .853) The work atmosphere encourages honest and open 3 15 9.18 3.10

Downloaded from rop.sagepub.com at COLUMBIA UNIV on February 4, 2015

communication. My ideas and opinions count at work.
People who challenge the status quo are valued

Note: All data are from the Survey of Organizational Excellence. Unless otherwise noted, all responses are based on a 1-5 scale: 1 = strongly disagree; 2 =
strongly agree; 3 = feel neutral; 4 = agree; 5 = strongly agree. Respondents also have the opportunity to choose don’t know/not applicable. Such responses
are excluded from the scale.

139
140 Review of Public Personnel Administration

Notes
1. In describing the level of turnover intention in Texas, we might ask whether this is too high. It
should be noted that there is no clear agreement on what an optimal level of turnover looks like. On one
hand, some turnover is helpful in creating space for internal promotions and introducing organizational
learning and innovation through new employees. On the other hand, voluntary turnover implies the loss
of an employee the organization would otherwise wish to continue to employ, as well as hiring and train-
ing costs. Overall, the bias in studies of turnover is that less is better. That said, we do not have any strong
empirical basis to determine whether turnover among Texas state employees is actually damaging state
government performance. However, recent research by Meier and Hicklin (in press) sheds some light on
this question. They test whether there is an inverted U–shaped relationship between turnover and perfor-
mance, that is, at relatively low levels turnover improves performance by bringing in new employees, but
that after a certain point turnover becomes too great and is negatively related to performance. Using two
measures of high school performance, Meier and Hicklin find mixed evidence. For test scores, turnover
has a linear and negative effect on performance, suggesting that any level of turnover is damaging.
However, turnover did have a nonlinear relationship for a secondary organizational goal characterized by
greater task difficulty, preparing students for college. This suggests that the relationship between turnover
and performance is contingent on the nature of the task. Given that most public organizations have tasks
of varying complexity, this makes it difficult to make definitive statements about what the optimal level
of turnover would be.
2. Of the state employees of Texas, 44% are non-White, which matches the proportion of non-Whites
among respondents. The mean age of a state of Texas employee was 42.4 years in 2002. Among respon-
dents, the most frequently selected age category was the 40-49 range. The experience of respondents in
our sample is also similar to the overall state workforce. Almost 40% have less than 5 years of experi-
ence. The equivalent number for the entire workforce is 43%. The one area where there is a sizeable dif-
ference between our respondents and the state workforce is in gender: 56% of state employees are
females, compared to 67% of respondents. However, this is primarily because our sample did not include
the Department of Criminal Justice, and of the remaining state agencies, 66% are women, consistent with
our respondents. Our model includes measures of relevant demographic factors, ensuring that the results
of other independent variables are not affected by any possible sample bias.
3. Given that it is still possible that some employees in their late 50s could be eligible for retirement,
we also ran the model excluding all employees over 50, reducing the N size to 17,470. The results for age
remain the same, although some other variables that are logically correlated with age fall from signifi-
cance: supervisor and salary.
4. For example, males in the sample were much more likely than females to be primary earners, and
so dropping the primary earner variable might be expected to affect our findings. If we drop the primary
earner variable for the model, the gender findings remain significant, although the gender probability dif-
ferential of quitting declines from 2.7 to 2.3 percentage points. The odds ratio also slightly declines as
women become 20% rather than 23% less likely than men to state an intention to quit.
5. As it is likely that family-friendly policies are more likely to be relevant for working mothers than
other groups, we tested the interaction of family-friendly policies and gender. Contrary to expectations,
we found that this interaction of female status and family-friendly policies was positively related to
turnover intention and significant at the .05 level. However, once we also incorporated size of house-
hold through a three-way interaction effect, this variable was significantly and negatively related to
turnover, that is, women from larger households who perceived a family-friendly work environment were
less likely to state an intention to quit. Overall, the interaction terms suggest that the findings reported in
Table 1 may somewhat underestimate the importance of family-friendly policies, especially for women
with familial responsibilities.
6. Following the example of Lee and Whitford (in press), we also explore some of the potential inter-
action effects between diversity practices and particular groups—do they have a greater effect for certain

Downloaded from rop.sagepub.com at COLUMBIA UNIV on February 4, 2015

 Moynihan, Landuyt / Turnover Intention in State Government 141

groups than for others in reducing turnover intention? Lee and Whitford did not find such an interaction,
and neither did our analysis. Interactions of diversity practices with White males, Hispanics, or Blacks did
not prove to be significant predictors of turnover intention.

References
Arthur, J. B. (1994). Effects of human resource systems on manufacturing performance and turnover.
Academy of Management Journal, 37(3), 670-687.
Becker, G. S. (1962). Investment in human capital: A theoretical analysis. Journal of Political Economy,
70(5), 9-49.
Bernhardt, A., & Dresser, L. (2002). Why privatizing government services would hurt women workers.
Washington, DC: Institute for Women’s Policy Research.
Bertelli, A. (2007). Bureaucratic turnover and democratic governance: Evidence from the U.S. Internal
Revenue Service. Journal of Public Administration Research and Theory, 17(2), 235-258.
Blau, F. D., & Kahn, L. M. (1981). Race and sex differences in quits by young workers. Industrial and
Labor Relations Review, 34(4), 563-577.
Bureau of Labor Statistics. (2005). Women in the labor force: A databook. Washington, DC: Author.
Cohen, A. (1993). Organizational commitment and turnover: A meta-analysis. Academy of Management
Journal, 36(5), 1140-1157.
Cotton, J. L., & Tuttle, J. M. (1986). Employee turnover: A meta-analysis and review with implications
for research. Academy of Management Review, 11(1), 55-70.
Curry, D., McCarragher, T., & Dellmann-Jenkins, M. (2005). Training, transfer, and turnover: Exploring
the relationship among transfer of learning factors and staff retention in child welfare. Children and
Youth Service Review, 27(8), 931-948.
Dalton, D. R., Johnson, J. L., & Daily, C. M. (1999). On the use of “intent to…” variables in organiza-
tional research: An empirical and cautionary assessment. Human Relations, 52(10), 1337-1350.
DeHart-Davis, L., Marlowe, J., & Pandey, S. K. (2006). Gender dimensions of public service motivation.
Public Administration Review, 66(6), 873-887.
Durst, S. (1999). The effect of family-friendly programs on public organizations. Review of Public
Personnel Administration, 19(3), 19-33.
Ezra, M., & Deckman, M. (1996). Balancing work and family responsibilities: Flextime and childcare in
the federal government. Public Administration Review, 56(2), 174-176.
Gornick, J. C., & Jacobs, J. A. (1998). Gender, the welfare state, and public employment: A comparative
study of seven industrialized countries. American Sociological Review, 63(5), 688-710.
Gould Williams, J. (2004). The effects of “high commitment” HRM practices on employee attitude: The
views of public sector workers. Public Administration, 82(1), 63-81.
Hirschman, A. O. (1970). Exit, voice, and loyalty: Responses to decline in firms, organizations, and states.
Cambridge, MA: Harvard University Press.
Hochwarter, W. A., Perrewé, P. L., Ferris, G. R., & Guercio, R. (1999). Commitment as an antidote to the ten-
sion and turnover consequences of organizational politics. Journal of Vocational Behavior, 55(3), 277-297.
Huang, I., Chuang, C. J., & Lin, H. (2003). The role of burnout in the relationship between perceptions
of organizational politics and turnover intentions. Public Personnel Management, 32(4), 519-532.
Ingersoll, R. M. (2001). Teacher turnover and teacher shortages: An organizational analysis. American
Educational Research Journal, 38(3), 499-534.
Ippolito, R. A. (1987). Why federal workers don’t quit. The Journal of Human Resources, 22(2), 281-299.
Ito, J. K. (2003). Career branding and mobility in the civil service: An empirical study. Public Personnel
Management, 32(1), 1-21.
Iverson, R. D., & Currivan, D. B. (2003). Union participation, job satisfaction, and employee turnover:
An event-history analysis of the exit-voice hypothesis. Industrial Relations, 42(1), 101-105.

Downloaded from rop.sagepub.com at COLUMBIA UNIV on February 4, 2015

142 Review of Public Personnel Administration

Kellough, J. E., & Osuna, W. (1995). Cross-agency comparisons of quit rates in the federal service:
Another look at the evidence. Review of Public Personnel Administration, 15(4), 15-68.
Kim, M. (1999). Where the grass is greener: Voluntary turnover and wage premiums. Human Relations,
52(10), 584-592.
Kim, S. (2005). Factors affecting state government information technology employee turnover intentions.
American Review of Public Administration, 35(2), 137-156.
Kolberg, J. E. (1991). The gender dimension of the welfare state. International Journal of Sociology,
21(1), 119-148.
Lazear, E. P. (1999). Personnel economics: Past lessons and future directions. Journal of Labor
Economics, 17(2), 199-236.
Lee, R. T., & Ashforth, B. E. (1993). A longitudinal study of burnout among supervisors and managers:
Comparisons between the Leiter and Maslach (1988) and Golembiewski et al. (1986) models.
Organizational Behavior and Human Decision Processes, 54, 369-398.
Lee, S., & Whitford, A. B. (in press). Commitment, voice, and the intent to leave: Evidence from the
public workforce. Journal of Public Administration Research and Theory.
Leonard, J. S. (1987). Carrots and sticks: Pay, supervision, and turnover. Journal of Labor Economics,
5(4), 136-152.
Lewis, G. B. (1991). Turnover and the quiet crisis in the federal civil service. Public Administration
Review, 51(2), 145-155.
Lewis, G. B., & Park, K. (1989). Turnover rates in federal white-collar employment: Are women more
likely to quit than men? American Review of Public Administration, 18(1), 13-28.
Llorens, J. J, Wenger, J. B., & Kellough, E. (in press). Choosing public sector employment: The impact
of wages on the representation of women and minorities in state bureaucracies. Journal of Public
Administration Research and Theory.
Long, J. S., & Freese, J. (2006). Regression models for categorical dependent variables using Stata.
College Station, TX: Stata Press.
March, J. G., & Simon, H. A. (1958). Organizations. New York: Wiley.
Meier, K. J., & Hicklin, A. (in press). Employee turnover and organizational performance: Testing a
hypothesis from classical public administration. Journal of Public Administration Research and Theory.
Michaels, C. E., & Spector, P. E. (1982). Causes of employee turnover: A test of the Mobley, Griffeth,
Hand, and Meglino model. Journal of Applied Psychology, 67(1), 53-59.
Mobley, W. H. (1977). Intermediate linkages in the relationship between job satisfaction and employee
turnover. Journal of Applied Psychology, 62(2), 237-240.
Mobley, W. H., Griffeth, R. W., Hand, H. H., & Meglino, B. M. (1979). A review and conceptual analy-
sis of the employee turnover process. Psychological Bulletin, 86(3), 493-522.
Mor Barak, M. E., Nissly, J. A., & Levin, A. (2001). Antecedents to retention and turnover among child
welfare, social work, and other human service employees: What can we learn from past research? A
review and meta-analysis. Social Science Review, 75(4), 625-661.
Moynihan, D. P., & Pandey, S. K. (2008). The ties that bind: Social networks, person-organization value
fit and turnover intention. Journal of Public Administration Research and Theory, 18(2), 205-227
Muchinsky, P. M., & Morrow, P. C. (1980). A multidisciplinary model of voluntary employee turnover.
Journal of Vocational Behavior, 17, 263-290.
Munn, E. K., Barber, C. E., & Fritz, J. J. (1996). Factors affecting the professional well-being of child life
specialists. Children’s Health Care, 25(2), 71-91.
Naff, K. C., & Crum, J. (1999). Working for America: Does public service motivation make a difference?
Review of Public Personnel Administration, 19(4), 5-16.
Newman, M., & Mathews, K. (1999). Family-friendly work-place policies: Barriers to effective imple-
mentation. Review of Public Personnel Administration, 19(3), 34-48.
Osterman, P. (1987). Turnover, employment security, and the performance of the firm. In M. M. Kleiner,
R. N. Block, M. Roomkin, & S. W. Salsburg (Eds.), Human resources and the performance of the firm
(pp. 227-313). Madison, WI: Industrial Relations Research Association.

Downloaded from rop.sagepub.com at COLUMBIA UNIV on February 4, 2015

 Moynihan, Landuyt / Turnover Intention in State Government 143

Park, H. Y., Ofori-Dankwa, J., & Bishop, D. R. (1994). Organizational and environmental determinants
of functional and dysfunctional turnover: Practical and research implications. Human Relations,
47(3), 353-366.
Porter, L. E., & Steers, R. M. (1973). Organizational, work and personal factors in employee turnover and
absenteeism. Psychological Bulletin, 80(2), 151-176.
Selden, S. C., & Moynihan, D. P. (2000). A model of voluntary turnover in state government. Review of
Public Personnel Administration, 20(2), 63-74.
Shaw, J. D, Delery, J., Jenkins, G. D., Jr., & Gupta, N. (1998). An organization-level analysis of volun-
tary and involuntary turnover. Academy of Management Journal, 41(5), 511-525.
Smith, C. B. (1979). Influence of internal opportunity structure and sex of worker on turnover patterns.
Administrative Science Quarterly, 24(3), 362-381.
State Auditor’s Office. (2002). A summary of the Texas state workforce for fiscal year 2002. Austin, TX:
Author.
State Auditor’s Office. (2007). An annual report on certified employee turnover for fiscal year 2007.
Austin, TX: Author.
State Auditor’s Office. (n.d.). Employee turnover by gender. Retrieved January 29, 2008, from
http://www.hr.state.tx.us/workforce/Turnover2004/Gender.html
Stier, H., & Lewin-Epstein, N. (2001). Welfare regimes, family-supportive policies, and women’s employ-
ment along the life-course. American Journal of Sociology, 106(6), 1731-1760.
Stivers, C. (2002). Gender images in public administration: Legitimacy and the administrative state (2nd
ed.). Thousand Oaks, CA: Sage.
Stremmel, A. J. (1991). Predictors of intention to leave child care work. Early Childhood Research
Quarterly, 6(2), 285-298.
U.S. Census Bureau. (2007). Statistical abstract of the United States 2007. Retrieved January 4, 2008,
from http://www.census.gov/compendia/statab/tables/07s0453.xls
U.S. Department of Defense. (2003). Demographics profile of the military community. Washington, DC:
Author.
U.S. Office of Personnel Management. (1997). The fact book. Washington, DC: Author.
U.S. Office of Personnel Management. (2004). The fact book. Washington, DC: Author.
Vandenberg, R. J., & Nelson, J. B. (1999). Disaggregating the motives underlying turnover intentions:
When do intentions predict turnover behavior? Human Relations, 52(10), 1313-1336.
Waldfogel, J., Higuchi, Y., & Abe, M. (1999). Family leave policies and women’s retention after childbirth:
Evidence from the United States, Britain, and Japan. Journal of Population Economics, 12, 523-545.
Wise, L., & Tschirhart, M. (2000.) Examining empirical evidence on diversity effects: How useful is
diversity research for public-sector managers? Public Administration Review, 60(5), 386-394.
Wright, B. E., & Davis, B. S. (2002). Job satisfaction in the public sector: The role of the work environ-
ment. American Review of Public Administration, 33(1), 70-90.
Zax, J. S. (1988). Quits and race. The Journal of Human Resources, 24(3), 469-493.

Donald Moynihan (dmoynihan@Lafollette.wisc.edu) is an associate professor of public affairs at the

University of Wisconsin–Madison. He is the author of the book The Dynamics of Performance
Management: Constructing Information and Reform, and has published articles in Public Administration
Review, Journal of Public Administration Research and Theory, and Governance.

Noel Landuyt (nlanduyt@austin.utexas.edu) is a lecturer and research associate in the School of Social
Work at the University of Texas at Austin. He is the project director for the Organizational Excellence
Group in the Center for Social Work Research.

Downloaded from rop.sagepub.com at COLUMBIA UNIV on February 4, 2015