Академический Документы
Профессиональный Документы
Культура Документы
Justin Marlowe
Sanjay K. Pandey
University of Kansas
Leisha DeHart-Davis is an assistant Feminist scholars of public administration have critiqued women, revolve around family). Leadership is cultur-
professor of public administration at the the dominance of masculine imagery in public admin- ally masculine not only in objective terms (more
University of Kansas. Her research interests
include the gender dimensions of public
istration theory and practice. However, public service leaders are men) but also in subjective terms (leaders
administration and the effects of motivation is one area of public administration discourse should look male and pale). Public virtue is portrayed
organizational structure and reforms on that contains both feminine and masculine imagery. in decidedly culturally masculine overtones, in terms
public employees.
E-mail: lddavis@ku.edu.
Focusing on Perry’s multidimensional public service of paternalistic guardians, autonomous fame seekers,
motivation scale, the authors borrow from a range of commanding heroes (not heroines), and citizens who
Justin Marlowe is an assistant social science literatures to contend that compassion is a have historically been identified as white and male.
professor of public administration at the
University of Kansas. His research interests
feminine dimension of public service motivation, whereas
include public budgeting and financial attraction to policy making and commitment to public Duerst-Lahti and Johnson suggest that public organi-
management, municipal management, and service are masculine dimensions. Data from a survey of
research methods.
zations may actually be more culturally masculine
E-mail: jmarlowe@ku.edu.
public managers in state health and human service agen- than other organizational types given the influence of
cies reveal that women score higher on Perry’s compassion the “manly” Weberian bureaucracy and its undervalu-
Sanjay K. Pandey is an assistant subscale but also on attraction to policy making. No
professor of public administration at the
ation of the culturally feminine. Though the critique
University of Kansas. He conducts research
statistically significant gender differences were found on of public administration practice and theory is both
on public management and health policy. commitment to public service. well founded and well articulated, recent public ad-
Recently, he directed Phase II of the
National Administrative Studies Project.
ministration scholarship has become more mindful of
F
E-mail: skpandey@ku.edu. eminist scholars of public administration have gender (see, e.g., Meier, O’Toole, and Goerdel 2006
uncovered and critiqued the dominance of on management functions and outcomes; Rubin and
culturally masculine imagery in public adminis- Bartle 2005 on budgetary impacts; and Wilkins and
tration theory and practice. In particular, expertise, Keiser 2006 on active and passive representation, to
virtue, leadership, and bureaucracy—lauded charac- name a few).
teristics of public administration—are based on male
perspectives on the world (Duerst-Lahti and Johnson Continuing in this vein, the study reported here as-
1990, 1992; Ferguson 1984; Stivers 2002). In Fergu- serts that public service motivation is one area of
son’s critique of bureaucracy, for example, public public administration discourse that contains both
administration discourse emphasizes the control and culturally feminine and culturally masculine imagery.
subordination of employees by upper management, Public service motivation can be defined as “an indi-
mirroring the control and subordination of women by vidual’s predisposition to respond to motives
men throughout history. This emphasis on control grounded primarily or uniquely in public institutions
and subordination leads lower-level bureaucrats to and organizations” (Perry and Wise 1990, 368). We
resort to “feminization” as a means of coping with argue that three motives—attraction to policy mak-
their secondary status, a strategy that includes defer- ing, compassion, and commitment to public
ring to superiors and conveying favorable images to interest—have distinct gender dimensions. Further-
them. more, we expect these gender dimensions to be
observable in comparisons of male and female public
From Stivers’s perspective, images of expertise, virtue, managers. Although some recent research has noted
and leadership are not only culturally masculine but gender differences in public service motivation, these
also deliberately unfeminine in their cultural con- efforts do not attempt an in-depth explanation of the
struction. Expertise based on science pits objectivity differences (Bright 2005; Moynihan and Pandey,
against emotion, culturally masculine autonomy forthcoming; Perry 1997).
against culturally feminine responsiveness, scientific
authority against lay perspectives, and professional Given assertions in the literature that public service
expectations against societal expectations (which, for motivation should increase organizational effectiveness
Gender Dimensions of Public Service Motivation 873
(Perry and Wise 1990; Rainey 1982; Romzek 1990), Borrowing from sociological research by Knoke and
lead to higher bureaucratic performance (Perry and Wright-Isak (1982), Perry and Wise delineated three
Wise 1990), and identify candidates for public service categories of motives: affective, rational, and norma-
(Moynihan and Pandey, forthcoming), it is important tive. Affective motives are driven by emotion, rational
to scrutinize the gender dimensions of the underlying motives by individual utility maximization, and
construct. Used in these ways, a purely culturally norm-based motives by the desire to fulfill societal
feminine or masculine conception of public service expectations (which, for public service, relates to
motivation risks generating incomplete and biased the desire to serve the common good). These opera-
assessments of organizations, current workers, and tional definitions clarified the public service motiva-
candidate employees. tion concept, which prior to that time had been
marked by vagueness and inconsistency of meaning
The data used to examine gender dimensions of the (Rainey 1982).
public service motivation construct were collected by
the National Administrative Studies Project during Following this research, Perry (1996) developed a
late fall 2002 and winter 2003. This study adminis- measurement scale for this new operational definition
tered a mail survey to managers in state health and of public service motivation. Scale development began
human service agencies nationwide. Approximately with the crafting of 35 survey items developed from a
half the respondents were women. The hypotheses literature review and focus groups with master of
were tested using an ordered logit model that included public administration candidates on conceptions of
a range of demographic variables as controls. public service. The 35 survey items were devised to
correspond to six dimensions of public service motiva-
The first section profiles the public service motivation tion: attraction to policy making, commitment to
concept, with particular attention to Perry’s (1996) public interest, civic duty, social justice, compassion,
conceptual and measurement scheme. The second and self-sacrifice. With the exception of self-sacrifice,
section, borrowing from a range of social science these individual motivations mapped onto the three
literature, outlines expectations for gender differences categories of motivations that Perry and Wise identi-
in different forms of public service motivation. The fied in 1990: attraction to policy making fell into the
third and fourth sections identify the data and mea- rational category; commitment to public interest,
sures used to test these hypotheses, and the fifth sec- civic duty, and social justice into the norm-based
tion describes the results. The final two sections category; and compassion into the affective category.
discuss the broader implications of these results. (Self-sacrifice was retained because of its presence in
the public administration literature.) These survey
Public Service Motivation items were included in a questionnaire that was ad-
Several authors have traced the history of public ser- ministered to small groups of students pursuing mas-
vice motivation as a scholarly topic, although they ter’s degrees in public administration and business.
vary in their emphasis on its intellectual precursors, Feedback from these groups triggered three revisions
such as those focusing on studies conducted during of the survey items (see Perry 1996 for details).
the 1960s to examine differences in reward prefer-
ences between public and private sector workers The final survey items were administered to a wide-
(Crewson 1997; Scott and Pandey 2005); those em- ranging sample (including master’s degree students,
phasizing Buchanan’s 1975 comparison of job involve- business executives, and state and county government
ment between the public and private sectors or employees) that yielded 376 responses. Confirmatory
Rainey’s 1982 examination of more direct indicators factor analysis of the resulting survey data detected
of the public service ethic (Brewer and Selden 1998; four dimensions of public service motivation: attrac-
Brewer, Selden, and Facer 2000); and those examining tion to policy making, commitment to public interest/
motivation scholarship in the social psychology and civic duty, compassion, and self-sacrifice. Interestingly,
sociology literature from the 1960s to the 1980s Perry found that commitment to public interest/civic
(Wise 2004). duty was highly correlated with self-sacrifice (suggest-
ing that it could be combined into one dimension)
Regardless of the starting point identified for public but that the four-dimension model provided a better
service motivation research, a turning point in this fit for the interdependence of the individual survey
line of inquiry occurred in 1990, when Perry and items than the three-dimension model.
Wise proposed a theoretical definition of public ser-
vice motivation. The two scholars defined public The Perry PSM scale is important to public adminis-
service motivation as “an individual’s predisposition to tration scholarship for several reasons. First, it repre-
respond to motives grounded primarily or uniquely in sents the most methodologically sophisticated
public institutions and organizations,” with a motive development of the public service motivation con-
defined as a “psychological deficiency or need that an struct, and it is a vast improvement over previous
individual feels some compulsion to eliminate.” indirect measures (Brewer and Selden 1998; Brewer,
874 Public Administration Review • November | December 2006
Selden, and Facer 2000). Second, it has been used for historical limitations on women’s participation in the
practical purposes, such as assessing attitudinal public sphere, along with the continuing tensions
changes among President Bill Clinton’s AmeriCorps between public and private loyalties experienced by
program participants (Perry 1996) and measuring modern women, we expect commitment to public
public service motivation in the 1996 Merit Principles service to be a culturally masculine trait. We will
Survey of the U.S. Merit Systems Protection Board. consider these issues in greater detail below.
Scholars have also used the scale to examine the rela-
tionship between public service motivation and indi- Attraction to Policy Making
vidual bureaucratic performance (Alonso and Lewis Attraction to policy making is a public service motive
2001), individual conceptions of public service moti- based on the desire to satisfy personal needs while
vation (Brewer, Selden, and Facer 2000), perceptions serving the public interest (Perry and Wise 1990).
of red tape (Scott and Pandey 2005), and organiza- Categorized as a rational public service motive because
tional effectiveness (Kim 2005), as well as the indi- it involves maximizing individual utility, policy mak-
vidual and organization determinants of public service ing appeals to those who seek a sense of achievement
motivation (Moynihan and Pandey, forthcoming). and self-importance (Wise 2000). Policy processes
provide opportunities to satisfy both desires through
Given the increasing use of the Perry PSM scale, it is competition for preferred outcomes that produces
important to explore its gender dimensions. We will winners and losers (Bardach 1977).1
now turn to a broad array of literature that underlies
expectations for gender differences in the PSM scale. The game-like nature of policy processes suggests that
attraction to policy making is a culturally masculine
Public Service Motivation and Gender construct, based on studies of masculinity that indi-
At the outset, the categorical drivers of public service cate an emphasis on self and individuality, as well as
motivation—rationality, emotion, and norms—are the pursuit of self-interest and competition (Gilligan
themselves laden with culturally constructed meanings 1982; Maccoby 1990; McGuffey and Rich 1999;
and understandings that pertain to gender. Rational Messner 2000). These attributes are reflected in boys’
motives are goal oriented and assumed to be a function games, which emphasize the quest for domination and
of individual choice based on a consideration of individual achievement (Hasbrook and Harris 1999;
potential gains and losses. This approach is problematic Messner 2000; Moller, Hymel, and Rubin 1992), as
from a gender perspective because women have well as men’s leadership styles, which are characterized
historically operated within narrower physical and by rationality, competitiveness, and the goal of win-
social boundaries than men, allowing them to consider ning (Loden 1985, 121; Rosener 1995, 73). Lips
and pursue fewer choices and goals (Kelly and Boutilier (1978) notes that the desire to achieve does not ap-
1978). Emotion has long been considered the strength pear to differ for males and females, but the motiva-
and weakness of femininity, appropriate for the tion to achieve is aroused under different conditions:
perpetuation of hearth and home but an inhibition to Men are motivated by instructions that stress compe-
clear thinking (Blum 1982) and an unacknowledged tition and leadership, whereas women are motivated
and undervalued skill in the marketplace (Guy and by instructions that stress social skills.
Newman 2004). As for societal norms, they differ for
women and men based on the historical relegation of Conversely, studies of culturally feminine attributes
women to the private sphere and their corresponding portray a struggle to establish identity, a tendency
exclusion from public life (Stivers 2002; Thomas toward collaboration over competition, and an avoid-
1994). ance of games with clear winners and losers in favor of
win–win scenarios (Beutel and Marini 1995;
We hypothesize that there are gender dimensions to Chodorow 1974; Gilligan 1982; Maccoby 1998;
three specific public service motivations within each Moller, Hymel, and Rubin 1992). Exemplifying the
of these categories: attraction to policy making, a latter attribute, Kohlberg (1969) noted girls’ prefer-
rational motive; compassion, an affective motive; and ences for games involving indirect competition, such
commitment to public interest, a normative motive. as jump rope or hopscotch. Lever (1976) concluded
Both feminist theorists (e.g., Gilligan 1982; Mumby that girls avoid conflict in game playing, preferring to
and Putnam 1992; Ruddick 1989) and “naïve theo- terminate the game when arguments break out. By
rists,” to use Fritz Heider’s coinage for laypersons, contrast, boys appeal to rules when disputes arise,
concur that reason or rationality is culturally con- resulting in short disruptions of game playing. A more
strued as a masculine concept, just as affective motives recent study of children’s play noted that girls tend to
are culturally construed as feminine. Although there is play in smaller same-sex groups in which players can
no overwhelming support for the gender dimensions be sensitive to their partners’ needs, in contrast with
of normative motives being either male or female, the boys, who tend to play in larger groups in which
concept of commitment to public interest makes conflict and competition dominate (Fabes, Martin,
explicit reference to the public sphere. Given the and Hanish 2003). Again, Loden (1985) argues that
Gender Dimensions of Public Service Motivation 875
women’s leadership styles follow the patterns exhibited Compassion
in children’s games, with female managers favoring Compassion is a public service motive that entails love
cooperation and empowerment over competition and and concern for others and a desire that others be
winning. Rosener (1990) and Helgesen (1995) make protected (Perry and Wise 1990). Compassion is
similar arguments, focusing on the tendency among categorized as an affective motive that involves re-
female leaders to favor empowerment and shared sponding emotionally to humankind (Frederickson
power, a finding that has received substantiation in a and Hart 1985). It is also construed as a culturally
meta-analysis of the scholarly literature (Van Engen feminine quality (Stivers 2002, 58), underscored by
and Willemsen 2004). the tendency for women to define themselves in rela-
tion to others more so than men (Chodorow 1974)
To illustrate how these attributes surface in a political and to make moral decisions based on the impact of
context, Thomas notes that the influx of women into those actions on others (Gilligan 1982). There is
the political arena during the 1970s was expected to scholarly disagreement as to whether culturally femi-
engender more cooperative methods of doing business nine compassion is the result of nature, nurture, or a
than the “cutthroat competition and battles for domi- combination of both (Chodorow 1974).2
nance” that were “the norm” (1994, 6). That such
norms did not change led some women to withdraw The notion of compassion as a culturally feminine
from political processes completely. trait is implied by historical and behavioral studies of
gender differences in female and male public adminis-
The historical exclusion of women from policy mak- trators. Women pioneered social policies and causes
ing (Stivers 2000) provides another source of expecta- during the late 19th and early 20th centuries in the
tion for gender differences in attraction to policy United States, a time when charitable work signified
making. This exclusion has resulted in the continuing the virtuous tendencies of women (Stivers 2000). The
domination of men in politics, leading to the message dominance of women in “caring” agencies continues
that politics is about men and for men (Elder 2004) to this day, a fact that is supported by the overrepre-
and that women are emotionally unsuitable for poli- sentation of women in public welfare, housing, and
tics (Fox and Lawless 2003). Thus, current political health agencies (Lewis and Nice 1994); in employ-
processes may seem alien to potential female partici- ment security, human resources, and civil rights agen-
pants, possibly explaining the low percentage of cies (Riccucci and Saidel 1997); and in positions that
women who run for elected office or seek higher-level require “emotional labor,” that is, work that requires
appointed positions (Elder 2004; Fox and Lawless responsiveness and a caring attitude (Guy and
2003). Though public managers are not necessarily in Newman 2004). Women are also well represented and
the thick of political competition, public organiza- sometimes overrepresented within federal, state, and
tions are not insulated from politics, and performance local redistributive agencies, which tend to involve
in top-management roles requires considerable politi- caring functions (Kerr, Miller, and Reid 2002;
cal acumen. Newman 1994; Reid, Kerr, and Miller 2000; Reid,
Miller, and Kerr 2004). These behavioral patterns lead
Fox and Schuhmann (1999) provide one of the few us to expect that
studies to illuminate gender differences in attraction to
policy processes in a public administration context. In H2: Female public managers will indicate a
interviews with city managers, these scholars found higher compassion motive than their male
that females and males similarly proclaimed a love of counterparts.
politics and a commitment to a specific policy area.
However, the female city managers were less likely It should be noted that compassion’s inclusion as
than their male counterparts to consider themselves a public service motive is significant in light of its
policy entrepreneurs; more likely than their male deliberate omission from Weber’s “ideal” bureaucracy,
counterparts to see city management as oriented to- as well as from the virtues listed by such moral
ward administration rather than public policy formu- philosophers as Kant and Hegel (Blum 1982).3
lation; and more likely than men to say that city Compassion has been shown to be valued equally by
managers should remain neutral on controversial city male and female public managers as an important trait
issues. These observations led Fox and Schumann to of bureaucracy (Duerst-Lahti and Johnston 1992,
conclude that female city managers are less likely to 144). Thus, the inclusion of compassion as a motive
perceive themselves as policy entrepreneurs and more for public service indicates a decisive shift in the way
likely to perceive themselves as managers or facilitators public service is conceived by both scholars and
than their male counterparts. Therefore, we expect that practitioners.
H1: Male public managers will be more at- Commitment to Public Service
tracted to policy making than their female Commitment to public service is based on one’s desire
counterparts. to fulfill a societal obligation or standard and thus is
876 Public Administration Review • November | December 2006
categorized as a norm-based motive (Perry 1997). In their historical participation in civic issues that have
particular, commitment to public service entails loy- shaped the construction of the public administration
alty to duty and country, as well as a desire to serve field (Stivers 2000, 8). To do so, however, would
the public interest (Perry and Wise 1990). In 1997, generate a false rival hypothesis: Our contention is not
Perry found that men scored higher than women on that women are less committed to public service but
the public interest construct but offered no explana- that declaring commitment to the public sphere is
tion as to why. Perhaps one explanation is that a more problematic for women, who face greater com-
norm-based public service motivation is problematic peting loyalties—and feel more ambivalence about
for gender when one considers that society holds those competing loyalties—than men (Hattery 2001;
significantly different expectations for women and Maushart 1999). The competing loyalties argument
men. These differing expectations pertain to the sepa- has been used to examine the gender aspects of civic
ration of life into the public and private spheres, with engagement: The disproportionate amount of unpaid
women being relegated to the latter for most of his- “care work” performed by women, ranging from
tory (Blum 1982; Ferguson 1984; Stivers 2002). housework to child care to elder care, interferes with
Confinement to the private sphere has meant that their ability to claim full citizenship, justice, and
women historically have had responsibility only for political participation (Lister 2003, 103; Okin 1989;
persons with whom they have emotional relationships; Phillips 1991)—that is, until unpaid care counts as a
by contrast, the execution of Weberian duties and form of civic engagement (Herd and Meyer 2002).
obligations requires one to execute duties independent
of one’s feelings about those duties or the person for Data and Methodology
whom they are being performed (Blum 1982). Thus, The data for this project were collected during Phase
commitment to public service is understood as a II of the National Administrative Studies Project
culturally masculine construct because it implies a (NASP-II). The sampling frame consisted of managers
single-minded pursuit of life in the public realm working in information management activities at the
(largely male dominated), to the exclusion of life in state level in health and human service agencies.
the private realm (largely female dominated). Primary human service agencies were identified
Feminist theorists have made a similar assertion, according to the definition used by the American
noting the culturally masculine overtones of citizen- Public Human Services Association (APHSA) and
ship as it is traditionally construed (Lister 2003; included agencies housing programs related to
Stivers 2002). Medicaid, Temporary Assistance to Needy Families,
and child welfare. Information management was
Though the argument may sound antiquated, women broadly defined to include a range of key managerial
still bear the lion’s share of housework and caregiving roles, such as the top program administrator, manag-
(Bianchi et al. 2000; Cancian and Oliker 2000; ers of information system applications, managers in
Coltrane 2000; Kroska 2004; Lee and Waite 2005; charge of evaluation and research, and managers deal-
Voydanoff and Donnelly 1999). And even though ing with public information and communication. The
both women and men now participate in the public sampling frame was developed with the aid of the
and private realms, evidence indicates that for women 2001 APHSA directory, the most widely used direc-
(more so than men), time invested in one sphere is tory of human service agency managers. Application
spent at the expense of the other sphere (Blair-Loy of the study criteria resulted in a sampling frame of
2001; Mennino, Rubin, and Brayfield 2005; Sanchez 570 managers from the 50 states and Washington,
and Thompson 1997). So although women have made D.C. Given the small size of the sampling frame, a
great strides in participating in both realms of life and decision was made to administer the survey to the
gender attitudes have become more egalitarian over entire sampling frame (i.e., conduct a census).
time (Rice and Coats 1995; Kozimer-King and Leicht
1999; Mason and Lu 1988), women are still less likely The survey-implementation process sought to mini-
than men to be able to devote themselves to public mize nonresponse to both the survey and specific
life alone (Stivers 2002, 93). Accordingly, we expect questionnaire items. Thus, the study employed
that Dillman’s (2000) comprehensive
tailored design method to maxi-
H3: Public managers who are One could argue that women mize the response rate. This
women will be less likely to should register a higher com- approach includes (1) a question-
indicate a commitment to
mitment to public service given naire with well-designed content;
public service than public (2) a survey questionnaire that is
managers who are men. their historical participation in formatted in accordance with the
civic issues that have shaped the latest thinking in cognitive re-
One could argue that women construction of the public search; (3) multiple personalized
should register a higher commit- administration field. contacts, each contact including
ment to public service given a carefully crafted message to
Gender Dimensions of Public Service Motivation 877
encourage the respondent to complete the survey tent with our expectations for a typical career manager
questionnaire; and (4) the use of real stamps on return in a health and human service agency, not the stereo-
envelopes and such features as a pre-notice letter, fax typical image of information technology staff as peripa-
message, and phone call at key points in the survey tetic young males who frequently change employment.
administration, as well as the use of special delivery (a
combination of two-day delivery by Airborne Express Of the 274 respondents, 114 had line responsibilities
and U.S. Postal Service Priority Mail). for managing programs, and the rest performed staff
functions. It should be noted, however, that even the
The data-collection phase of the study began in fall managers performing staff functions were senior man-
2002. First, sample members were sent an alert letter agers operating in a large agency; they were respon-
informing them about the study and requesting their sible for providing both operational and strategic
cooperation in completing a questionnaire to be support on an ongoing basis for a wide range of orga-
mailed later. Approximately a week after the initial nizational operations.
alert letter, the survey questionnaire was mailed to the
sample. The cover letter accompanying the survey Public Service Motivation Measures
questionnaire outlined the study objectives, indicated Public service motivation was measured based on
the voluntary nature of the study, requested participa- scales of survey items designed by Perry (1996). Three
tion, and provided contact details for the project types of public service motivation served as dependent
director for further informational needs and clarifica- variables: attraction to policy making, compassion,
tion. About 10 days later, a combination thank you/ and commitment to public interest.
reminder postcard was sent to all respondents, thank-
ing those who had responded and encouraging those Attraction to policy making was measured as the reverse
who had not to respond as soon as they possibly sum of the survey participant’s level of agreement with
could. Nearly a month after mailing this postcard, a three questions, all of which took the form of a
new cover letter and replacement survey were sent to Likert-type scale ranging from “strongly disagree”
nonrespondents. The cover letter emphasized the fact (1) to “strongly agree” (5):
that it was important for everyone to respond (unless
for some reason or other the respondent chose not to • “Politics is a dirty word.”
respond). To ensure that target participants were aware • “The give and take of public policy making does
of the second mailing, we also faxed the second mail- not appeal to me.”
ing’s cover letter to nonrespondents. The final step in • “I don’t care much for politicians.”
survey administration took place about two months
The Cronbach’s alpha, which measures scale reliability
later, when nonrespondents were sent a new cover
from 0 to 1, was 0.72 for this scale. This score com-
letter and a second replacement survey with a request
pares well with Perry’s (1997) 0.77 correlation among
to complete the survey. This final mailing pointed out
the same survey items.
that it was the final opportunity to complete the sur-
vey questionnaire and used a combination of two-day
Compassion was measured as the sum of the survey
delivery by an express carrier and U.S. Postal Service
participant’s level of agreement with three of Perry’s
Priority Mail. By the time survey administration con-
eight original Likert-type items measuring compassion
cluded in winter 2003, a total of 274 responses had
(1 = strongly disagree to 5 = strongly agree):
been received, a 53 percent response rate.
on individual attitudes toward these particular public which each independent variable was held at its mean
service motivations. value and gender was assigned a value of either 0 or 1.
As table 3 shows, the general trend for compassion is
Education and professionalism, by contrast, are that males are more likely to report lower levels of
important predictors of public service motivation. compassion, whereas females are more likely to report
The magnitude of those effects can be better under- higher levels of compassion. Those effects were stron-
stood by examining how changes in the independent gest at both the “weak moderate” level, at which males
variables affected respondents’ predicted scores on were 8 percent more likely to report, and the “strong
each of the public service motivation scales. Those moderate” level,” at which females were 7 percent
effects are presented in table 2, which reports the more likely to report.
effect of an increase in each independent variable
from its mean to 1 standard deviation above its mean There is also evidence of gender-based differences in
on the odds that a respondent will score higher on attraction to policy making, although that evidence
each of the public service motivation scales. Odds contradicts our expectations and is not as strong as the
above 1 suggest that a higher motivation score is findings for compassion. In this case, the regression
more likely, and odds below 1 suggest that a stronger coefficient shows that females were 28 percent more
motivation score is less likely. In the model of attrac- likely to report a stronger attraction to policy making
tion to policy making, for example, an increase in than men, but that relationship was only marginally
the professionalism index from just over 1 to just significant at the .052 level. In terms of predicted
over 3 increased the likelihood of a respondent re- probabilities, females were only 3 percent more likely
porting a stronger attraction to policy making by 29 to strongly agree that they were attracted to policy
percent. For the compassion model, that change was making, and males were only 1 percent to 2 percent
even stronger—34 percent. By comparison, an in-
crease in education from its average of 1.38 to its
Table 3 Predicted Probabilities of Public Service Motivation
highest value of 2 resulted in a 9 percent increase in Outcomes by Gender (Percent)
the likelihood of a respondent reporting a stronger
Males Females
attraction to policy making. Somewhat surprisingly,
that same increase in education decreased the likeli- Attraction to policy making
hood of a stronger compassion motivation by ap- Weak 14 12
Weak moderate 13 12
proximately 6 percent (or 1 minus the reported
Moderate 15 15
change in odds of 0.94).6 Strong moderate 9 11
Strong 12 15
The results offer mixed support for our hypotheses. A Very strong 6 8
respondent’s gender was shown to be a significant Compassion
predictor of his or her reported levels of both attrac- Weak 5 2
tion to policy making and compassion. The results Weak moderate 19 11
Moderate 19 14
were strongest for compassion, where gender was
Strong moderate 21 21
statistically significant, and its coefficient indicates Strong 19 26
women were 57 percent more likely than men to score Very strong 8 13
higher on the compassion index. These results are
Note: Weak scores are 1 standard deviation below the mean;
further illustrated in table 3, which compares the weak moderate scores are 0.5 standard deviation below the
predicted probabilities of each compassion outcome mean; moderate scores are at the mean; strong moderate
by gender. These probabilities were calculated by scores are 0.5 standard deviation above the mean; and strong
examining the ordered logit model predictions in scores are 1 standard deviation above the mean.