Content uploaded by Marlene Walk
Author content
All content in this area was uploaded by Marlene Walk on Jun 04, 2019
Content may be subject to copyright.
1
“Don't you want to stay?” The impact of training and
recognition as HR practices on volunteer turnover
Marlene Walk
School of Public and
Environmental Affairs
Indiana University-Purdue
University Indianapolis
mwalk@iupui.edu
Ruodan Zhang
School of Public and
Environmental Affairs
Indiana University,
Bloomington
rz3@umail.iu.edu
Laura C. Littlepage
1) School of Public and
Environmental Affairs
Indiana University, Bloomington
2) Public Policy Institute, Indiana
University
llittlep@iu.edu
Note: this is the pre-copyedited version as submitted to the publisher.
Please cite as: Walk, M., Zhang, R., & Littlepage, L. (online first) “Don't you want to
stay?” The impact of training and recognition as human resource practices on volunteer
turnover. Nonprofit Management and Leadership.
Financial support: The first author would like to thank the IUPUI Office of the Vice
Chancellor for Research, the IUPUI Office for Women, and Indiana University’s Public Policy
Institute for their generous support.
Acknowledgements: We extend our gratitude to the Boy Scouts Crossroads of America
Council for their data access. We also thank Beth Gazley for comments on an earlier version of
this paper, Megan Bowers for her support in data cleaning, and Charity Scott for her help in
manuscript preparation. Finally, we also thank our anonymous reviewers for their constructive
feedback during the review process.
2
Abstract
Managing volunteers is a difficult undertaking. This study draws on human resource
management theory and literature to investigate the effect of two human resource (HR)
practices—training and recognition—on volunteer turnover. We use longitudinal administrative
data collected by an Indiana nonprofit organization, which contains individual volunteer
characteristics, organizational HR practices, and information on actual turnover behavior. We
found that recognizing volunteer contributions with awards predicted volunteer retention in the
following year. Training did not have a direct effect on volunteer turnover, but interacted with
gender; men who received training were more likely to stay than women. The study contributes
to the literature on human resource management in the volunteer context, adds to the emerging
literature on awards as incentives for volunteers, and addresses the common method bias by
using longitudinal data.
Keywords: turnover, volunteers, HR practices, human resource management, volunteer
management
Background and Significance
In 2015, about 25% of US adults donated 8.7 billion hours of their time to nonprofits
(Bureau of Labor Statistics, 2016). Volunteering, defined as activities performed in formal
settings in which time is freely given to benefit others without remuneration (Cnaan, Handy, &
Wadsworth, 1996), is, thus, a major source of (unpaid) labor for nonprofits. Indeed, beyond
universal board service, an estimated 80% of all charitable nonprofit organizations rely on
volunteers for service provision and management (Hager & Brudney, 2008). Nonprofits, in turn,
3
are better able to provide services and offer programs in diverse areas such as education, the
environment, or human services.
Managing volunteers, however, is a difficult undertaking. Due to their non-contractual
relationship with nonprofits, volunteers ‘vote with their feet’ and leave organizations when they
are dissatisfied. As such, the volunteer nature of work challenges traditional retention strategies,
which operate under the assumption of mutual dependency between employee and organization
(e.g., exchange of labor for a wage; Simon, 1951). Indeed, volunteer turnover has been regarded
as an increasing challenge to nonprofits’ provision of services (Garner & Garner, 2011). Reliable
statistics are scarce, but the Corporation for National and Community Service (2015) reports
annual turnover rates between 24 – 47% for the organizations with which they work.
Contributions to the volunteer management literature in the past few years has led to a
better understanding of the intricacies of volunteer turnover. Scholars have identified a range of
significant predictors for volunteer retention such as voice (Garner & Garner, 2011), volunteer
role identity (Grube & Piliavin, 2000), perceived competence and efficacy (Ripamonti,
Pasquarelli, Ravasi, & Sala, 2016; Wu, Li, & Khoo, 2016), distributive justice (Hurst, Scherer, &
Allen, 2017), as well as volunteer motivation (e.g., Garner & Garner, 2011; Gazley, 2013;
Nencini, Romaioli, & Meneghini, 2016). Aside from individual volunteer motives and
dispositions, effective management within the organization also contributes to volunteer
retention (Hager & Brudney, 2008; Gazley, 2013). Scholars have also examined volunteers’
perceptions of the organizational context impacting retention success such as organizational
climate (Nencini et al., 2016), the design of volunteer roles (Alfes, Shantz, & Saksida, 2015),
organizational support (Alfes, Shantz, & Bailey, 2016), and job resources (Presti, 2013). Rather
than relying on individual perceptions of volunteer management practices, this study builds on
4
these findings and adopts a management perspective by drawing on objective measures of human
resource (HR) practices that impact volunteer retention. Our research question is: To what extent
do HR practices of volunteer training and recognition influence actual turnover of volunteers? In
answering this question, we particularly focus on the effects of training and recognition on
volunteer turnover while using longitudinal administrative data containing information on 7,595
volunteers and controlling for their demographic information (such as age, gender, race)
alongside volunteer context information (such as roles, ranks).
Scholars have increasingly investigated the applicability and transferability of human
resource management practices to the volunteer context. Their findings indicate that certain HR
practices can impact volunteer attitudes and behaviors, and therefore, improve volunteering
outcomes, such as retention rates (Saksida, Alfes, & Shantz, 2016), as well as organizational
outcomes, such as client satisfaction ratings (Rogers, Jiang, Rogers, & Intindola, 2016). Even
though efforts have been made to develop a theoretical understanding of the potential for HR
practices to impact volunteer retention, this research area remains underdeveloped (Alfes,
Antunes, & Shantz, 2016) leading Bartram, Cavanagh, and Hoye (2017) to note that there is an
“absence of research on the impact of HRM within volunteer and grassroots community
organizations” (p. 1907-1908). As such, this study responds to calls for research on the impact of
HR practices on volunteer outcomes (Alfes, Shantz, & Saksida, 2015; Studer & von Schnurbein,
2013) by focusing on two prominent HR practices—training and recognition—and their impact
on volunteer turnover. Particularly, we test the HR bundling perspective (Delery & Doty, 1996;
MacDuffie, 1995) in the context of volunteers, indicating that HR practices work best if utilized
in well-aligned bundles.
We also contribute to the emerging awards literature (Frey & Gallus, 2017; Frey &
Gallus, 2018; Gallus, 2016). Originally stemming from economics, awards have been proposed
5
as a valuable instrument of recognition for volunteers as they create benefits for those who win
(Frey & Gallus, 2017). This study contributes to the knowledge on awards and their particular
impact on subsequent performance (Frey & Gallus, 2017; Frey & Gallus, 2018; Gallus, 2016).
We hereby respond to calls to study awards in non-virtual fields in the context of volunteer
retention (Gallus, 2016; Zhang & Zhu, 2011).
In trying to understand volunteer turnover, most of the previous literature relies on cross-
sectional, self-reported data collected from volunteers or volunteer managers. There are three
particular reasons why this should be concerning. First, self-reports are a problematic tool to
make claims about certain phenomena, since they are prone to measurement error (Podsakoff,
MacKenzie, Lee, & Podsakoff, 2003). Second, studies relying on self-reported data most
frequently assess intentions to stay/quit as proxy rather than actual turnover, since data on actual
turnover is difficult to collect in a systematic fashion (Henderson & Sowa, 2017). Self-reporting,
however, may be difficult for respondents due to challenges of recall over a certain time span
(Schwarz & Oyserman, 2001). Moreover, in a study on volunteer management practices, those
surveyed may not want to draw attention to potential volunteer retention problems, which
potentially leads to social desirability bias (Podsakoff et al., 2003). Scholars have therefore
called for research incorporating objective measures to mitigate these limitations (Newton,
Becker, & Bell, 2014). Finally, most studies on volunteer retention collect data for the
independent and dependent variables from the same source at one point in time, which likely
causes common method bias (Podsakoff et al., 2003). Our longitudinal administrative data allow
us to eliminate these previous methodological limitations.
6
Literature Review - The role of human resource management in volunteer retention
Similar to paid employee retention, volunteer retention is one of the greatest challenges
for volunteer-dependent nonprofit organizations (Garner & Garner, 2011). These organizations
have to invest time and money into volunteer replacement since new volunteers need to be
properly onboarded and get acquainted with organizational processes and culture in order to
perform their tasks well (Clary, Snyder, & Ridge, 1992; Handy & Srinivasan, 2004). Traditional
human resource management focuses on HR practices such as selection/recruitment,
performance management, or training/development to effectively and efficiently manage human
resources (Huselid, 1995). Even though these HR practices have been proven effective in the
case of paid employees, it is unclear whether or not these can be applied to the context of
volunteers (Alfes et al., 2016), since volunteers differ significantly from paid employees in key
areas including formal relationship with the organization, attitudes, behaviors, and motivations
(Studer & von Schnurbein, 2013). For example, volunteers are not bound to the organization by a
contractual agreement, volunteer only for a few hours per week and often for more than one
organization, and are not motivated by receiving a paycheck or health benefits (Alfes & Langner,
2017; Cnaan & Cascio, 1998; Fallon & Rice, 2015). The extent to which volunteers and paid
employees are different may depend on the specific roles and the work environment (Liao‐Troth,
2001) as well as on organizational culture (Schein, 2010). Volunteers, compared to paid
employees, can have weaker ties that bind them to the organization and, therefore, easily ‘vote
with their feet’ and leave when they are dissatisfied (Pearce, 1983). Therefore, it is important for
nonprofits to engage in strategic HR planning for volunteers and paid staff individually, being
cognizant of the HR practices in place that will best recruit and retain both groups (Alfes et al.,
2016). This study particularly focuses on two HR practices—training and recognition—that are
frequently employed to support volunteer performance.
7
Recognition as HR practice
Recognition activities, broadly defined as activities to reward volunteers for their labor,
play an important role in fostering positive relationships between volunteers and the organization
(Fallon & Rice, 2015). Recognition activities come in different forms and range from
acknowledgements in newsletters or during public events, to awards for the service provided, to
a simple ‘thank you’ from an organization member or service recipient (Fallon & Rice, 2015;
Kovacs & Black, 2000; Smith & Grove, 2017). Recognition is especially important for long-term
volunteers (Schlesinger, Egli, & Nagel, 2013) and positively impacts volunteer satisfaction,
commitment, and tenure (Cnaan & Cascio, 1998; Fallon & Rice, 2015; Smith & Grove, 2017).
Different types of recognition lead to different desired outcomes. For instance, Cnaan and Cascio
(1998) find that receiving thank you letters is related to volunteer satisfaction, whereas free
medical care and meals as well as the ability to attend conferences are associated with
commitment. These authors also show that certificates of appreciation as well as thank you
luncheons are related to tenure with the organization. Farmer and Fedor (1999) study the impact
of perceived organizational support—a measure that captures recognition activities alongside
other aspects—on turnover intentions and find a negative relationship, indicating that volunteers
are less likely to want to leave if they feel supported and recognized in the organization.
Similarly, a lack of or inadequate recognition from the organization can lead to dissatisfaction
and volunteer turnover (Smith & Grove, 2017).
This paper particularly focuses on awards as a form of recognition. Awards have a
history in film (such as the Oscars) and academia (such as the Nobel Prize) but have only
recently received attention in the volunteer space (Frey & Gallus, 2018). Frey and Gallus (2018)
propose awards as valuable incentives for volunteers as they create benefits for those who win.
8
For instance, volunteers who receive awards feel recognized and honored while those who do not
receive awards would like to do so in the future and, thus, may increase their performance efforts
(Frey & Gallus, 2018). Generally, the literature distinguishes between discretionary and
confirmatory awards; confirmatory awards are given out following clearly defined and
observable achievements, whereas discretionary awards are given out at the discretion of the
giver aiming to recognize exceptional behavior (Frey & Gallus, 2017).
To date, the impact of awards on subsequent performance has been understudied (Frey &
Gallus, 2017) as disentangling the causal mechanism between awards and performance is
challenging, given the apparent selection bias: awards tend to be given out to those individuals
displaying high performance, however, receiving an award may also lead to higher performance.
A recent natural field experiment in the volunteer context further explored the directionality of
the causal effect. Gallus (2016) randomly selected new volunteer editors on Wikipedia to receive
awards while others were excluded from receiving awards. Her findings indicate that receiving
discretionary awards lead to higher subsequent performance as well as increased retention rates
(Gallus, 2016). Particularly, volunteers who received awards were more active in the months
following the award receipt, indicating a direct impact on performance. Moreover, awards also
had a strong and sustained impact on retention; one month after award receipt retention rates
were 20% higher, and two months later retention rates were 14% higher for award recipients as
compared to non-recipients. While we cannot fully eliminate selection bias in our study, the
findings from Gallus (2016) increase our confidence in the discretionary award-subsequent
performance relationship and also point towards the potential positive impact of awards on
volunteer retention in the context of our study over time. As the findings in Gallus (2016) are
limited to Wikipedia volunteers, we further test the relationship between discretionary awards
9
and volunteer retention in a non-virtual setting. Our paper, thus, further disentangles the award-
performance relationship (Gallus, 2016; Zhang & Zhu, 2011).
Training as HR Practice
Training is another commonly used HR practice. Training increases the volunteers’
abilities and skills to perform their volunteer tasks well. Moreover, training helps volunteers to
further increase their knowledge about the organization, its goals, and their own role in achieving
these goals (Saksida et al., 2016). After receiving training, volunteers experience higher levels of
confidence in their abilities to perform their roles (Newton, Becker, & Bell, 2014). The literature
on the impact of training on volunteer retention is limited but, generally, there is a positive effect
of training on retention. For instance, research among volunteers of five different organizations
finds that perceived learning and development opportunities were positively related to intentions
to stay (Newton et al., 2014). Henderson and Sowa (2017), in their study of volunteer fire
fighters, find that training opportunities were positively related to volunteers’ intentions to
remain in one year and in five years. Other studies investigating the relationship between
recognition and training and the impact of both on volunteers’ intention to stay, present similar
positive findings (Fallon & Rice, 2015). However, the above-mentioned studies measured
perceptions of training opportunities rather than actual training.
Rather than assessing perceptions, Hager and Brudney (2008) focus on the impact of
actual volunteer management practices on volunteer retention. Volunteer managers of a
randomly drawn sample of nearly 3,000 charities were asked about the extent to which their
organizations adopted volunteer management practices such as supervision and communication
with volunteers, screening and matching volunteers to jobs, recognition activities, training and
professional development, among others. Recognition activities were most commonly employed
10
(82% adopted those practices to some or to a large degree) closely followed by training and
development opportunities (74% adopted those practices to some or to a large degree). Findings
from regression analysis indicate that both recognition and training activities positively influence
volunteer retention in the sampled organization.
Similar to Hager & Brudney (2008), the present study draws on actual training and
recognition information and can, therefore, provide a more objective assessment of the
effectiveness of these HR practices on volunteer retention. Based on the reviewed literature, we
hypothesize a negative relationship between HR practices of training and recognition on actual
volunteer turnover indicating that volunteers who receive training and/or recognition are more
likely to remain with the organization the following year.
Hypothesis 1: Recognition in form of discretionary awards will be negatively related to
actual volunteer turnover.
Hypothesis 2: Training will be negatively related to actual volunteer turnover.
Training and recognition as HR bundle
Human resource management scholars have argued that individual HR practices work
best if integrated into HR bundles (Delery & Doty, 1996; MacDuffie, 1995). HR bundles consist
of individual HR practices that are complementary and reinforce each other (MacDuffie, 1995).
If individual HR practices are well-aligned and internally consistent (Delery, 1998), HR bundles
can have a greater effect on performance and other organizational outcomes such as job
satisfaction, turnover, or organizational commitment, exceeding the effect of individual practices
(MacDuffie, 1995; Macky & Boxall, 2007).
There are different ways to operationalize HR bundles. The two most prominent
approaches are the additive—using indices—and the multiplicative—using interaction terms
11
(MacDuffie, 1995; Delery, 1998). The additive approach assumes that utilizing more HR
practices is better and will increase the net effect on the desired outcomes, whereas the
multiplicative approach assumes that one HR practice depends on the extent of other practices in
the system (MacDuffie, 1995; Delery, 1998). Since we reason that training and recognition are
not dependent on each other in the context of our study, we pursue the additive approach.
Hypothesis 3: An HR bundle consisting of training and recognition will be negatively
related to actual volunteer turnover.
Data and Methods
Our data stem from the Boy Scouts Crossroads of America Council (BSA) which
operates in 26 counties in Indiana and serves 33,000 youth through the support of about 7,500
volunteers. These BSA volunteers are considered long-term and are mostly involved in providing
weekly programming. Consequently, volunteer retention is a critical management issue for the
BSA. The BSA collects administrative data from volunteer application forms and regularly
enters more information into the system (e.g., changes in volunteer roles). Moreover, all BSA
volunteers have to renew their commitment annually in a process that is called “rechartering.”
The rechartering data can be considered actual turnover data as those who have not renewed their
commitment have essentially dropped out of volunteer service.
This paper draws on de-identified longitudinal administrative data from 2016-2017. The
original 2016 data consist of 7,597 volunteers and the 2017 data of 7,036 volunteers. Each
observation was assigned a unique identifier composed of the first letter of the volunteer’s first
name, the last letter of the volunteer’s last name, gender designation, birthday (1-31), and birth
year. Only 3.45 percent of the 2016 data and 3.21 percent of the 2017 data were duplicate
records using the original identifier. We added information on district—the BSA divides the
12
Indiana counties into 13 districts, most of which serve multiple counties—unit numbers, and
awards received to the identifier, in order to remove all remaining duplicates within the
respective data sets.
After the data cleaning process,i a total of 6,629 entries were matched across both 2016
and 2017 data sets. The remaining 968 cases out of the 7,597 entries in the 2016 data set were
identified as volunteers who did not renew their commitment for the following year. We also
excluded 2 observations that suggest significant data entry error—one with no age value or
identifier information, and the other with an age of 14 and substantial missing values. In
summary, the data reported 6,628 (87.27%) rechartered volunteers from 2016, and 967 (12.73%)
discontinued volunteers (see Table 1). This turnover rate (12.73%) was well below findings from
other organizations (Corporation for National and Community Service, 2015).
Volunteers within the dataset volunteer either at a local BSA council, at a district within a
council, or in a Scout unit. A majority (about 70%, N = 5,316) volunteer in a Scout unit where
they directly interact with the Scouts (hereafter, “unit-level volunteers”). Considering possible
heterogeneity in volunteer experience given different volunteer task assignment, it is important
to control for volunteer roles and ranks. In addition, while the administrative data do not contain
information on the length of volunteering, volunteer rank information best suggests volunteers’
previous involvement and progress with the BSA. However, this data set only records such
information for unit-level volunteers. There is no other clear identifier to separate rank and role
information for volunteers serving on the council or at the district level. In total, the data set
recorded complete information, including volunteer roles and ranks, on 4,310 unit-level
volunteers (56.75% of the total observations), who constitute the majority of the BSA’s
volunteer workforce and are most relevant to the organization’s management practice.
13
Overall, the full data set (N = 7,595) and the subset of unit-level volunteers do not differ
significantly on the explanatory variables (including the proportion of Eagle Scouts). However,
the full data set reported a slightly older population (mean = 45.5 years, versus 43.2 years in the
unit-level sample), lower level of training (mean = 0.34, versus 0.49), slightly higher proportion
of discretionary award recipients (mean = 0.11, versus 0.095), but lower turnover rates (mean =
0.13, versus 0.18). (Analyses are available upon request.)
<insert Table 1 about here>
Measures
Dependent variable
This study uses actual volunteer turnover as the dependent variable. The variable was
constructed by merging the data across 2016 and 2017. A rechartered volunteer (Turnover = 0)
was identified if the volunteer had matched entries in both 2016 and 2017 data sets. A
discontinued volunteer (Turnover = 1) was identified if the entry was only available in 2016 with
no match in 2017. Observations only available in 2017 were not part of this analysis, because
these account for new volunteers joining in 2017.
Independent variables
This study uses volunteer training and recognition as independent variables. Volunteer
training was measured as whether the volunteer participated in optional unit-level training
(Training = 1; no training = 0) in the 2016 administrative record. The BSA offers general
training, such as orientation, and position-specific trainings (e.g., Pack committee member
position-specific training, Tiger Cub Den leader position-specific training) for adult volunteers.
Training is available to all individuals, regardless of their volunteer role (unit, district, or
14
council). We exclude mandatory training, such as youth protection training, that all registered
volunteers must retake every two years.
Volunteer recognition was measured as whether the volunteer received any discretionary
award (Discretionary award = 1; no award = 0) in the 2016 administrative record. Possible
awards include scout leader awards (e.g., District Award of Merit) and other awards (e.g., World
Conservation Award) which are given out at the discretion of the BSA. Boy Scout volunteers can
also pursue awards that are rather confirmatory in nature, such as the Mile Swim or Tiger Cub
Den Leader Award—awards for which there are clear criteria necessary for achieving them (Frey
& Gallus, 2017). We excluded confirmatory awards from the analyses because, in these cases,
the award givers do not enjoy the discretion to recognize top performers, and top performers are
guaranteed recognition by meeting certain criteria.
We operationalized the HR bundles of training and recognition using a Training and
Recognition Index variable, following the additive approach suggested by MacDuffie (1995).
Specifically, we constructed an index of these two HR practices with values ranging from 0 (no
practice present), to 1 (either practice present), to 2 (both practices present).
Control variables
We controlled for volunteer roles, ranks, and organizational sponsor at the contextual
level. Volunteer roles differ by the type of unit (Troop/Pack/Crew/Ship/Team) and respective
ranks within the unit. Within the 2016 data set, Troop (N = 2,478; 47%) and Pack (N = 2,586;
49%) units had the largest shares of volunteers, followed by Crew units (N = 222; 4%). We
constructed Unit type: Pack as a dichotomous variable that compares Pack units with other units
(Pack = 1; Other units = 0). We isolated Pack units for comparison because volunteers in the
Pack units serve a younger population (“Cub Scouts”—generally from grade 1 to grade 5),
compared with other scouting units.
15
We constructed the unit rank (ordinal) variable to measure differences in rank positions
within the units in 2016. For unit rank (ordinal), we assume that there are no hierarchical
differences between volunteers in leadership teams and committee teams—leadership and
committee are only different in terms of functions. However, within leadership/committee, we
coded the hierarchical relationships for each position. For example, within Troop Committee, the
chairman is coded as 1; the Committee member is coded as 2; unit secretary and training
chairman is coded as 3; and chartered organization representative is coded as 4.
Faith-based sponsor is a dichotomous variable indicating whether the volunteer unit is
sponsored by a faith-based organization (sponsored by a faith-based organization = 1; otherwise
= 0), usually a church or a parish. Other organizational sponsors include foundations (e.g., Camp
Sertoma Foundation), public charities (e.g., local parent-teacher associations and youth centers),
and public organizations (e.g., fire departments, parks and recreation departments, etc.).
We also included individual level control variables of age (in years, reported in 2016),
gender (female = 1; male = 0), and race (white = 1; other = 0). Finally, we used Eagle Scout as a
proxy for volunteer commitment (received Eagle Scout rank = 1; no Eagle Scout rank = 0)
among the male volunteers, because Eagle Scout is the highest designation one can obtain in the
scouting program (beyond the achievement of Life Scout). Nationally, only about 5 percent of all
Boy Scouts receive the Eagle Scout rank (National Eagle Scout Association, 2017). Therefore,
this achievement signals strong commitment to the BSA and the scouting program, and Eagle
Scouts were found to be more involved in community activities (Kim, Jang, & Johnson, 2016;
Polson, Kim, Jang, Johnson, & Smith, 2013).
16
Data analysis
We tested differences of means in the bivariate analysis, to examine the relationship
between independent/control variables and the turnover outcome. We estimated the association
of the variables using logistic regression models. Model 1 (N = 6,292) and Model 2 (N = 5,297),
as the baseline models, only include individual-level and contextual-level control variables,
respectively. Model 3 (N = 4,310) examines the effect of volunteer training (optional unit-level
training) and recognition (discretionary awards) on the likelihood of volunteer turnover,
controlling for individual and contextual characteristics. As a robustness check, Model 4 adds the
Eagle Scout variable as a measure of organizational commitment on the subset of male
volunteers (N = 2,912). Findings from this model led us to conduct a post-hoc analysis as
presented in Model 5 (N = 4,310), where we included an interaction effect between training and
gender to test whether volunteers in different gender categories responded to training
differently.ii The results for the abovementioned models are reported in Table 3 as odds ratios
(eβ). The Additive Approach Model (N = 4,310) in Table 4 examines the bundling effect of HR
practices. We included the Training and Recognition Index, controlling for individual and
contextual characteristics.
Results
The descriptive statistics (Table 1) illustrate that approximately 34 percent of volunteers
participated in training, and about 11 percent of volunteers received formal recognition in the
form of discretionary awards through the BSA.iii There were 277 (about 4%) volunteers who
received both training and recognition. The 2016 volunteer population was mostly male (69%)
and mostly white (95%), with an average age of 46 years. A majority (70%) of the volunteers
were sponsored by a faith-based organization.
17
The bivariate analysis in Table 2 shows statistically significant differences between those
volunteers who stayed (i.e., no turnover) and those who left for all variables. However, when
considering effect sizes (Cohen’s d), the magnitude of the differences was mostly small-to-
medium. In particular, both Eagle Scout status (t = 2.061, p < 0.05, Cohen’s d = 0.09) and race (t
= 2.872, p < 0.01, Cohen’s d = 0.109) present a significant but rather trivial effect. This means
that, despite the differences, the distribution of the independent variables overlaps to a large
extent for the groups of volunteers who stayed and for those who left. There are likely other
factors that play a more important role in determining volunteer turnover but were not captured
by our data set. We further discuss this in our recommendations for future research.
Preliminary results suggest that recognition, Eagle Scout status (for male volunteers),
faith-based organizational sponsorship, age, and race were all positively correlated with retention
and contributed to lower turnover. Higher turnover is associated with lower ranks in the units.
However, training, volunteer assignment into Pack units, and gender were negatively associated
with retention. For example, female volunteers were more likely to discontinue volunteering than
male volunteers. Finding gender differences is consistent with previous research; Wymer and
Samu (2002) find that, among volunteers in two U.S. mid-western cities, male volunteers
contributed more hours than female volunteers. The BSA context and our initial findings,
together with results from multivariate analysis, lead us to consider whether volunteers of
different genders responded differently to organizational management strategies, particularly in
an organization with such a strong gender imbalance.
<Table 2 about here>
We hypothesized that both recognition (H1) and training (H2) would be negatively
related to the likelihood of actual volunteer turnover. Results from Model 3 (see Table 3) show
18
dominant patterns arising from recognition and training, controlling for demographics, volunteer
roles and ranks, and type of sponsoring organization. We found supportive evidence for
Hypothesis 1: recognition through discretionary awards showed a negative effect on volunteer
turnover. In contrast to the bivariate analysis, we did not find evidence supporting Hypothesis 2
in Model 3. Although the model reported a negative effect for training on turnover, the effect
was not significant at the 0.1 level.
Results from Model 1 to Model 3 consistently suggest that white volunteers were less
likely to leave. In terms of unit type, volunteers serving the lower age group of scouts (i.e., Cub
Scouts) in the Pack units were more likely to leave, compared with those in other types of units.
Based on their volunteer position rank within the unit, in general, higher ranking volunteers were
more likely to stay than lower ranking volunteers. Both Model 1 and Model 3 reported that
female volunteers were more likely to leave.
We conducted post-hoc analyses in order to answer two questions: Do recognition and
training still show significant effects for more committed volunteers? Are the effects of
recognition and training moderated by gender? We took the subsample of male volunteers in
Model 4 to control for the effect of Eagle Scout status, which captures the most committed Boy
Scout volunteers. The results show that both recognition and training significantly decreased the
likelihood of turnover, after controlling for their Eagle Scout status. Moreover, Eagle Scout
status had a similar effect size as recognition and training. Specifically, for the male volunteer
group, Eagle Scouts were 29 percent less likely to turnover, discretionary award recipients were
also 29 percent less likely to leave, and training recipients were 21 percent less likely to leave.
We further explored a possible interaction effect between gender and training. Based on
Model 5 with the interaction term, findings show that (1) training (odds ratio = 0.779, p < 0.05)
19
was significantly associated with volunteer turnover, (2) the effect of training was also
moderated by gender (training*gender, odds ratio = 1.409, p < 0.05), and (3) the discretionary
award recipients (odds ratio = 0.659, p < 0.05) were 34 percent less likely to turnover.iv Overall,
the average marginal effects for gender (dy/dx = 0.037, p = 0.005) was significant at 0.01 level;
but the average marginal effects for training was only significant for the male population (dy/dx
= - 0.034, p = 0.022).
To illustrate the interaction effect of training and gender, Figure 1 shows that both male
and female volunteers who had not participated in training shared similar turnover results.
However, while training effectively reduced the probability of turnover for male volunteers, it
did not help retain female volunteers in the same way. Although we note that the magnitude of
the training effect as shown in Figure 1 is small, the finding implies that training for the male
volunteers at the BSA was more effective with regards to turnover as compared to their female
counterparts.
<Table 3 about here>
<Figure 1 about here>
As a robustness check for recognition, we estimated Model 3 to Model 5 using
“Confirmatory Award” instead of “Discretionary Award.” We have found no significant
evidence (at the 0.1 level) for a negative link between recognition through confirmatory awards
and actual turnover. This finding suggests that volunteer recognition through discretionary
awards is more effective than confirmatory awards for the general volunteer population. In
addressing the possible selection bias, we also estimated Model 3 on the subset of Eagle Scout
volunteers (N = 668), who are some of the most committed volunteers in the BSA. We found no
effect of discretionary awards and training among these top performers on actual turnover. The
20
results suggest that discretionary awards may be most effective for less committed volunteers
(results available upon request).
Finally, we hypothesized that using both training and recognition as a bundle of HR
practices would decrease actual volunteer turnover (H3). Results from the Additive Approach
Model (see Table 4) suggest that the presence of one more HR practice significantly lowers the
likelihood of actual turnover by 17 percent at the 0.05 level, holding the individual and
contextual factors constant. Alternatively, we tested the robustness by generating two dummy
variables from the Training and Recognition Index to estimate the effects of receiving either one
practice or both practices, compared with no HR practices at all, and we found similar effects for
the HR bundle. Since the two models—the Additive Model in Table 4 and Model 3 in Table 3—
are not nested, we compared the model fit using AIC and BIC (reported in both tables). Although
Model 3 reports a lower AIC value, the difference (0.428) between the two models suggests that
both are similar with regards to the goodness-of-fit. On the other hand, the BIC statistics show
positive evidence for the Additive Model (difference = 5.94) as the better fit (Kass & Raferty,
1995; Kuha, 2004).
Since the effect of training varied by gender, we further estimated two models for each
gender category separately and found that the bundling effect was largely driven by male
volunteers. While we found a significant bundling effect (odds ratio = 0.733; p < 0.01) for male
volunteers, the effect for female volunteers was inconclusive (odds ratio = 1.059, p = 0.67).
Discussion and Contributions
Human resource practices can be vital tools for mitigating turnover in the context of paid
employees. To date, we do not fully understand the mechanism between HR practices and
volunteer turnover (Alfes et al., 2016); this paper, therefore, provides an important step towards
21
this endeavor. Our findings show that HR practices play an essential role in volunteer retention;
consequently, this study contributes to existing literature and research in multiple ways.
First, this study contributes to the underdeveloped knowledge based regarding the impact
of HR practices on volunteer retention and, thus, answers calls for more research (Alfes et al.,
2016). Recognizing volunteers through discretionary awards and providing them with training,
as evident in this study, are two HR practices that improve volunteer retention. Particularly, we
found that recognizing volunteers through discretionary awards significantly decreased rates of
turnover. This finding contributes to the emerging literature investigating the impact of awards
on performance (Frey & Gallus, 2017; Frey & Gallus, 2018; Gallus, 2016), indicating that
discretionary awards positively impact subsequent performance. As we study awards in a non-
virtual field, we also contribute to closing this previously identified research gap (Gallus, 2016;
Zhang & Zhu, 2011).
A more nuanced picture emerged for training. Training has a positive effect on retention
for men, however, training was not an effective retention strategy for female volunteers in this
study. There are a few potential reasons for this finding. For instance, men may be more attuned
to the training delivery methods (e.g., online vs. face-to-face, the specific trainer, etc.) as well as
the content of the training. Moreover, men may be reinforced in their commitment to the Boy
Scouts after training. On the other hand, women may have certain expectations regarding their
volunteer role and find out during training that their expectations will not be met and, thus,
decide to drop out. Furthermore, the Boy Scouts is a highly gendered membership organization;
as such, the male-dominated culture of the organization may lead decision makers to design and
implement trainings that are implicitly more targeted towards the needs and expectations of men.
We discussed our findings with BSA staff members during focus groups and, overall, they
22
supported these potential reasons. Given the data sources we have, it is not entirely clear if one
of these potential reasons holds true from the perspective of the volunteers; future research is
needed to unpack this finding. We especially recommend studying volunteers’ evaluation of
training, since there is prior evidence that volunteers tend to evaluate training more negatively
when they do not perceive opportunities to express their opinions during training, which
ultimately influences their commitment and satisfaction (Costa, Chalip, Green, & Simes, 2006).
In addition to our above speculations , it may be likely that positive evaluations of training lead
to increased retention among men, but not among women.
We also explored the HR bundling perspective in the volunteer context and found that the
HR bundle index consisting of training and recognition had a significant effect on volunteer
turnover. We found that this model had slightly better model fit than testing for HR practices
separately; however, future research might want to replicate our study to verify this finding. In
order to be effective, HR bundles should be “interrelated and internally consistent” (MacDuffie,
1995, p. 204). Having internally consistent HR practices implies a certain level of intentionality
or strategic intent. Whereas we were able to test the additive impact of recognition and training
in this sample of Boy Scout volunteers, our data did not allow us to assess the level of strategic
intent behind the configuration of these HR practices. Further qualitative research is needed to
learn about the design of these HR practices.
Overall, our findings verify previous research in the volunteer context that investigated
self-reported intentions to quit/stay (Henderson & Sowa, 2017) or that used information provided
by organization members about retention problems (Hager & Brudney, 2008). Even though
training and recognition are prevalent HRM strategies used for volunteers as well as for paid
employees (Guo, Brown, Ashcraft, Yoshioka, & Dong, 2011; Walk, Schinnenburg, & Handy,
23
2014), there are other HR practices predominantly used among paid employees—such as
performance management (Walk & Kennedy, 2017)—that could potentially be adapted to and
tested in the volunteer context.
HR research has emphasized the importance of individual perceptions of HR practices on
the effectiveness of those practices (Nishii, Lepak, & Schneider, 2008). Given the use of
administrative data, we were not able to capture individual perceptions. However, we agree with
Studer and von Schnurbein (2013) that organizational and individual factors have to be
considered alongside each other to successfully manage volunteers. Whereas organizations
should focus on strategically developing HR practices to improve volunteer outcomes, they also
have to be mindful of volunteer attitudes, assumptions, and expectations towards volunteer work.
For instance, previous research indicates the importance of studying volunteer motives (Willems
& Walk, 2013) and volunteer satisfaction (Hurst, et al., 2017), especially in relationship to
volunteer retention (Garner & Garner, 2011; Gazley, 2013). In particular, Newton and colleagues
(2014) find that volunteer motives moderate the relationship between perceived training and
intentions to stay; therefore, future studies should complement administrative data with survey
data to obtain both objective and subjective information on volunteer motives and satisfaction,
HR practices, and other salient constructs. Survey data has two other potential benefits: first,
models integrating both organizational and individual factors will likely yield higher levels of
explained variance as compared to our findings. Second, survey instruments can also be designed
to further control for potential selection bias between awards and performance as discussed
earlier.v
As obtaining systematic information on volunteer retention has been regarded as “near to
impossible” (Henderson & Sowa, 2017, p. 7), this study also makes important methodological
24
contributions. In contrast to previous research, which mostly relied on cross-sectional, self-
reported data, we were able to use information on actual HR practices as well as turnover
information by drawing on administrative data collected by one organization over time, thus
overcoming previous methodological limitations (Podsakoff et al., 2003).
The use of an administrative data set with all the associated benefits, however, still
includes some unique limitations. For instance, we are limited to the information the
organization collects on a regular basis. For example, this data set does not capture whether or
not volunteers have children who are or have been Boy Scouts. Anecdotal evidence, however,
suggests that volunteers may leave once their children drop out of the program. Future studies
may want to capture that information to establish an empirical relationship. We also do not know
when volunteers received the awards (during 2016 or prior to that) and if volunteers received
more than one award. Similarly, it is unclear when training took place and if the volunteers
participated in one or multiple trainings. We encourage researchers to study the time effect
between recognition/training and subsequent turnover as one could speculate that the impact of
receiving an award or participating in training may wane over time. Moreover, just reporting
whether or not volunteers have received training does not give an indication of the perceived
quality of training, which may be an important indicator that future studies might want to include
(McMullen & Schellenberg, 2003). Another limitation pertains to the fact that we only studied
one nonprofit organization and, even though our data set is large and provides rich information,
findings may not be generalizable to other organizations or volunteer populations. We encourage
other researchers to replicate our findings in different contexts to explore whether or not similar
relationships emerge. As such, we would at least expect to see similar tendencies in other
25
gendered membership organizations (e.g., Girl Scouts, professional associations that have
majority female membership such as nursing, etc.).
Our study also has implications for practice. Understanding the factors that contribute to
volunteer turnover allows nonprofit organizations to develop strategies to improve volunteer
retention and to channel their resources into the most effective strategies. Similar to prior
research (Fallon & Rice, 2015; Kovacs & Black, 2000; Gallus, 2017), we found that recognizing
volunteers through awards is an effective management tool to increase future volunteer retention
rates. We recommend that volunteer managers pay particular attention to recognition strategies,
especially as these can be easily implemented and tend to be cost-effective. However, we
recommend being intentional about the kind and number of awards to be implemented, because
‘more is not always better.’ It is important to ensure that receiving an award is considered an
honor. Awards, as a recognition activity, can also backfire and negatively impact those who did
not receive awards; those individuals may feel discouraged in their volunteer commitment (Frey
& Gallus, 2017). Future studies might want to include measures of the importance of recognition
as an additional factor to tease out this effect.
Training, as an HR practice, might be more difficult to implement, as it requires more
time and monetary commitment from organizations. However, training provides volunteers with
the opportunity to increase their skills and knowledge and helps to prepare them for their
volunteer tasks (Saksida et al., 2016). We were only able to detect that women and men behave
differently after having received training due to a large enough data set. Generally, we encourage
volunteer managers to collect data on their volunteers in a systematic fashion and to critically
analyze the collected information for potentially adverse impacts.
26
In addition to the aforementioned suggestions for future research, some of the control
variables warrant further exploration. Particularly, we find that being affiliated with a faith-based
organization predicted lower turnover as compared to being affiliated with a secular
organization. Prior research has long shown that religious individuals (or “churchgoers”) are
more likely to volunteer (Ruiter & De Graaf, 2006) and that health benefits of volunteering are
more positive for religious volunteers (McDougle, Konrath, Walk, & Handy, 2016). Future
studies may want to disentangle this finding and look at potential variance within faith-based
sponsoring organizations. As previous research has established that members of different faith-
based groups have different volunteer rates (Cnaan & Curtis, 2013; Ruiter & De Graaf, 2006), it
is not yet clear if a similar pattern emerges for faith-based sponsoring organizations in the
context of the BSA.
Managing volunteers is a difficult undertaking. Despite the aforementioned limitations,
this study contributes to a more nuanced understanding of volunteer turnover by assessing the
effect of two HR practices. Particularly we show that recognition and training can have a positive
impact on future volunteer retention.
References
Alfes, K., Antunes, B., & Shantz, A. D. (2016a). The management of volunteers – what can
human resources do? A review and research agenda. The International Journal of Human
Resource Management, 28(1), 62-97.
Alfes, K., & Langner, N. (2017). Paradoxical leadership: Understanding and managing
conflicting tensions to foster volunteer engagement. Organizational Dynamics, 46(2), 96-
103.
27
Alfes, K., Shantz, A., & Bailey, C. (2016b). Enhancing volunteer engagement to achieve
desirable outcomes: What can non-profit employers do? VOLUNTAS: International
Journal of Voluntary and Nonprofit Organizations, 27(2), 595-617.
Alfes, K., Shantz, A., & Saksida, T. (2015). Committed to whom? Unraveling how relational job
design influences volunteers’ turnover intentions and time spent volunteering.
VOLUNTAS: International Journal of Voluntary and Nonprofit Organizations, 26(6),
2479-2499.
Bartram, T., Cavanagh, J., & Hoye, R. (2017). The growing importance of human resource
management in the NGO, volunteer and not-for-profit sectors. The International Journal
of Human Resource Management, 28(14), 1901-1911.
Bureau of Labor Statistics. (2016). Volunteering in the United States - 2015. U.S. Department of
Labor.
Clary, E. G., Snyder, M., & Ridge, R. (1992). Volunteers' motivations: A functional strategy for
the recruitment, placement, and retention of volunteers. Nonprofit Management and
Leadership, 2(4), 333-350.
Cnaan, R. A., & Cascio, T. A. (1998). Performance and commitment: Issues in management of
volunteers in human service organizations. Journal of Social Service Research, 24(3-4),
1-37.
Cnaan, R. A., & Curtis, D. W. (2013). Religious congregations as voluntary associations: An
overview. Nonprofit and Voluntary Sector Quarterly, 42(1), 7-33.
Cnaan, R. A., Handy, F., & Wadsworth, M. (1996). Defining who is a volunteer: Conceptual and
empirical considerations. Nonprofit and Voluntary Sector Quarterly, 25(3), 364-383.
28
Corporation for National and Community Service. (2015). State Rankings by Volunteer
Retention Rate. Retrieved from https://www.nationalservice.gov/vcla/state-rankings-
volunteer-retention-rate
Costa, C. A., Chalip, L., Green, B. C., & Simes, C. (2006). Reconsidering the role of training in
event volunteers’ satisfaction. Sport Management Review, 9(2), 165-182.
Delery, J. E., & Doty, D. H. (1996). Modes of theorizing in strategic human resource
management: Tests of universalistic, contingency, and configurational performance
predictions. Academy of Management Journal, 39(4), 802-835.
Delery, J. E. (1998). Issues of fit in strategic human resource management: Implications for
research. Human Resource Management Review, 8(3), 289-309.
Fallon, B., & Rice, S. (2015). Investment in staff development within an emergency services
organisation: Comparing future intention of volunteers and paid employees. The
International Journal of Human Resource Management, 26(4), 485-500.
Farmer, S. M., & Fedor, D. B. (1999). Volunteer participation and withdrawal. Nonprofit
Management and Leadership, 9(4), 349-368.
Frey, B. S., & Gallus, J. (2017). Towards an economics of awards. Journal of Economic Surveys,
31(1), 190-200.
Frey, B. S., & Gallus, J. (2018). Volunteer Organizations: Motivating with Awards. In R.
Ranyard (Ed.), Economic Psychology (pp. 273-286): John Wiley & Sons, Ltd.
Gallus, J. (2016). Fostering public good contributions with symbolic awards: A large-scale
natural field experiment at Wikipedia. Management Science, 63(12), 3999-4015.
29
Garner, J. T., & Garner, L. T. (2011). Volunteering an opinion: Organizational voice and
volunteer retention in nonprofit organizations. Nonprofit and Voluntary Sector Quarterly,
40(5), 813-828.
Gazley, B. (2013). Predicting a volunteer’s future intentions in professional associations: A test
of the Penner Model. Nonprofit and Voluntary Sector Quarterly, 42(6), 1245-1267.
Guo, C., Brown, W., Ashcraft, R., Yoshioka, C. F., & Dong, D. (2011). Strategic Human
Resource Management in Nonprofit Organizations. Review of Public Personnel
Administration, 31, 248-269.
Grube, J. A., & Piliavin, J. A. (2000). Role identity, organizational experiences, and volunteer
performance. Personality and Social Psychology Bulletin, 26(9), 1108-1119.
Hager, M., & Brudney, J. (2008). Management Capacity and Retention of Volunteers. In M.
Liao-Troth (Ed.), Challenges in Volunteer Management (pp. 9-27). Charlotte:
Information Age Publishing.
Handy, F., & Srinivasan, N. (2004). Valuing volunteers: An economic evaluation of the net
benefits of hospital volunteers. Nonprofit and Voluntary Sector Quarterly, 33(1), 28-54.
Henderson, A. C., & Sowa, J. E. (2017). Retaining Critical Human Capital: Volunteer
Firefighters in the Commonwealth of Pennsylvania. VOLUNTAS: International Journal
of Voluntary and Nonprofit Organizations, 1-26.
Hurst, C., Scherer, L., & Allen, J. (2017). Distributive justice for volunteers: Extrinsic outcomes
matter. Nonprofit Management and Leadership, 27(3), 411-421.
Huselid, M. A. (1995). The impact of human resource management practices on turnover,
productivity, and corporate financial performance. Academy of Management Journal,
38(3), 635-672.
30
Kass, R. and Raftery, A. (1995). Bayes Factors. Journal of the American Statistical Association,
90, 773–795.
Kim, Y.-I., Jang, S. J., & Johnson, B. R. (2016). Tying Knots With Communities: Youth
Involvement in Scouting and Civic Engagement in Adulthood. Nonprofit and Voluntary
Sector Quarterly, 45(6), 1113-1129.
Kovacs, P. J., & Black, B. (2000). Volunteerism and older adults: Implications for social work
practice. Journal of Gerontological Social Work, 32(4), 25-39.
Kuha, J. (2004). AIC and BIC comparisons of assumptions and performance. Sociological
Methods & Research, 33(2), 188-229.
Liao-Troth, M. A. (2001). Attitude differences between paid workers and volunteers. Nonprofit
Management and Leadership, 11(4), 423-442.
Macky, K., & Boxall, P. (2007). The relationship between ‘high-performance work practices’
and employee attitudes: an investigation of additive and interaction effects. The
International Journal of Human Resource Management, 18(4), 537-567.
McDougle, L., Konrath, S., Walk, M., & Handy, F. (2016). Religious and Secular Coping
Strategies and Mortality Risk among Older Adults. Social Indicators Research, 125(2),
677-694.
McMullen, K. E., & Schellenberg, G. (2003). Job quality in non-profit organizations (Vol. 1).
Ottawa: CPRN= RCRPP.
National Eagle Scout Association. (2017). Trail to Eagle. Retrieved from
http://www.nesa.org/trail.html
Nencini, A., Romaioli, D., & Meneghini, A. M. (2016). Volunteer motivation and organizational
climate: Factors that promote satisfaction and sustained volunteerism in NPOs.
31
VOLUNTAS: International Journal of Voluntary and Nonprofit Organizations, 27(2),
618-639.
Newton, C., Becker, K., & Bell, S. (2014). Learning and development opportunities as a tool for
the retention of volunteers: a motivational perspective. Human Resource Management
Journal, 24(4), 514-530.
Nishii, L. H., Lepak, D. P., & Schneider, B. (2008). Employee Attributions of the "why" of HR
Practices: Their Effects on Employee Attitudes and Behaviors, and Customer
Satisfaction. Personnel Psychology, 61, 503-545.
Pearce, J. L. (1983). Job attitude and motivation differences between volunteers and employees
from comparable organizations. Journal of Applied Psychology, 68(4), 646.
Podsakoff, P. M., MacKenzie, S. B., Lee, J.-Y., & Podsakoff, N. P. (2003). Common method
biases in behavioral research: a critical review of the literature and recommended
remedies. Journal of Applied Psychology, 88(5), 879-903.
Polson, E. C., Kim, Y. I., Jang, S. J., Johnson, B. R., & Smith, B. (2013). Being prepared and
staying connected: Scouting's influence on social capital and community involvement.
Social Science Quarterly, 94(3), 758-776.
Presti, A. L. (2013). The interactive effects of job resources and motivations to volunteer among
a sample of Italian volunteers. VOLUNTAS: International Journal of Voluntary and
Nonprofit Organizations, 24(4), 969-985.
Ripamonti, C. A., Pasquarelli, L., Ravasi, S., & Sala, F. C. (2016). Dropout of Hospital
Volunteers in Italy. VOLUNTAS: International Journal of Voluntary and Nonprofit
Organizations, 28(1), 44-68.
32
Rogers, S. E., Jiang, K., Rogers, C. M., & Intindola, M. (2016). Strategic human resource
management of volunteers and the link to hospital patient satisfaction. Nonprofit and
Voluntary Sector Quarterly, 45(2), 409-424.
Ruiter, S., & De Graaf, N. D. (2006). National context, religiosity, and volunteering: Results
from 53 countries. American Sociological Review, 71(2), 191-210.
Saksida, T., Alfes, K., & Shantz, A. (2016). Volunteer role mastery and commitment: can HRM
make a difference? The International Journal of Human Resource Management, 1-23.
Schein, E. H. (2010). Organizational culture and leadership (Vol. 2). San Francisco, CA: John
Wiley & Sons.
Schlesinger, T., Egli, B., & Nagel, S. (2013). ‘Continue or terminate?’ Determinants of long-
term volunteering in sports clubs. European Sport Management Quarterly, 13(1), 32-53.
Schwarz, N., & Oyserman, D. (2001). Asking questions about behavior: Cognition,
communication, and questionnaire construction. American Journal of Evaluation, 22(2),
127-160.
Simon, H. A. (1951). A formal theory of the employment relationship. Econometrica: Journal of
the Econometric Society, 19(3), 293-305.
Smith, S. L., & Grove, C. J. (2017). Bittersweet and Paradoxical: Disaster Response
Volunteering with the American Red Cross. Nonprofit Management and
Leadership, 27(3), 353-369.
Studer, S., & von Schnurbein, G. (2013). Organizational factors affecting volunteers: A literature
review on volunteer coordination. VOLUNTAS: International Journal of Voluntary and
Nonprofit Organizations, 24(2), 403-440.
33
Walk, M., & Kennedy, T. A. (2017). Making Nonprofits more Effective: Performance
Management and Performance Appraisals. In J. E. Sowa & J. Word (Eds.). The Nonprofit
Human Resource Management Handbook: From Theory to Practice. (Vol. 1, pp. 250-
267). New York, NY: Taylor & Francis.
Walk, M., Schinnenburg, H., Handy, F. (2014). Missing in Action: strategic human resource
management in German nonprofits. VOLUNTAS: International Journal of Voluntary and
Nonprofit Organizations, 25(4): 991-1021.
Willems, J., & Walk, M. (2013). Assigning volunteer tasks: The Relation between task
preferences and functional motives of youth volunteers. Children and Youth Services
Review, 35, 1030-1040.
Wu, Y., Li, C., & Khoo, S. (2016). Predicting future volunteering intentions through a self-
determination theory perspective. VOLUNTAS: International Journal of Voluntary and
Nonprofit Organizations, 27(3), 1266-1279.
Wymer Jr, W. W., & Samu, S. (2002). Volunteer service as symbolic consumption: Gender and
occupational differences in volunteering. Journal of Marketing Management, 18(9-10),
971-989.
Zhang, X., & Zhu, F. (2011). Group size and incentives to contribute: A natural experiment at
Chinese Wikipedia. American Economic Review, 101(4), 1601-1615.
34
Table 1. Descriptive Statistics, BSA, 2016-2017
+ The Eagle Scout variable is only applicable to male volunteers, within a subsample of 5,234 volunteers.
(1)
(2)
(3)
(4)
(5)
(6)
(7)
(8)
VARIABLES
N
mean
sd
min
max
skewness
kurtosis
p50
Dependent Variable
Turnover
7,595
0.127
0.333
0
1
2.236
6.000
0
Independent Variables
Training ( = 1)
7,595
0.335
0.472
0
1
0.699
1.488
0
Discretionary award ( = 1)
7,595
0.106
0.308
0
1
2.562
7.565
0
Training and Recognition Score
7,595
0.441
0.565
0
2
0.839
2.684
0
Eagle Scout ( = 1)+
5,234
0.203
0.402
0
1
1.480
3.192
0
Control Variables: Contextual Factors
Unit type: Pack ( = 1)
5,316
0.486
0.500
0
1
0.0549
1.003
0
Unit rank (ordinal)
5,297
2.545
1.109
1
5
0.581
2.205
2
Faith-based sponsor ( = 1)
5,316
0.698
0.459
0
1
-0.864
1.746
1
Control Variables: Individual Factors
Age
7,595
45.53
12.55
18
106
0.270
3.730
45
Gender (Female = 1)
7,595
0.311
0.463
0
1
0.817
1.667
0
Race (White = 1)
6,292
0.945
0.227
0
1
-3.924
16.40
1
35
Table 2. BSA Turnover Results – Differences of Means, 2016-2017
Note:
***p<0.001, **p<0.01, *p<0.05.
+ The Eagle Scout variable is only applicable to male volunteers, within a subsample of 5,234
volunteers.
† The unit rank variable ranges from 1 to 5, with 1 indicating the highest rank and 5 the lowest.
¶ Absolute values were used to report the Cohen’s d effect sizes. Following the convention, an
effect of 0.2 is considered as small; an effect of 0.5 is considered as medium.
No Turnover
Turnover
Effect size
VARIABLES
N
mean
N
mean
t-test
Cohen’s d ¶
[95% C.I.]
Independent Variables
Training ( = 1)
6,628
0.318
967
0.451
-8.201***
0.282
(0.21, 0.35)
Discretionary award ( = 1)
6,628
0.112
967
0.063
4.634***
0.160
(0.092, 0.23)
Training and
Recognition Score
6,628
0.430
967
0.514
-4.305***
0.148
(0.081, 0.22)
Eagle Scout ( = 1)+
4,647
0.207
587
0.170
2.061*
0.0903
(0.0044, 0.18)
Control Variables: Contextual Factors
Unit type: Pack ( = 1)
4,359
0.469
957
0.563
-5.272***
0.188
(0.12, 0.26)
Unit rank (ordinal)†
4,341
2.503
956
2.738
-5.968***
0.213
(0.14, 0.28)
Faith-based sponsor ( = 1)
4,359
0.708
957
0.654
3.288**
0.117
(0.047, 0.19)
Control Variables: Individual Factors
Age
6,628
45.96
967
42.54
7.957***
0.274
(0.21, 0.34)
Gender (Female = 1)
6,628
0.299
967
0.393
-5.908***
0.203
(0.14, 0.27)
Race (White = 1)
5,505
0.949
787
0.924
2.872**
0.109
(0.035, 0.18)
36
Table 3. Logistic regression results – Odds ratios reported (Dependent Variable = Turnover
result in 2017)
Model 1
Model 2
Model 3
Model 4
Model 5
VARIABLES
Individual
Contextual
Predictor
+ Control
Eagle Scout
Training
* Gender
Independent Variables
Training ( = 1)
0.888
0.787*
0.779*
(0.0778)
(0.0883)
(0.0850)
Training
1.409*
* Gender
(0.238)
Discretionary Award
0.654*
0.708†
0.659*
( = 1)
(0.114)
(0.141)
(0.115)
Control Variables: Contextual Factors
Unit type: Pack ( = 1)
2.036***
1.848***
1.616**
1.873***
(0.191)
(0.200)
(0.230)
(0.203)
Unit rank order = 2
3.206***
3.351***
2.953***
3.357***
(0.490)
(0.582)
(0.646)
(0.583)
Unit rank order = 3
6.768***
9.208***
8.504***
9.480***
(1.319)
(2.048)
(2.318)
(2.116)
Unit rank order = 4
2.234***
2.708***
2.610***
2.691***
(0.350)
(0.483)
(0.589)
(0.480)
Unit rank order = 5
3.180***
3.597***
3.053***
3.551***
(0.664)
(0.860)
(0.935)
(0.849)
Faith-based sponsor
0.834*
0.873
0.830
0.873
organization ( = 1)
(0.0651)
(0.0772)
(0.0949)
(0.0772)
Control Variables: Individual Factors
Age
0.980***
1.001
0.994
1.000
(0.00311)
(0.00375)
(0.00494)
(0.00376)
Gender (Female = 1)
1.532 ***
1.272**
1.089
(0.121)
(0.110)
(0.126)
Race (White = 1)
0.654**
0.711*
0.660*
0.712*
(0.0975)
(0.113)
(0.137)
(0.114)
Eagle Scout ( = 1)
0.713*
male volunteer
(0.120)
Constant
0.452***
0.0620***
0.0777***
0.152***
0.0830***
(0.0907)
(0.0105)
(0.0234)
(0.0602)
(0.0251)
Observations
6,292
5,297
4,310
2,912
4,310
Pseudo R2
0.0166
0.0296
0.0407
0.0385
0.0417
LR Chi2
78.56
148.1
165.4
99.47
169.5
AIC
4,672.664
4,867.514
3,925.659
2,508.523
3,923.568
BIC
4,699.652
4,913.539
4,002.083
2,580.242
4,006.361
37
Note: Standard errors in parentheses
*** p<0.001, ** p<0.01, * p<0.05, †p<0.1 (two-tailed).
Figure 1. Predictive Margins for Training and Gender with 95% Confidence Interval
38
Table 4. HR practices bundle: logistic regression results – Odds ratios reported
(Dependent Variable = Turnover result in 2017)
Note: Standard errors in parentheses
*** p<0.001, ** p<0.01, * p<0.05, †p<0.1 (two-tailed).
VARIABLES
Additive Approach
Model
Male Only
Subsample
Female Only
Subsample
Independent Variable
Training and Recognition Index
0.829*
0.733**
1.059
(0.0627)
(0.0683)
(0.143)
Control Variables: Contextual Factors
Unit type: Pack ( = 1)
1.879***
1.721***
2.215***
(0.202)
(0.239)
(0.393)
Unit rank order = 2
3.309***
2.949***
4.015***
(0.574)
(0.644)
(1.148)
Unit rank order = 3
9.165***
8.374***
11.43***
(2.039)
(2.280)
(4.551)
Unit rank order = 4
2.687***
2.626***
2.840***
(0.479)
(0.591)
(0.832)
Unit rank order = 5
3.595***
3.106***
4.557***
(0.859)
(0.951)
(1.766)
Faith-based sponsoring
organization ( = 1)
0.874
0.824*
0.943
(0.0772)
(0.0939)
(0.132)
Control Variables: Individual Factors
Age
1.003
0.999
1.007
(0.00348)
(0.00417)
(0.00714)
Gender (Female = 1)
1.292**
(0.111)
Race (White = 1)
0.712*
0.694†
0.737
(0.113)
(0.142)
(0.188)
Constant
0.0717***
0.107***
0.0492***
(0.0212)
(0.0394)
(0.0256)
Observations
4,310
2,912
1,398
Pseudo R2
0.0401
0.0365
0.0400
LR Chi2
163
94.21
58.56
AIC
3,926.087
2,509.779
1,425.615
BIC
3,996.143
2,569.545
1,478.043
39
Notes:
i Using the modified identifier, we were able to merge 6,538 entries in both data sets. The
remaining unmerged entries (1,059 entries from 2016 and 498 entries from 2017) were
investigated to ensure proper merging procedures. Initially, some entries could not be merged as
the more complex code—due to duplication in one data set—did not correspond to the identifier
in the other data set. Moreover, for some entries, the age variable in 2017 was not properly coded
and, as such, individuals retained the same age across years, which meant an automatic merge
did not happen. After investigating these issues manually, an additional 91 records could be
merged.
ii Considering wide-ranging volunteer ages, we ran the same set of models on several
subsamples, for example, excluding observations at the top and the bottom 1 or 5 percent in
terms of age values. We did not find qualitative differences except for the discretionary award
variable in Model 3 as we excluded the top and the bottom 5 percent. The effect of discretionary
awards became insignificant possibly as a significant number of discretionary award recipients
were younger than 20 years old (at the bottom 5%).
iii About 24 percent of volunteers received either a confirmatory award or a discretionary award.
iv Odds ratios calculated using Stata.
v We acknowledge that experiments where participants are randomized into control and treatment
groups are superior to survey data in order to mitigate selection bias.
