Does Public Service Motivation Matter in Volunteering Domain Choices?

Review of Public Personnel Administration http://rop.sagepub.com/

Does Public Service Motivation Matter in Volunteering Domain Choices? A Test of Functional Theory David Coursey, Jeffrey L. Brudney, Laura Littlepage and James L. Perry Review of Public Personnel Administration 2011 31: 48 DOI: 10.1177/0734371X10394405 The online version of this article can be found at: http://rop.sagepub.com/content/31/1/48

Published by: http://www.sagepublications.com

On behalf of:

Section on Personnel Administration and Labor Relations of the American Society for Public Administration

Additional services and information for Review of Public Personnel Administration can be found at: Email Alerts: http://rop.sagepub.com/cgi/alerts Subscriptions: http://rop.sagepub.com/subscriptions Reprints: http://www.sagepub.com/journalsReprints.nav Permissions: http://www.sagepub.com/journalsPermissions.nav Citations: http://rop.sagepub.com/content/31/1/48.refs.html

Downloaded from rop.sagepub.com at INDIANA UNIV on May 16, 2012

>> Version of Record - Jan 31, 2011 What is This?


ROP39440 5 ROPPA31110.1177/0734371X10394405Coursey et al.Review of Public Personnel Administration

Does Public Service Motivation Matter in Volunteering Domain Choices? A Test of Functional Theory

Review of Public Personnel Administration 31(1) 48–66 © 2011 SAGE Publications Reprints and permission: http://www. sagepub.com/journalsPermissions.nav DOI: 10.1177/0734371X10394405 http://roppa.sagepub.com

David Coursey1, Jeffrey L. Brudney2, Laura Littlepage3, and James L. Perry4

Abstract Functional theory suggests that people choose activities based on their perception of how well the work matches their personal motives.This process implies that worker motivations vary by activity even when controlling for typical motivational antecedents. Although this perspective is common in the volunteering literature, the public service motivation (PSM) literature has not formally considered functional theory hypotheses.Yet PSM theory asserts a relationship not to government work itself, but to activities related to public service such as volunteering.This study evaluates functional theory in regard to PSM based on using a survey of recipients of the Daily Point of Light Award (DPOL) and the President’s Community Volunteer Award (PCV). After controlling for common PSM antecedents such as gender, level of education, religious participation, and age, PSM and its subdimensions exhibit some variance across volunteering domains (viz., religious, educational/school, human services, and other), though most differences involve religious organizations. Keywords volunteering, public service motivation, functional theory, religious volunteering 1

Arizona State University Cleveland State University 3 Center for Urban Policy and the Environment School of Public and Environmental Affairs, Indiana University–Purdue University 4 Indiana University 2

Corresponding Author: David Coursey, Visiting Scholar, Decision Theater, Arizona State University, 21 E 6th Street, Suite 126A, Tempe, AZ 85287 Email: [email protected]


49

Coursey et al.

Since Perry and Wise (1990) defined public service motivation (PSM) two decades ago, a substantial body of research has evaluated their proposition that workers with higher PSM will be more attracted to public organizations (e.g., Perry, Hondeghem, & Wise, 2009). Studies testing relationships between antecedent variables (such as sociodemographic, social institutions, organizational) and dependent variables (including reward preferences, organizational commitment, and job satisfaction) demonstrate mostly supportive results (e.g., Pandey & Stazyk, 2008; Perry et al., 2009). Nevertheless, existing research on PSM overlooks two significant theoretical concerns. First, research focuses far more on formal organizations by legal status than originally intended by Perry and Wise (1990). To Perry and Wise, the thesis is not that people with higher PSM are attracted to public organizations per se but that the typical nature of this work is expected to conform to those values. People with higher PSM are more attracted to altruistic public service opportunities and activities regardless of the formal organization or sector. Second, much of PSM theory rests on the attraction-selection paradigm, which postulates that workers are attracted to organizations providing opportunities to fulfill their motivations. Both of these issues raise the importance of mission and work domain as well as work outside formal government agencies. If attraction-selection theory holds, potential workers select employment based on their perception of how much the job and the organization allow them to fulfill their motivational needs. An organization’s work domain and mission are key signals to a potential employee, yet rarely is domain considered in PSM research. This oversight may arise, in part, due to the presumption that PSM is more influential for those already employed because entry and exit decisions are based more on economic needs and limited employment choices. Hence, recent research predominately examines an individual’s perception of organizational fit via PSM to outcomes such as job satisfaction and performance (e.g., Bright, 2007; Perry & Thomson, 2004; Wright & Pandey, 2008). One solution to these issues is examining volunteers’ selection of volunteer opportunities. First, volunteering is altruistic or prosocial behavior yet has been neglected in PSM studies. Second, volunteers are expected to be more motivated by altruism. Third, unlike paid employment, volunteers have far greater freedom of choice (Wuthnow, 1995). Fourth, volunteers who find that an organization does not fulfill their motivational needs can more easily exit than employees (see Wilson & Musick, 1999). In sum, attraction-selection via intrinsic rewards and motivations should be more powerful for volunteers (Knoke & Wood, 1981; Wilson & Musick, 1999). Indeed, volunteer recruitment research examines how organizations market themselves by promoting their work and mission as meeting presumed motivations of their target volunteer population (e.g., Stukas, Worth, Clary, & Snyder, 2009)—a level of practical application not yet achieved by PSM researchers. Volunteering and public work are not as distinctive as they may appear. Perry and Wise’s (1990) proposition that individuals with high PSM will be more attracted to public work relates not to the organization per se but to the public nature of the work (Perry et al., 2009). The nature of public work, according to Perry and Wise, is grounded


50

Review of Public Personnel Administration 31(1)

in altruism. As such, volunteering activities are arguably more “public” than government employment. Volunteerism scholars argue that volunteering ranges along several dimensions (including level of remuneration) and that government work can be considered a form of volunteering since public sector employees are willing to accept lower financial rewards than in the private sector (e.g., Cnaan, Handy, & Wadsworth, 1996; Francois, 2000; Smith, 1982). In addition, these employees are often involved in producing “public” goods and may be more likely motivated to help others). By any reasonable definition, volunteering is associated and accounted for by similar theories and values as public work (see Musick & Wilson, 2008, for an extensive review). Despite these similarities, PSM research rarely considers volunteering and its related literature as we do here. This oversight is detrimental since many of the same theoretical questions and issues apply to both PSM and a related concept, volunteer motivation. In this study we examine variation in PSM among elite volunteers devoted to specific organizational domains (i.e., religious, education/school, and human services) primarily within the context of volunteer research and theory. This study examines whether the variance in the typical opportunities and nature of volunteer work in those domains is related to PSM values.

The Role of Work Domain in PSM Though mostly supportive of the hypothesis that PSM will be higher among public employees, research findings are mixed (e.g., Pandey & Stazyk, 2008; Perry et al., 2009). Partly, these differences may be attributable to the varying work domains sampled (e.g., human services, criminal justice, financial/revenue, etc.). For example, according to functional theory we might expect public work that appears less public service oriented or altruistic to generate lower PSM scores so that the differences may be a function of the work domain more than formal sector. Researchers have been applying person–environment fit theories toward more complex models (e.g., Perry et al., 2009). In short, these theories (for a review, see Coursey & Pandey, 2007; Perry & Vandenabeele, 2008) postulate that how an individual fits with respect to his or her goals, motives, and other criteria with an organization’s culture and opportunities will affect workplace behavior. Most of these studies focus on employees’ perceptions of how well their PSM values fit those of the organization toward various outcomes, such as job satisfaction and performance. Missing, however, is a rather obvious extension: If fit matters, individuals can also be expected to consider how well their motivations fit that of a potential volunteer agency or employer. For example, military recruitment studies evaluating marketing effectiveness generally find that tailoring messages to the needs and values of recruits is important (e.g., Hanssens & Levien, 1983; Lovell & Morey, 1991). Military recruitment research has also found that the armed services are not viewed generically so that recruitment campaigns for one branch (e.g., army) are more effective than general, across-branch efforts (Brockett, Cooper, Kumbhakar, & Kwinn, 2004).


51

Coursey et al.

The volunteer literature is also influenced by person–environment fit perspectives through functional-fit theory. Functional theory argues that people act, or volunteer, with the expectation that the activity fulfills some psychological function (e.g., motivation; Snyder, Clary, & Stukas, 2000). This premise suggests that people may volunteer for the same activity, but for quite different reasons based on their motivations and perceptions of work opportunities. By contrast to research in public administration, volunteer research concentrates far more on fit as it relates to attraction and selection: How does fit affect the decision to volunteer, including where, how much, and how long? Volunteer researchers use a scaled motivation construct very similar to PSM, the Volunteer Functions Inventory (VFI). Much of this research tests relationships between VFI and choices among volunteer activities and host organizations. For example, do volunteers vary their choice in where (and how much) they volunteer among the wide variety of volunteer organizations such as religious, educational, civic, and so forth based on their perception of how much the activities match their motivations? Research supports that volunteers do vary choices according to the perceived matching (Clary et al., 1998; O’Brien et al., 2000; Stukas et al., 2009). Within public administration, Steinhaus and Perry (1996) found that the industry in which an organization operates is a better predictor of organizational commitment (a construct that should be related to person–environment fit) than broad public/private categorizations. Functional-fit theory argues that individuals choose organizations based on how well they think the agency provides opportunities related to their motivations. But how does a potential recruit evaluate such a fit? Some recruits are influenced in their choices by their social networks, which also tend to share their values (Clary et al., 1998; Stukas et al., 2009). Otherwise, recruits have only general impressions of the types of opportunities and work available to make their choices. The domain of volunteering is a reasonable indicator. Domain is related to institutional networks as well. A volunteer motivated by compassion for the poor will be more attracted to social service groups. A volunteer motivated by opportunities to influence a public policy area will prefer civic and political action groups. Tailoring recruitment messages to individuals whose motives fit the agency’s mission has practical value (e.g., Snyder, Omoto, & Lindsay, 2004; Stukas et al., 2009): Laboratory and field studies have found recruitment messages tailored to individual motivations are more influential than those that are not (Clary et al., 1998). Surprisingly, though, research testing domain differences is rather limited even within the volunteer literature. In a Gallup Organization in-home sample of 2,775 American adults, Sundeen (1992) found 6 personal goal orientations that linked to specific types of volunteer activity. A laboratory experiment involving 535 university psychology students found recruitment messages matching personal motivations were more persuasive (Clary, Snyder, & Stukas, 1996). Clary and colleagues also found in a field study of 467 Minneapolis/St. Paul volunteers that the extent to which their experiences matched their motivations predicted satisfaction and future volunteering intentions—a result replicated in further studies (e.g., Stukas et al., 2009). Using the Independent Sector Giving


52


and Volunteering in the United States (1990-1999) surveys, Musick and Wilson (2008) demonstrated that volunteer motivations for religious volunteering varied from other domains (viz., education, human services, advocacy, and youth development), controlling for various demographical variables (viz., family status, education, income, race, age, gender, and religious differences). Unfortunately, aside from religious volunteering, they did not test any other paired comparisons of functional domains. The few studies that do analyze domain differences offer little theoretical explanation and are primarily conducted post hoc with regard to expected differences and associated rationale. Nonetheless, domain differences, especially controlling for demographic variables, are noteworthy and call for further delineation. One such explanation is grounded in institutional differences, particularly support and sense of community. Individuals within social networks tend to share values. Organizations that foster a greater sense of belonging and family through social networks (Wilson, 2000), thereby reinforcing core ideals, tend to have more distinctive volunteer values (e.g., Wilson & Musick, 1999; see also Wuthnow, 1991). This trait characterizes religious organizations. Membership itself is strongly tied to volunteer activity far more than in other domains (Wilson & Musick, 1999). Hence, religious organizations offer clearer values with reinforcing traits that may better attract individuals with specific motivations. Another possibility is the level of formalization. Stukas and colleagues (2009) argued that more structured organizations offer less flexibility for individual behavior to be driven by personal motives (see Snyder & Ickes, 1985). Structural limitations also affect turnover. Individuals who remain with the organization are more likely to experience a better fit to the types of motivational fulfillment offered. Typically, religious organizations are far less structured in opportunities and activities than human service and educational organizations, which tend to rely more on paid or stipended workers and a formal workplace. Hence, prospective volunteers to religious organizations may view less formal structures as more appealing to fulfill their motivational needs. Motivational matching is thus stronger and, due to organizational flexibility, opportunities can be better tailored to individual motivation, leading to a more common set of values among volunteers with longer tenure. Although research testing domain differences is limited, religious organizations have been found to be the most distinctive and to foster the greatest commitment, as would be predicted by these two explanations (e.g., Musick & Wilson, 2008; Wilson & Musick, 1999). Related to formalization is the level of person-to-client activity. First, such work, controlling for other variables, can be viewed as more rewarding as it is usually easier for volunteers to see the results of their efforts as related to their motivations. Personto-client activity is characteristic of human service volunteering, and, hence, compassion, and possibly self-sacrifice PSM dimensions, would be higher even in comparison to religious organizations. Musick and Wilson found such an ordering for compassion in their analysis of the Independent Sector data (1990-1999). Individuals do not volunteer only to satisfy intrinsic motivational factors. Extrinsic personal rewards, such as career advancement and personal enhancement (dimensions


53

Coursey et al.

in the VFI scale), are related to volunteering domains as predicted by exchange theory (Smith, 1982). In addition, friendships and social networks encourage volunteering regardless of value systems (e.g., Musick & Wilson, 2008). Educational volunteering provides an example. Research shows that educational volunteering is related to lifecourse needs, particularly parenting (e.g., Sundeen, 1990). Parents certainly have a personal stake in their child’s education and, hence, are motivated by extrinsic interests (Wilson, 2000) independent of PSM, which is more strongly anchored in altruistic motivation. Civic and political activities and to a lesser extent, perhaps, educational volunteering, relate to a commitment to public involvement/interest/service, which is both extrinsic and intrinsic. Sundeen (1992) found that both educational and social service volunteering were related to public involvement and charity values and that religious volunteering was not associated with public involvement. Given the possible explanations for domain differences, what are the implications for PSM and its subdimensions? First, we would expect volunteers in religious organizations to score higher than those in education on the compassion and self-sacrifice dimensions of PSM. As a result of the typical person-to-client activity in human service organizations, volunteers in this domain should score higher than those in education and, to a lesser extent, those performing religious volunteering on the compassion and self-sacrifice dimensions of PSM (see Musick & Wilson, 2008). Lastly, the dimensions of commitment to public service and attraction to public policy making should be higher for volunteers in education and social service organizations. In this research we examine empirically whether mean scores for PSM and three of its subdimensions (compassion, self-sacrifice, and commitment to public service) vary across four types of volunteer organization domains: religious organizations, school/ education, human services, and “other.” Our data are from a national survey of elite volunteers. The data set includes four of the most commonly supported antecedents of PSM (age, level of education, gender, and religious participation) so that we can control for them in the analysis (e.g., Pandey & Stazyk, 2008). Our research question is whether PSM varies across organizational domains controlling for these variables. If so, an implication is that volunteers in those areas have different motivations possibly due to the self-selection predicted by functional and person–environment fit theories.

Research Design The empirical basis for this inquiry is a sample of volunteers who have been awarded high honors for their volunteer activity. The following sections describe our sample, variables, and analytical techniques.

Sample The sample for the empirical study consists of recipients of the Daily Point of Light (DPOL) award (DPOL) and the President’s Community Volunteer Award (PCV),


54


presented on behalf of the President of the United States.1 The DPOL award honors individuals and volunteer groups that have made a commitment through service to help meet critical needs in their communities. These nonpartisan awards are given to those who find innovative ways to meet community needs, efforts that often lead to long-term solutions and impact social problems in their local communities. Now called the President’s Community Volunteering Award, it is the nation’s highest honor for volunteerism. It recognizes outstanding individuals, families, groups, organizations, businesses, and labor unions engaged in community services that address unmet human service, educational, environmental, and public safety needs. Traditionally, the President of the United States presents these awards at a White House ceremony to signify and symbolize their importance. Given their level of commitment, these award winners are not typical volunteers. We concentrate on such a specific group and not volunteers in general (or even volunteers vs. nonvolunteers) for two main reasons. First, very few studies have been undertaken on such volunteers, and other research we have in progress focuses on whether these elite, highly committed volunteers differ in their characteristics and motivations compared with other samples. Second, based on functional theory, more committed volunteers have greater motivational needs. We might also expect larger differences across chosen domains. For example, highly dedicated volunteers working solely for religious organizations might choose to do so because they perceive an even greater match between their motivational needs and the organization’s activities than less committed volunteers.

Survey A survey instrument was developed that included a short version of Perry’s (1996) public service motivation scale and multiple sources for measures of religiosity, volunteerism, motivations to volunteer, and family influences. Because the primary purpose of the survey was to examine the role of faith in volunteering, we convened an advisory board of clergy to review and comment on the draft survey instrument. Before finalizing the survey, we incorporated the advisory group’s input and received approval for administration from the institutional review boards at our respective universities. The survey was designed for administration through U.S. postal mail. The first mailing of the surveys occurred in early 2004 with a follow-up mailing approximately 4 weeks later. The final response rate was 38% (525 surveys returned from a total of 1,386 possible). As could be expected with any survey, a few cases had missing data, which led us to eliminate from analysis any case with fewer than three of the PSM indicators present. Next, we used expectation maximization (e.g., McKnight, McKnight, Sidani, & Figueredo, 2007) to estimate values for missing data. About 8% of the cases had missing data on at least one variable, mostly income (a variable not used in this study). No significant differences were found between cases with and without missing data on the study variables (t test or Mann–Whitney U; p > .05). This


55

Coursey et al.

procedure yielded a sample of 511. Furthermore, in order to attribute any differences found to a specific functional area we selected cases for analysis only when a volunteer indicated a single work domain. Not surprisingly, given such active and elite volunteers, most respondents contributed time in several domains. Hence, this restriction reduced the final sample to 136 cases. Respondents were compared to all award winners to verify representativeness through two available indicators: percent of respondents by state and percent of respondents by award year. The comparison showed that respondents were widely distributed and not concentrated by state. As might be expected, a higher percentage of respondents came from more recent award years, although the percentage is not dramatically higher.

Variables Public service motivation and three of its subdimensions—compassion, commitment to public service, and self-sacrifice—are the dependent variables in the analysis. The fourth PSM subdimension, attraction to public policy making, was unfortunately not included in the survey because it was presumed to be a poor predictor of volunteer, nonprofit activity. It should be noted that the policy-making subdimension items as originally worded have often been the most problematic for measurement (see Coursey, Perry, Brudney, & Littlepage, 2008, a-b). Independent variables in the analysis include the volunteer domain as well as various controls for commonly asserted PSM antecedents. The appendix presents all survey items included in the analysis. Public service motivation. Public service motivation and its three measured subdimensions were calculated via a second-order reflective confirmatory factor analysis. This procedure generates values far more representative of the theoretical assumptions concerning subdimension relations and their associations with overall PSM. By contrast, simply adding scores on the subdimensions, or alternatively creating principal component factors (PC), so commonly used for scales such as PSM, are invalid approaches, for they presume a very different measurement model (e.g., formative). For example, PC would presume that PSM subdimensions are independent and that they are measured without error. PSM theory presumes the dimensions are related since they all reflect core altruistic behavior. For details on the analysis used for the PSM values and the related measurement issues, see Coursey and colleagues (2008, a-b) and Perry, Brudney, Coursey, and Littlepage (2008). Volunteer domain. Five commonly cited volunteer domains, religious organizations, political/civic, schools/educational, human services, and all others (such as arts) were defined in the survey (see Sundeen, 1992). The analysis is based on the 136 individuals who volunteered in a single domain (12 religious, 36 schools, 3 political/civic, 50 human resources, 38 other). Unfortunately, civic and political volunteering had to be eliminated from the analysis given the minute sample size, just three cases. Controls or PSM antecedents. Four PSM predictors are included in the analysis: level of education, gender, age, and religious participation. These variables were chosen


56


based on supporting research findings (see Pandey & Stazyk, 2008, for a thorough review) and because as background variables they would be less likely to have reciprocal relationships with PSM. Level of education is measured by the highest degree earned, with high school or less combined into one category. Age is calculated as the survey year minus birth year. Religious participation is measured in two ways: whether the respondent indicated being a member of a church, mosque, and so forth, and how many times per week she or he attends religious services. Separate analyses were conducted for each variable. Because the differences are marginal, only the results based on the attendance variable are reported here. We chose not to use religious activity (such as how many times a person engages in various church functions outside of religious services). Research on volunteers find that respondents, particularly those in evangelical faiths where most church social service activities are closely tied to missionary work or church membership alone, have difficulty separating volunteer work from activities specific to the church or organization (for an overview, see Musick & Wilson, 2008, pp. 281-282).

Analysis Analysis of Covariance (ANCOVA) was used to test for mean differences on PSM and its subdimensions across the volunteer domains controlling for the various antecedents. A few of the analyses displayed heterogeneity across the domains; these analyses were adjusted with an appropriate weighting by domain.2 Tables 1 to 4 provide the results for mean differences adjusted for the model covariates. Least significant difference tests are used to compare values.3 Covariates are not included as they are of no interest to the analyses other than as controls. Probability or p values up to .10 are judged statistically noteworthy given the small samples and low test powers (viz., chance of Type 2 error). Table 1 provides the findings for overall PSM. Volunteers for religious organizations have higher mean PSM values than volunteers in all other domains; no other statistically significant differences are found. The overall F test is statistically significant (p < .01) and accounted for variance uniquely attributable to domain is reasonable at 0.08 (partial Eta2). Test power is just at the typically used 0.80 cutoff to presume a Type 2 error is not operative (viz., falsely retaining hypothesis or presuming there are no differences when in fact they exist in the population). The results for the compassion subdimension appear in Table 2. Volunteers for religious organizations have higher mean compassion values than education and human services volunteers. Education and human services volunteers both score lower than those in the “other” category. The overall F test is marginal (p < .06), and variance uniquely attributable to domain is rather low (partial Eta2 = 0.040). However, test power is low (0.614), well below the typically used 0.80 cutoff. Table 3 presents the findings for the commitment subdimension. Religious organization volunteering ranks higher than human services and the “other” category, but not education/school volunteering. Overall domain significance is very poor though test


57

Coursey et al. Table 1. Estimated Marginal Mean Differences for PSM Across Volunteering Domains ANCOVA Results With Least Squares Difference Testsa Estimated marginal mean

School/education (significance)

Human services (significance)

3.64 3.44

.199 (.035)* —

.334 (.002)** .135 (.116)

3.31

—

—

Religious School/ education Human services Other

3.32

Other (significance) .320 (.007)** .121 (.220) -.014 (.102)

—

—

—

Overall significance = .014

Partial Eta2 = 0.079

Power = 0.789

a. Analysis adjusted by weight for volunteer domains due to heterogeneity (Levine test p < .01). Group sample sizes in order (religious to other) are 12, 36, 50, 38. *p < 0.10.** p < 0.01.

Table 2. Estimated Marginal Mean Differences for Compassion Across Volunteering Domains ANCOVA Results With Least Squares Difference Testsa Estimated marginal mean



Other (significance)

3.81 3.50

.309 (.053)* —

.315 (.051)* .006 (.957)

.126 (.438) -.183 (.087)*

3.49

—

—

-.189 (.053)*

3.68

—

—

—



2

Partial Eta = 0.056

Power = 0.614

a. Group sample sizes in order (religious to other) are 12, 36, 50, 38.

power is as well. Hence, the likelihood of falsely concluding that no differences exist in the population is high (i.e., Type 2 error). As shown in Table 4, the self-sacrifice subdimension exhibits a pattern similar to overall PSM. Volunteers to religious organizations score higher than all other volunteers, with overall statistical significance strong and reasonably accounted for partial variance and test power.

Discussion Overall, the findings of the empirical analysis are mixed. First, where differences occur in mean PSM scores the variance accounted for by volunteer domain is rather low. However, such low partial variances are very common in the volunteer literature


58


Table 3. Estimated Marginal Mean Differences for Commitment Across Volunteering Domains ANCOVA Results With Least Squares Difference Testsa


Estimated marginal mean



3.73 3.64

.094 (.295) —

.195 (.044)* .100 (.147)

3.54

—

—

-.009 (.899)

3.55

—

—

—


2

Partial Eta = 0.040

Other (significance) .185 (.065)* .091 (.228)

Power = 0.450

a. Analysis adjusted by weight for volunteer domains due to heterogeneity (Levine test p < .01). Group sample sizes in order (religious to other) are 12, 36, 50, 38.

Table 4. Estimated Marginal Mean Differences for Self-Sacrifice Across Volunteering Domains ANCOVA Results With Least Squares Difference Testsa


Estimated marginal mean



Other (significance)

4.74 4.53

.210 (.031)* —

.348 (051)** .138 (.957)

.333 (.006)** -.123 (.225)

4.40

—

—

4.41

—

—


2

Partial Eta = 0.081

-.015 (.887) — Power = 0.800

a. Analysis adjusted by weight for volunteer domains due to heterogeneity (Levine test p < .01). Group sample sizes in order (religious to other) are 12, 36, 50, 38.

that incorporates Volunteer Functions Inventory (VFI) motives and activity (see Stukas et al., 2009) and, in comparison to those findings, are relatively high in the present analysis (see Stukas et al., 2009; Wilson & Musick, 1999). Volunteer researchers argue that motivations alone may be important, but not extremely powerful, given a host of other explanations for volunteer behavior. For example, personal values may have more to do with what volunteering means to an individual than why they decide to contribute (e.g., Wilson, 2000). In addition, we would expect elite volunteers such as those sampled to have relatively high PSM levels (as the mean scores typically indicate) and, hence, far less variance to account for by any explanatory variable.


59

Coursey et al.

The clearest finding is that differences by domain mostly involve religious organizations. It should be noted that these differences exist even when controlling for church membership or attendance levels. This finding is very similar to Musick and Wilson’s (2008, p. 79) results based on the Independent Sector Giving and Volunteering in the United States (1990-1999) survey data, which encompass a much broader sample of volunteers and even nonvolunteers. The most notable difference they found was on the VFI “values” subdimension, which is very similar to PSM’s compassion subdimension, between volunteering for religious versus other organizations. One might well expect that people with higher motivations grounded in compassion would see volunteering for religious organizations as a better match than the other organizational domains. By contrast, the commitment subdimension of PSM demonstrates the weakest findings. Again, we might expect this result as the dimension specifically concerns public service, unlike the other two subdimensions, although we anticipated that volunteers in the education/school domain would score higher. The results also differ from Sundeen (1992), who using different measures of commitment (“public involvement”), found that education/schools volunteering ranked higher on this concept. “Commitment to public service” should apply more to civic and political organizations, which unfortunately had too small a sample (n = 3) for adequate comparison and analysis. We would, therefore, presume little difference between the four studied domains in volunteers’ expectations of how well each meets commitment motivation. Education/school volunteering should also be more strongly linked with extrinsic motivations as a result of parental self-interest; PSM as conceptualized only has one extrinsic-related subdimension, unlike the VFI that better incorporates extrinsic incentives. Although they are not always statistically significant, all mean orderings have religious volunteering surpassing volunteering in education/schools, followed by human services. It seems surprising that human services volunteering ranks well below the other domains on the commitment and self-sacrifice subdimensions (albeit not significantly). The commitment result could be attributable to school and educational organizations having greater similarity to public organizations than to nonprofit agencies. Mean scores on the self-sacrifice subdimension, however, are not so easily explained. One possibility is that many volunteers are employed in human service organizations, a very common type of volunteering, and discount their paid activities. If such paid work were considered, the result would likely differ as someone with a higher self-sacrifice score might well be more likely to work, not just volunteer, for such organizations.4 Another possibility is that human service volunteers may be far more varied in their motivations due to the wide range of activities and missions encompassed by that domain; they may also attach different motivations to the same activity (Wilson, 2000). Lastly, organizational formalization could be responsible. Participation (and employment) in human services as a volunteer may be more structured, regulated, and “bureaucratic,” and thus as theorized earlier, depress motivational levels.


60


Limitations and Future Research Overall, this study provides some evidence that functional theory may apply for public service motivation in relation to choices made by volunteers concerning domain of activity. Although the differences are not large, they are similar both in direction and strength to other volunteering research. The clearest finding is for religious organizations, a result paralleling Musick and Wilson’s (2008) findings using different motivational items though similar in construct definitions to PSM. Many issues await further inquiry. First, this study only examines volunteering. Another research question is what differences might be found across government domains, and do the differences hold for paid work, where entry and exit is much more restricted. One approach, though it has its own limitations, is to study the views of precareer MPA students concerning their preferred public work area, as exemplified by Light (1999). Preservice studies are quite rare compared to the volunteer literature, possibly because individuals have considerably less flexibility in paid employment. As a practical result, the findings suggest that, as in the volunteer recruitment literature (e.g., Clary et al., 1998), promotional messages stressing how work and opportunities meet the typical motivations of potential volunteers/employees may be more successful. Such a practical application is a noted deficiency in PSM research (see Perry & Vandenabeele, 2008; Perry et al., 2009). PSM literature, to date, does not test or expressly consider the practical use of the instrument (e.g., Perry et al., 2009). In addition, due in part to an initial proposition by Perry and Wise (1990), research on PSM tends to compare public organizations generically with private ones. The initial formulation of PSM did not presume any differences within sector—or at least did not account for them. Given the differences in PSM across volunteering domains found in the present study, research should consider the possibility that PSM varies across paid work in public agencies. PSM is nestled in altruistic behavior, not formal public service. Functional theory would hold that public employees may select agencies based on their perception of the altruistic motivations they may offer workers. Second, the theoretical justification for domain differences rests on formal structure and extrinsic motivations. These factors are not directly measured but instead assumed, though reasonably so, based on domain characteristics described in the volunteer literature. However, the validity of measuring the formal structure of an organization may be questionable. In theory, potential volunteers have scant information about the work organization of a specific entity. Instead, they likely rely on broad characterizations. Actual, organization-specific measures are likely more relevant once a volunteer joins, for example, in relation to their job satisfaction and retention (e.g., McCurley & Lynch, 2005). Overall, although this article represents progress in a scarcely researched area, theoretical explanations for domain differences require significant elaboration. Third, only three domains were tested in this analysis with a broad collapsed category for other types (civic and political organizations had to be excluded from analysis due to the very small sample used). A broader set of domains more consistent with the volunteer literature would be helpful. The “attraction to public policy-making”


61

Coursey et al.

subdimension of PSM was not included in the survey. It could be expected that the political domain would demonstrate the strongest differences (likely lower) on the studied dimensions as well as on the policy-making dimension (likely higher). In addition, unlike the VFI, PSM does not encapsulate extrinsic motivations very well (with the arguable exception of policy-making); future studies should consider measuring VFI extrinsic dimensions (such as career) along with PSM. Fourth, the samples are small. This limitation is attributable in part to the studied population: Finding elite volunteers working exclusively in one domain is challenging. In the statistical analysis, only tests for overall PSM and self-sacrifice have fair power. A larger sample with greater statistical power may reveal more statistically significant differences. However, it is noteworthy that many differences are large enough to be significant even for such small samples. One possibility is to compare motivations of volunteers by domain regardless of the number of domains in which they work. Although this procedure would increase the sample size, it would surely complicate the findings by making the attribution of motivational levels to a specific domain extremely difficult; some control for the number of domains would also be necessary. Qualitative, in-depth studies may be desirable yet raise their own methodological difficulties. Fifth, this study only examines elite (award-winning) volunteers. Virtually no other studies of volunteers specifically consider such individuals (most research compares nonvolunteers to volunteers or concentrates on more typical volunteers in terms of volunteering effort). Award-winning volunteers may differ from their more typical counterparts. One possibility is that such intense volunteers may be more prone to sunk cost explanations (e.g., Wilson & Musick, 1999) or postrecruitment justifications for their motivations tied to the organization’s activity (see Smith, 1982; Wuthnow, 1995). This sociological explanation proposes that motives may not account for behavior but instead act as rationalizations for an activity. However, Wilson and Musick did not find a relationship between the level of volunteering (in terms of hours and estimated monetary value) and attachment, a finding that disputes the sunk cost explanation. Attrition may be more of a concern since volunteers tend to leave if they find that an organization fails to meet their values (Stukas et al., 2009). The motivations of longer-term volunteers may be more domain specific, a result that would accentuate differences in PSM and its subdimensions. The sociological explanation potentially confounds all volunteer motivation research, not just the sampled population examined here. Further inquiry requires a broader sample of volunteering levels to evaluate whether findings are stable by volunteer effort and recognition (i.e., awards). Both sunk cost and sociological explanations require time series or experimental approaches. Thus, future research testing functionalfit theory should compare motivations of volunteers at entry (recruitment) to later times.

Conclusion The results of this analysis suggest that public service motivation (PSM) varies by volunteering domain, even when controlling for antecedents. Most observed differences


62


involved religious organizations, with higher PSM and dimensional levels in this domain. PSM research incorporating domain is nascent, however, and numerous research questions and avenues await inquiry. Interestingly, more complex explanations derived from established theory are more common in the analogous volunteering literature. Whatever direction future research takes, this study provides a useful extension of the PSM literature using functional theory and/or person–environment fit as an explanation for how PSM relates to work choices. In assuming that people choose work based on how well they perceive it fits their needs, functional theory offers a potentially practical application of PSM as an agency could adjust its recruiting and work activities to better fit the functional needs of its potential and current employees. Indeed, PSM researchers should strongly consider the applicability of volunteer studies and theory as volunteer motivation is a very similar concept to PSM, and the volunteer literature has a practical orientation. Our understanding of the role of PSM in public organizations is contingent on developing a greater theoretical and empirical sophistication analogous to that found in volunteering research.

Appendix Public Service Motivation The following statements ask for your opinions. Please read each statement carefully and check the column that best reflects your view. (Response options—agree, somewhat agree, neither agree nor disagree, somewhat disagree, disagree). Compassion a. I seldom think about the welfare of people I don’t know personally. (REVERSED) b. I have little compassion for people in need who are unwilling to take the first step to help themselves. (REVERSED) c. Most social programs are too vital to do without. d. It is difficult for me to contain my feelings when I see people in distress. e. I am often reminded by daily events about how dependent we are on one another. Self-sacrifice a. Much of what I do is for a cause bigger than myself. b. I am one of those rare people who would risk personal loss to help someone else. c. Making a difference in society means more to me than personal achievements. d. I think people should give back to society more than they get from it. (continued)


63

Coursey et al.

Appendix (continued) Commitment/civic duty a. I unselfishly contribute to my community b. Meaningful public service is very important to me. c. I consider public service my civic duty. Authors’ Note This article is a revised version of a paper originally presented at the International Public Service Motivation Research Conference, Indiana University, Bloomington, Indiana, June 7-9, 2009. Authors are listed alphabetically after the first author, and this ordering does not indicate contribution levels.

Declaration of Conflicting Interests The authors declared no potential conflicts of interests with respect to the authorship and/or publication of this article.

Funding The fourth author’s participation in this research was supported by the World Class University (WCU) program through the Korea Science and Engineering Foundation funded by the Ministry of Education, Science and Technology (R32-20002).

Notes 1. Details about these awards including nomination processes, selection criteria, and recent awardees are available from the Points of Light Foundation website: Daily Points of Light Award, http://www.pointsoflight.org/awards/dpol/; President’s Community Volunteer Award. Retrieved August 17, 2006, from http://www.pointsoflight.org/awards/prescommunityvol/ 2. Levine test (p < .01). Although ANCOVA as a form of ANOVA design can be fairly robust to nonconstant error variance, it is especially problematic for small samples and where the groups sizes are quite divergent. The overall model tests are too conservative in such cases where the smallest group size has the lowest variance, which was the case for all analyses. As diagnostics indicated volunteer domain was the source of the heterogeneity, a weight for domain was calculated using a weighted regression and then applied. 3. We elected not to use a Bonferroni or some other commonly used adjustment for Type 1 error inflation from pairwise comparisons. The reason is that, despite the reputation of such tests, they tend to be overly conservative and do not really control for Type 1 error as appropriately as usually assumed. Such pairwise test controls are discouraged where all pairs are compared, for a small number of comparisons, and where there is a small sample (all of which apply here; for a discussion of these issues, see Fiese, 2002).


64


4. This problem is common in empirical analysis in the volunteer literature where employees of organizations utilizing volunteers have a difficult time separating their paid versus volunteer work ( Musick & Wilson, 2008), and/or the commitment attributable to employment is not well conceptualized or captured empirically.

References Bright, L. (2007). Does person-organization fit mediate the relationship between public service motivation and the job performance of public employees? Review of Public Personnel Administration, 27(4), 361-379. Brockett, W., Cooper, S., Kumbhakar, M., & Kwinn, D. (2004). Alternative statistical regression studies of the effects of joint and service specific advertising on military recruitment. Journal of the Operational Research Society, 55(10), 1039-1048. Clary, E., Snyder, M., Ridge, R., Copeland, J., Stukas, A., Haugen, J., et al. (1998). Understanding and assessing the motivations of volunteers—a functional approach. Journal of Applied Social Psychology, 24, 1129-1150. Clary, E., Snyder, M., & Stukas, A. (1996). Volunteers, motivations: Findings from a national survey. Nonprofit and Voluntary Sector Quarterly, 25, 485-505. Cnaan, R., Handy, F., & Wadsworth, M. (1996). Defining who is a volunteer: Conceptual and empirical considerations. Nonprofit and Voluntary Sector Quarterly, 25(1), 364-383. Coursey, D., & Pandey, S. (2007). Public service motivation measurement: Testing an abridged version of Perry’s Scale. Administration and Society, 39(5), 547-568. Coursey, D., & Pandey, S. (2009). PSM and support for citizen participation: A test of Perry and Vandenabeele’s reformulation of PSM theory. Paper presented at the International Public Service Motivation conference, Bloomington, IN. Coursey, D., Perry, J., Brudney, J., & Littlepage, L. (2008a). Psychometric verification of Perry’s Public Service Motivation Instrument: Results for volunteer exemplars. Review of Public Personnel Administration, 28(1), 79-90. Coursey, D., Perry, J., Brudney, J., & Littlepage, L. (2008b). Deconstructing public service motivation: Vital measurement and discriminant validity issues in volunteerism. Minnowbrook III, September 2008, Lake Placid, NY. Fiese, R. (2002). Do multiple outcome measures require p-value adjustment? BMC Medical Research Methodology, 2, 8. Francois, P. (2000). Public service motivation as an augment for government provision. Journal of Public Economics, 78(3), 275-299. Hanssens, D., & Levien, H. (1983). An econometric study of recruitment marketing in the U.S. Navy. Management Science, 29(10), 1167-1184. Knoke, D., & Wood, J. (1981). Organized for action: Commitment in voluntary associations. New Brunswick, NJ: Rutgers University Press. Light, P. (1999). The new public service. Washington, DC: Brookings. Lovell, C., & Morey, R. (1991). The allocation of consumer incentives to meet simultaneous sales quotas: An application to U.S. Army recruiting. Management Science, 37(3), 350-367.


65

Coursey et al.

McCurley, S., & Lynch, R. (2005). Keeping volunteers: A guide to retention. Olympia, WA: Fat Cat Publications. McKnight, P., McKnight, K., Sidani, S., & Figueredo, A. (2007). Missing data: A gentle introduction. New York: Guilford. Musick, M., & Wilson, J. (2008). Volunteers: A social profile. Bloomington: Indiana University Press. O’Brien, L., Crain, T., Omoto, L., & Snyder, M. (2000, May). Matching motivations to outcomes: Implications for persistence in service. Paper presented at the annual meeting of the Midwestern Psychological Association, Chicago. Pandey, S., & Stazyk, E. (2008). Antecedents and correlates of public service motivation. In J. Perry & A. Hondeghem (Eds.), Motivation in public management (pp. 101-117). Oxford, UK: Oxford University Press. Perry, J. (1996). Measuring public service motivation: An assessment of construct reliability and validity. Journal of Public Administration Research and Theory, 6(1), 5-22. Perry, J. (1997). Antecedents of public service motivation. Journal of Public Administration Research and Theory, 7(2),181-197. Perry, J., Brudney, J., Coursey, D., & Littlepage, L. (2008). What drives morally committed citizens? A study of the antecedents of public service motivation. Public Administration Review, 68(3), 445-458. Perry, J., Hondeghem, A., & Wise, L. (2009, June). Revisiting the motivational bases of public service: Twenty years of research and an agenda for the future. Paper presented at the International Public Service Motivation conference. Bloomington, IN. Perry, J., & Thomson, A. (2004). Civil service: What difference does it make? Armonk, NY: M. E. Sharpe. Perry, J., & Vandenabeele, W. (2008). Behavioral dynamics: Institutions, identities, and selfregulation. In J. Perry & A. Hondeghem (Eds.), Motivation in public management: The call of public service, 56-79. Oxford, UK: Oxford University Press. Perry, J., & Wise, L. (1990). The motivational bases of public service. Public Administration Review, 50(3), 367-373. Smith, D. (1982). Altruism, volunteers, and volunteerism. In J. Harman (Ed.), Volunteerism in the eighties. Washington, DC: University Press American. Snyder, M., Clary, E., & Stukas, A. (2000). The functional approach to volunteerism. In G. Maio & J. Olson (Ed.), Why we evaluate (pp. 365-393). Mahwah, NJ: Lawrence Erlbaum. Snyder, M., & Ickes, W. (1985). Personality and social behavior. In G. Lindzey & E. Aronson (Eds.), The handbook of social psychology (pp. 883-947). New York: Random House. Snyder, M., Omoto, A., & Lindsay, J. (2004). Sacrificing time and effort for the good of others: The benefits and costs of volunteerism. In A. Miller (Ed.), The social psychology of good and evil (pp. 444-468). New York: Guilford. Steinhaus, C., & Perry, J. (1996). Organizational commitment: Does sector matter? Public Productivity and Management Review, 19(3), 278-288. Stukas, A., Worth, K., Clary, E., & Snyder, M. (2009). The matching of motivations in affordances in the volunteer environment: An index for assessing the impact of multiple matches on volunteer outcomes. Nonprofit and Voluntary Sector Quarterly, 38(1), 5-28.


66


Sundeen, R., (1992). Differences in personal goals and attitudes among volunteers. Nonprofit and Voluntary Sector Quarterly, 21(3), 271-291. Sundeen, R. (1990). Family life course status and volunteer behavior: Implications for the single parent. Sociological Perspectives, 33(4), 483-500. Wilson, J., & Musick. M. (1999). Attachment to volunteering. Sociological Forum, 14(2), 243-272. Wright, B., & Pandey, S. (2008). Public service motivation and the assumption of personorganization fit: Testing the mediating effect of value congruence. Administration and Society, 40(5), 502-521. Wuthnow, R. (1991). Acts of compassion. Princeton, NJ: Princeton University Press. Wuthnow, R. (1995). Learning to care: Elementary kindness in an age of indifference. New York: Oxford University Press.

Bios David Coursey is a Visiting Scholar at Arizona State University’s Decision Theater. He specializes in public management and information systems and quantitative methods. Jeffrey L. Brudney, Ph.D., is the Albert A. Levin Chair of Urban Studies and Public Service at the Maxine Goodman Levin College of Urban Affairs, Cleveland State University. He has published widely in the areas of volunteerism, nonprofit management, and public management. Laura Littlepage is a faculty member of both the School of Public and Environmental Affairs and Philanthropic Studies at Indiana University Purdue University Indianapolis as well as a senior researcher with the Center for Urban Policy and the Environment. She is the author of numerous journal articles and book chapters. James L. Perry is Distinguished Professor, Indiana University, Bloomington, and World Class University Distinguished Professor, Department of Public Administration, Yonsei University. He is recipient of the 2008 Dwight Waldo Award and co-editor of Motivation in Public Management: The Call of Public Service (Oxford University Press, 2008).
