Volunteering Predicts Health Among Those Who Value Others: Two National Studies

Abstract

Objective:
The purpose of these studies was to examine the role of positive views of other people in predicting stress-buffering effects of volunteering on mortality and psychological distress.

Method:
In Study 1, stressful life events, volunteering, and hostile cynicism assessed in a baseline Detroit-area survey (N = 846) predicted survival over a 5-year period, adjusting for relevant covariates. In Study 2, stressful life events, volunteering, and world benevolence beliefs assessed in a baseline national survey (N = 1,157) predicted psychological distress over a 1-year period, adjusting for distress at baseline.

Results:
In Study 1, a Cox proportional hazard model indicated that for individuals low in cynicism, stress predicted mortality at low levels of volunteering but not at high levels of volunteering. This effect was not present among those high in cynicism. In Study 2, multiple regression analysis revealed that among individuals high in world benevolence beliefs, stress predicted elevated distress at low levels of volunteering but not at high levels of volunteering. This effect was absent for those lower in world benevolence beliefs.

Conclusions:
Consistent with prior research on helping behavior, these studies indicate that helping behavior can buffer the effects of stress on health. However, the results of these studies indicate that stress-buffering effects of volunteering are limited to individuals with positive views of other people. Not all individuals may benefit from volunteering, and health-promotion efforts seeking to draw on health benefits of helping behavior may need to target their approach accordingly.

Full text

Volunteering Predicts Health Among Those Who Value Others:
Two National Studies
Michael J. Poulin
University at Buffalo
Objective: The purpose of these studies was to examine the role of positive views of other people in
predicting stress-buffering effects of volunteering on mortality and psychological distress. Method: In
Study 1, stressful life events, volunteering, and hostile cynicism assessed in a baseline Detroit-area
survey (N⫽846) predicted survival over a 5-year period, adjusting for relevant covariates. In Study 2,
stressful life events, volunteering, and world benevolence beliefs assessed in a baseline national survey
(N⫽1,157) predicted psychological distress over a 1-year period, adjusting for distress at baseline.
Results: In Study 1, a Cox proportional hazard model indicated that for individuals low in cynicism,
stress predicted mortality at low levels of volunteering but not at high levels of volunteering. This effect
was not present among those high in cynicism. In Study 2, multiple regression analysis revealed that
among individuals high in world benevolence beliefs, stress predicted elevated distress at low levels of
volunteering but not at high levels of volunteering. This effect was absent for those lower in world
benevolence beliefs. Conclusions: Consistent with prior research on helping behavior, these studies
indicate that helping behavior can buffer the effects of stress on health. However, the results of these
studies indicate that stress-buffering effects of volunteering are limited to individuals with positive views
of other people. Not all individuals may benefit from volunteering, and health-promotion efforts seeking
to draw on health benefits of helping behavior may need to target their approach accordingly.
Keywords: trust, hostility, volunteering, stress, mortality
Supplemental materials: http://dx.doi.org/10.1037/a0031620.supp
To a surprising degree, providing help or support to others
appears to benefit the helper. Volunteering or providing care for
close others predicts reduced morbidity and mortality (e.g., Brown,
Nesse, Vinokur, & Smith, 2003;Brown et al., 2009;O’Reilly,
Connolly, Rosato, & Patterson, 2008) and increased psychological
well-being (e.g., Brown, Brown, House, & Smith, 2008;Poulin et
al., 2010; see Post, 2007, for a recent overview). Diverse mecha-
nisms have been proposed to explain these effects, with research-
ers noting that volunteering facilitates social integration, distrac-
tion from one’s troubles, a sense of meaning, self-efficacy, positive
mood, and physical activity, all of which may promote health
(Midlarsky & Kahana, 1994;Oman, Thoresen, & McMahon,
1999).
The apparent benefits of engaging in prosocial behavior suggest
the possibility of interventions or policies directed toward encour-
aging volunteering. However, volunteering is not always benefi-
cial: In a review of the literature on volunteering and health, Oman
(2007) reported the results of several studies that indicate that the
benefits of volunteering depend on levels of other social contact
volunteers have, such as visiting friends or attending religious
services. Specifically, the evidence from these studies indicated
that volunteering is most effective among individuals who have the
greatest amount of social contact.
Why volunteering is more beneficial for some individuals than
for others is currently unknown. The present research attempted to
address this problem, examining the links between helping and
health through the lens of the caregiving behavioral system and
testing the prediction that helping valued others, specifically, buf-
fers the association between stress and health.
Prosocial Behavior, Stress-Buffering, and
Valuing Others
Recent evidence suggests that many of the benefits of volun-
teering may stem from the stress-buffering role of the caregiving
behavioral system, a set of emotions, cognitions, and physiological
states that evolved to facilitate actions to benefit valued others
(Brown & Brown, 2006;Goetz, Keltner, & Simon-Thomas, 2010).
Activation of this system leads to the emotional state of empathic
concern or compassion, defined as feelings of warmth and tender-
ness toward an individual in need (Batson, Fultz, & Schoenrade,
1987;Goetz et al., 2010), as well as physiological stress reduction
by way of parasympathetic nervous system activation (Goetz et al.,
2010) and neurochemicals such as oxytocin, prolactin, and endog-
enous opioids (Brown & Brown, 2006).
Consistent with the effects of the caregiving behavioral system,
helping or supporting others predicts reduced associations between
stress and mortality (Krause, 2006;Okun, August, Rook, & New-
som, 2010;Poulin, Brown, Dillard, & Smith, in press) or depres-
sion (Brown et al., 2008). Moreover, in the laboratory, helping
Correspondence concerning this article should be addressed to Michael
J. Poulin, Department of Psychology, University at Buffalo, State Univer-
sity of New York, Park Hall 206, Buffalo, New York 14260. E-mail:
mjpoulin@buffalo.edu
Health Psychology © 2013 American Psychological Association
2013, Vol. 32, No. 3, 000 0278-6133/13/$12.00 http://dx.doi.org/10.1037/a0031620
1
AQ: au
AQ: 1
AQ: 8
tapraid5/zg1-hea/zg1-hea/zg100313/zg12786d13z
xppws S⫽1 1/9/13 4:17 Art: 2012-1369
APA NLM

behavior leads to reductions in physiological responses to stress
(e.g., Floyd et al., 2007). Further evidence that the caregiving
behavioral system substantially accounts for the health benefits of
volunteering comes from a recent large study of volunteering and
mortality: Volunteering predicted reduced mortality only when it
was motivated by other-focused motives, which are at the core of
the caregiving behavioral system (Konrath, Fuhrel-Forbis, Lou, &
Brown, 2012).
If the caregiving behavioral system does explain much of the
beneficial effects of volunteering, volunteering should have the
greatest benefits among individuals for whom this system is active.
Prior research on the caregiving behavioral system suggests that
this should be true specifically for individuals who value the
welfare of those they help. The caregiving behavioral system-
linked emotion of empathic concern most strongly predicts helping
behavior toward family members (Maner & Gailliot, 2007),
friends (Schlenker & Britt, 2001), and people perceived as similar
to oneself (Stürmer, Snyder, & Omoto, 2005). In the absence of a
prior relationship with a person in need, simply viewing that
person as good and trustworthy facilitates empathic concern (Bat-
son, Eklund, Chermok, Hoyt, & Ortiz, 2007). Similarly, securely
attached individuals, who have positive views of others and the
self, are more likely than others to experience empathic concern
toward strangers (Gillath et al., 2005).
Together, prior research on volunteering and the caregiving
behavioral system raise the possibility that individual differences
in views of others might predict caregiving behavioral system
activation, and therefore that the stress-buffering benefits of vol-
unteering may be strongest among those with the most positive
views of other people. For example, volunteering may not be
beneficial for individuals high in hostile cynicism, who tend to
believe that others are selfish and greedy (e.g., Smith, 1994), but
may be quite beneficial for those with high world benevolence
beliefs, who believe that people are more good than bad (Janoff-
Bulman, 1989).
The Present Research
The present research tested the prediction, derived from research
on the caregiving behavioral system, that having positive views of
others will predict stress-buffering features of prosocial behavior.
Study 1 examined the stress-buffering implications of helping
behavior for physical health, examining stressful events and vol-
unteering as predictors of mortality risk. Specifically, Study 1
tested the moderating role of hostile cynicism. Study 2 tested the
stress-buffering features of helping behavior for psychological
adjustment following stressful life events, examining stressful
events and volunteering as predictors of psychological distress. To
test the moderating role of views of others, Study 2 specifically
examined the contributions of world benevolence beliefs. Predic-
tions for these studies were as follows:
1. For Study 1: Volunteering should more strongly predict
reduced associations between life stress and mortality risk among
individuals low in hostile cynicism than among those high in
hostile cynicism.
2. For Study 2: Volunteering should more strongly predict
reduced associations between life stress and psychological distress
among individuals high in world benevolence beliefs than among
those low in world benevolence beliefs.
Study 1
Method
Sample. Data were examined from the Changing Lives of
Older Couples (CLOC) study, a large prospective study of 1,532
members of married couples in which the husband was at least 65
years of age from the Detroit Standard Metropolitan Statistical
Area (see Carr et al., 2000, for a complete report). Of those
selected for participation in the CLOC study, 65% agreed to
participate, a response rate consistent with response rates in other
Detroit-area studies (Carr et al., 2000).
The CLOC study as a whole was primarily designed not to
examine predictors of mortality, but to assess predictors of a
surviving spouse’s experience of widowhood. Thus, mortality data
were not collected on focal respondents, but on their spouses.
However, more than one half of the CLOC sample (n⫽846)
consisted of married couples in which both members were focal
respondents and, thus, for whom baseline data can be used to
predict mortality of each individual. Therefore, members of this
subsample composed the sample for the present study.
Procedure. All participants completed a face-to-face baseline
interview, all of which were conducted over an 11-month period in
1987 and 1988. Subsequently, mortality of participants was mon-
itored over a 6-year period by checking obituaries in three Detroit-
area newspapers and monthly death-record tapes obtained from the
State of Michigan. If a participant died, the surviving spouse (if
any) was reinterviewed 3 months after the deceased spouse’s
death. These procedures were approved by the Institutional Re-
view Board of the University of Michigan, and all participants
provided informed consent in writing.
Measures.
Mortality data. The death of any participant was recorded
during the course of the study. In addition, the researchers created
a “gap” variable that represented how many months passed be-
tween the baseline interview and the follow-up interview of the
surviving spouse (at 3 months postdeath). This variable was used
as an indicator of time of postbaseline survival and was used to
create a variable representing age at death.
Stressful events. Recent life stress was measured at baseline
by asking respondents whether they had experienced any of the
following events in the past 12 months: serious, non–life-
threatening illness, burglary, job loss, financial difficulties, or
death of a family member. The number of these events (potentially
ranging from zero to five) was used as an index of recent stressful
events.
Volunteering. At baseline, respondents also reported the num-
ber of hours over the past year they had engaged in volunteer
(unpaid) work “for a church or other religious organization, for a
political group, a senior citizens’ group, or for any other type of
organization.” Hours were reported as the total amount of time
respondents spent in all of these activities combined on a scale
with the following anchors: 0 (no volunteering),1(less than 20
hr),2(20 –39 hr), 3 (40 –79 hr), 4 (80 –159 hr), and 5 (160 hr or
more).
Hostile cynicism. Cynicism was assessed at baseline using a
shortened form of the Cook–Medley hostility scale (Barefoot,
Dodge, Peterson, & Dahlstrom, 1989), a measure that has been
characterized as predominantly reflecting hostile cynicism about
2POULIN
tapraid5/zg1-hea/zg1-hea/zg100313/zg12786d13z
xppws S⫽1 1/9/13 4:17 Art: 2012-1369
APA NLM

other people (e.g., Smith, 1994). This shortened scale consisted of
four items: “When I look back on what has happened to me, I feel
cheated”; “I don’t seem to get what should be coming to me”;
“Other people always seem to get the breaks”; and “Getting too
attached to people is unwise.” These were all scored using the
same 4-point scale (1 ⫽not true at all,4⫽very true), and the
mean score was standardized and used as an index of hostile
cynicism (␣⫽.70).
Control variables. Several other variables were assessed at
baseline that were not directly related to the caregiving behavioral
system-related hypotheses of the present study, but were important
potential confounds of the hypothesized associations between key
study variables and mortality. These control variables included
demographics (age, gender, race, and education), health status
(self- and interviewer-rated), personality traits, and social network
characteristics (see Supplemental Material for details).
Self-rated physical health was measured by satisfaction with
health, functional health, and health behaviors. Satisfaction with
health was assessed with three items: the extent to which partici-
pants rated their health as excellent, good, fair, or poor; the extent
to which they thought their health limited their daily activities; and
the extent to which they were satisfied with their health. The mean
score of these items was standardized and used as an index of
satisfaction with health (␣⫽.84). The functional health index
measured the extent to which participants had difficulty with a
variety of tasks, including walking, climbing stairs, bathing, and
housework. The mean score of these items was standardized and
was used as an index of functional difficulties (␣⫽.78). Health
behaviors included measures of smoking (number of cigarettes per
day), drinking (number of drinks in the past month), and exercise
(frequency of taking walks or other forms of exercise).
Interviewers answered several questions about the respondents’
condition, three of which directly relate to physical health and
functioning. First, interviewers explicitly rated the respondent’s
apparent health (1 ⫽gravely or terminally ill,5⫽excellent).
Second, interviewers reported how much respondents had diffi-
culty walking at home (1 ⫽none,5⫽could not do at all). Third,
interviewers reported how much respondents found the interview
tiring (1 ⫽not tiring,4⫽very tiring).
Respondents also completed measures of their mental health and
well-being. Mental health was assessed in the form of standardized
scales for depression (Center for Epidemiological Studies Depres-
sion scale; Radloff, 1977;␣⫽.84) and subjective well-being
(Bradburn, 1969;␣⫽.79).
Individual difference variables included modified scales from
the NEO Five-Factor Personality Inventory (i.e., Extraversion,
␣⫽.53; Agreeableness, ␣⫽.62; Conscientiousness, ␣⫽.73;
Openness to Experience, ␣⫽.51; and Neuroticism, ␣⫽.70;
Costa & McCrae, 1992), as well as measures of self-esteem (␣⫽
.72; Rosenberg, 1965) and internal locus of control (␣⫽.71;
Levenson, 1973).
Respondents reported on several features of their social net-
work, including received support (instrumental and emotional) and
social contact (formal and informal). Received instrumental sup-
port was assessed with a single item: “If you and your husband
[wife] needed extra help with general housework or home main-
tenance, how much could you count on friends or family members
to help you?” (1 ⫽not at all,4⫽a great deal). Received
emotional support was assessed as the extent to which three
sources (spouse, children, and friends/relatives) provided love and
were willing to listen (1 ⫽not at all,5⫽a great deal); the mean
score of these items was standardized and was used as an index of
received emotional support (␣⫽.66).
Informal social contact was assessed with the following three
questions: “In a typical week, about how many times do you talk
on the phone with friends, neighbors, or relatives?”; “How often
do you get together with friends, neighbors, or relatives and do
things like go out together or visit in each other’s homes?”; and
“How often do you go out socially, by yourself, or with people
other than your husband [wife]?” These items were scored on a
5-point scale (1 ⫽not at all,5⫽a great deal), and the mean score
of these items was standardized and was used as an index of
informal social contact (␣⫽.51). Formal social contact was
assessed with two questions: “How often do you attend meetings
or programs of groups, clubs, or organizations that you belong to?”
and “How often do you usually attend religious services?” These
items were scored on a 5-point scale (1 ⫽never,5⫽more than
once a week), and the mean score of these items was standardized
and was used as an index of informal social contact (␣⫽.51).
Results
Sample demographics and characteristics. Exactly 50.0%
of the sample was female, and the ethnic composition of the
sample was 87.7% White, 11.7% African American, and less than
1.0% other ethnicity. The mean age of the sample was 71 years
(range ⫽34 –93 years). Of the subsample of 846 respondents, 134
died over the course of the study. A minority of respondents (35%)
reported having engaged in volunteering for an average of less
than 20 hr. A majority of respondents (70%) had experienced no
recent stressful life events, and 26% had experienced one such
event and 4% had experienced two or three events; the latter were
grouped for the present analyses. Descriptive statistics and corre-
lations for key variables can be found in supplementary online
materials (see Supplementary Table 1).
Associations between prosocial behavior and mortality.
Associations among stress, volunteering, hostile cynicism, and
mortality were examined with Cox proportional hazard modeling
(Cox, 1972) using the stcox module in Stata 11.0 (Stata Corp.,
College Station, TX). Cox modeling is a statistical procedure that
tests predictors of mortality risk over time while requiring minimal
assumptions about underlying distributions. This procedure yields
hazard ratios (HRs), representing the degree of change in mortality
risk over time for a unit change in a predictor. For the present
study, time was best represented by age (Korn, Graubard, &
Midthune, 1997). Analyses adjusted for within-couple shared vari-
ance using Stata’s “cluster” option, a modified form of the Huber–
White sandwich variance estimate (Huber, 1967;White, 1980) that
accounts for fact that although individuals within clusters are not
independent, the clusters are (Rogers, 1993;Williams, 2000).
To better estimate the unique associations of key variables with
mortality, analyses controlled for several potential confounds,
screened for entry into the model as controls in four blocks: first,
demographics (age, gender, education, non-White race); second,
mental and physical health (satisfaction with health, functional
difficulties, smoking, drinking, exercise, interviewer ratings of
apparent health and functioning, depression, and subjective well-
being); third, social network characteristics (informal and formal
3
VOLUNTEERING AND VALUING OTHERS
tapraid5/zg1-hea/zg1-hea/zg100313/zg12786d13z
xppws S⫽1 1/9/13 4:17 Art: 2012-1369
APA NLM

social contact, received instrumental and emotional support); and
fourth, individual differences (five-factor model personality traits,
self-esteem, and internal locus of control). Any significant vari-
ables from each block were entered into the Cox regression models
before volunteering, stress, or hostile cynicism.
The study hypothesis implied a three-way interaction among
volunteering, stressful events, and hostile cynicism. To test this
hypothesis, regression models included all three variables, all three
two-way product-term interactions among those variables, and the
three-way product-term interaction. In addition, because interac-
tion terms can be spurious in the presence of nonlinear effects of
constituent variables (Cohen, Cohen, West, & Aiken, 2002), ex-
ploratory models also tested for quadratic terms of volunteering,
stress, and hostile cynicism; no quadratic effects were detected.
The final model was examined for validity of assumptions of the
Cox model (e.g., proportionality of hazards, normality and ho-
moscedasticity of residuals) and for multivariate outliers; no prob-
lems were detected. Results of this model are reported in Table 1.
Examining the three-way interaction (Hostile Cynicism ⫻
Stress ⫻Volunteering) revealed that hostile cynicism and stressful
events significantly interacted with volunteering (interaction
HR ⫽1.18, p⫽.04). Simple slopes analyses indicated that the
Volunteering ⫻Stress interaction was significant at low (mean –
1SD) levels of hostile cynicism (interaction HR ⫽0.73, p⫽.049),
but not at high (mean ⫹1SD) levels of hostile cynicism (inter-
action HR ⫽1.03, p⫽.69). Next, the Volunteering ⫻Stress
interaction was examined at low (mean – 1 SD) levels of hostile
cynicism. Simple slopes analyses showed that, as hypothesized,
volunteering buffered the stress–mortality association at low levels
of hostile cynicism. Stress did not significantly predict mortality at
high levels of volunteering (HR ⫽0.73, p⫽.19), but stress
predicted a marginally higher rate of mortality at low levels of
volunteering (HR ⫽1.37, p⫽.07). This pattern is illustrated in
Figure 1.
Alternative models. To test the robustness of these results, I
examined a series of alternative models. First, because the cross-
sectional, self-report nature of the data suggested the need for
control variables, especially to isolate the effects of stressful events
from individual differences such as prior health and personality,
the model reported in Table 1 was examined without control
variables. In the resulting model, the three-way interaction (Hostile
Cynicism ⫻Stress ⫻Volunteering) was no longer significant
(HR ⫽1.08, p⫽.32). One plausible interpretation for this pattern
of results is that the stress-buffering effect of volunteering is
conditioned on baseline health status. That is, individuals with
better baseline health may not require stress-buffering as much as
do individuals with poorer baseline health, but adjusting for vari-
ables associated with baseline health (satisfaction with health, tired
appearance, age, and even gender) allows the stress-buffering
effect to be revealed. To probe this possible explanation, I exam-
ined the three-way interaction (Hostile Cynicism ⫻Stress ⫻
Volunteering) independently for individuals below the median
(n⫽417) and above the median (n⫽429) in satisfaction with
health. Results indicated that the three-way interaction was mar-
ginally significant for the less healthy individuals (HR ⫽1.15, p⫽
.06), but not at all significant for the healthiest individuals (HR ⫽
0.86, p⫽.78), suggesting that the stress-buffering effects of
volunteering may indeed be conditioned on baseline health.
Second, although the recommended way to evaluate interactions
with continuous moderators is to recenter the moderator, as was
done in Figure 1 (Cohen et al., 2002), an additional analysis
explored the Volunteering ⫻Stress interaction by dichotomizing
cynicism at the median and examining the Volunteering ⫻Stress
interaction in the two resulting groups. Consistent with the results
derived from recentering, in the low-cynicism group (n⫽408), the
interaction was significant (HR ⫽0.71, p⫽.03), but in the
high-cynicism group (n⫽435), it was not (HR ⫽1.01, p⫽.92).
Within the low-cynicism group, volunteering was also dichoto-
mized and the association between stress and mortality was exam-
ined in the two resulting groups. Among low-cynicism individuals
who did not engage in volunteering in the past year (n⫽243),
stress was a significant predictor of mortality (HR ⫽1.44, p⫽
.02), but among low-cynicism individuals who did engage in
volunteering in the past year (n⫽165), stress did not predict
mortality (HR ⫽0.71, p⫽.13).
Finally, to evaluate the distinctiveness of volunteering from
social engagement more generally, a series of models examined
associations of formal and informal social engagement with mor-
tality. There were no main effects of these variables, nor did they
significantly interact with stress; moreover, there were no signif-
icant three-way interactions of these variables with volunteering
and stress or with cynicism and stress (ps⬎.05).
Discussion
The results of Study 1 provided support for the study hypothesis.
Specifically, volunteering buffered the association between stress-
ful events and mortality, but only among individuals with rela-
tively positive views of others—that is, those low in hostile cyn-
icism. This pattern of findings is consistent with the notion that
positive views of others serve to facilitate the stress-buffering role
of prosocial behavior. Interestingly, the results of Study 1 also
provide tentative evidence that this stress-buffering role may be
particularly important among individuals who are less healthy at
baseline. However, these findings are limited in several respects.
First, even though analyses controlled for several important indi-
Table 1
Study 1: Cox Proportional Hazard Model With Volunteering,
Stress, and Hostile Cynicism as Predictors of Mortality
(N⫽843)
Variable HR 95% CI
Age 0.71
ⴱⴱⴱ
[0.67, 0.75]
Female gender 1.83
ⴱⴱ
[1.23, 2.73]
Satisfaction with health 0.55
ⴱⴱⴱ
[0.47, 0.64]
Interview tiring 1.41
ⴱⴱⴱ
[1.16, 1.70]
Neuroticism 1.24
ⴱ
[1.02, 1.51]
Stressful events 1.03 [0.88, 1.21]
Volunteering 0.97 [0.80, 1.16]
Hostile cynicism 1.14 [0.96, 1.36]
Volunteering ⫻Stress 0.86 [0.72, 1.04]
Cynicism ⫻Stress 1.03 [0.90, 1.19]
Cynicism ⫻Volunteering 1.03 [0.86, 1.24]
Cynicism ⫻Volunteering ⫻Stress 1.18
ⴱ
[1.01, 1.38]
Note.HR⫽hazard ratio. All variables used in interactions are standard-
ized. Interaction terms represent product terms of standardized constituent
variables. Model fit: Wald ␹
2
(12, 831) ⫽210.19, p⬍.001.
ⴱ
p⬍.05.
ⴱⴱ
p⬍.01.
ⴱⴱⴱ
p⬍.001.
4POULIN
T1
F1
AQ: 9
tapraid5/zg1-hea/zg1-hea/zg100313/zg12786d13z
xppws S⫽1 1/9/13 4:17 Art: 2012-1369
APA NLM

cators of baseline health status, it is not clear that these variables
identified all potential confounds of the observed stress-buffering
effects, meaning that it is possible that the stress-buffering effects
of hostile cynicism and volunteering may have reflected the action
of some other, unmeasured, variable. Similarly, the fact that all
self-report measures were assessed only at one time point prevents
any more direct examination of possible physiological or psycho-
logical mechanisms for the observed results. Finally, the measure
of views of others used in Study 1— hostile cynicism—is arguably
an incomplete measure because by design it is more sensitive to
very negative views of others than to positive views of others.
Study 2 was designed to address these issues.
Study 2
The goal of Study 2, like that of Study 1, was to test the
hypothesis that positive views of others predict stress-buffering
characteristics of prosocial behavior. However, unlike Study 1,
Study 2 focused on a more proximal outcome of stress buffering
than mortality—namely, psychological distress. In addition, Study
2 used a prospective design, with psychological distress and views
of others assessed before the occurrence of stressful events and
prosocial behaviors. By assessing variability in the dependent
variable (distress) and a key moderator (view of others) before the
assessment of stressful events, this approach reduces, although
could not eliminate, concerns that any stress-buffering effects
reflect confounding of predictors and outcomes. Finally, the mea-
sure of views of others used in Study 2—world benevolence
beliefs— differed from that used in Study 1 in that it was more
sensitive to both positive and negative views of others. Together,
these aspects of Study 2 were aimed at strengthening and extend-
ing the findings of Study 1.
Method
Sample and procedure. The study sample (N⫽1,157) com-
prised adult members of a nationally representative, Web-enabled
research panel established by Knowledge Networks, Inc., who
were randomly selected to participate in the present study (see
Silver, Holman, McIntosh, Poulin, & Gil-Rivas, 2002, for panel
details). Participants completed two waves of surveys conducted
approximately 1 year apart. At each wave, respondents were
notified via password-protected e-mail accounts that a survey was
available. The first survey (Wave 1) was fielded between Decem-
ber 28, 2006, and January 18, 2007, to 2,142 people, and 1,613
completed it (a 73.5% completion rate). The second survey (Wave
2) was fielded between December 28, 2007, and February 18,
2008, to all Year 1 respondents and was completed by 1,157
(71.7% retention). All procedures were approved by the Institu-
tional Review Board of the University of California, Irvine, and all
participants provided informed consent in writing.
Measures.
Psychological distress. Distress was assessed at Wave 1 and
Wave 2 using the 18-item Brief Symptom Inventory (Derogatis &
Savitz, 2000), a measure of general psychological distress that
includes depression, anxiety, and somatization subscales. Internal
consistency was excellent at both waves (␣s⬎.92).
Figure 1. Study 1: Product-limit estimator survival probability curves predicted for low levels (M–1SD)of
hostile cynicism. Curves represent low or high (M⫾1SD) numbers of stressful events in the past year, graphed
separately for low or high (M⫾1SD) amount of volunteering in the past year. Vertical axis represents value
of survival estimator function.
5
VOLUNTEERING AND VALUING OTHERS
tapraid5/zg1-hea/zg1-hea/zg100313/zg12786d13z
xppws S⫽1 1/9/13 4:17 Art: 2012-1369
APA NLM

Stressful life events. Exposure to stressful events was assessed
both at Wave 1 and Wave 2 as the total number of 37 negative
events (e.g., serious illness or injury, natural disaster) a respondent
reported experiencing (see Silver et al., 2002). The Wave 1 mea-
sure assessed events that occurred anytime up to that point in
participants’ lives. The Wave 2 measure specifically requested
respondents to report stressful events in the past 12 months,
yielding a measure of recent stressful events or those experienced
between Waves 1 and 2.
Volunteering hours. At Wave 2, respondents reported how
much time they had spent volunteering in the past year (i.e., since
Wave 1) in response to the following item: “In the past 12 months,
about how much time, if any, have you spent doing volunteer work
for religious, educational, health-related or other charitable orga-
nizations?” (1 ⫽none,2⫽1–50 hr,3⫽51–100 hr,4⫽101–200
hr,5⫽more than 200 hr).
Benevolence beliefs. Perceived benevolence of the world was
assessed at Wave 1 using a shortened form of the Benevolence of
the World subscale of Janoff-Bulman’s (1989) World Assump-
tions Scale. This scale consists of six items (e.g., “People are
basically kind and helpful,” “Human nature is basically good”; 1 ⫽
strongly disagree,5⫽strongly agree). The mean of these items
was used as an index of benevolence beliefs and exhibited good
internal consistency (␣⫽.85).
Control variables. To test for the presence of confounding in
the data, I assessed several plausibly confounding variables, not
related to the function of the caregiving behavioral system, as
potential controls at Wave 1. These variables included personality
traits from the five-factor model of personality (Neuroticism, Ex-
troversion, Openness, Agreeableness, Conscientiousness), as-
sessed using the Ten-Item Personality Inventory (Gosling, Rent-
frow, & Swann, 2003), as well as two variables associated with
mental health: number of prior mental illnesses (anxiety, depres-
sion, both, or neither) and number of lifetime stressful events. In
addition, the number of social groups individuals in which had
participated in the past year (e.g., school club, sports team, neigh-
borhood association) was assessed.
Results
Sample demographics and characteristics. The Wave 1
sample was 51.2% female, with a mean age of 46.0 years (range ⫽
18 –91 years). In total, 73.3% identified as White, 11.2% identified
as Hispanic, 9.7% identified as Black, and 5.7% were of other
ethnicities. Most respondents (54%) engaged in some volunteering
between Waves 1 and 2, for an average of approximately 50 hr.
Most of the sample (62%) also experienced some sort of stressful
life event between waves, for a sample-wide average of 1.48 such
events. Descriptive statistics and correlations for key variables can
be found in supplementary online materials (see Supplementary
Table 2).
Nonresponse and attrition analysis. A multiple logistic re-
gression examined differences between those who participated at
Wave 1 and those who did not respond. Compared with respon-
dents, those who did not respond to the Wave 1 survey were
younger and less likely to be African American (ps⬍.001), but
did not differ in terms of income or other ethnic differences. A
second multiple logistic regression examined differences between
those who participated in both waves compared with those who
completed Wave 1 only. Individuals who dropped out were more
likely to be African American (p⬍.01) and less likely to have
completed college (p⬍.05). Wave 2 respondents did not differ
from nonrespondents on any other variables, including Wave 1
distress or benevolence beliefs.
Associations between prosocial behavior and distress. As
in Study 1, the study hypothesis was that prosocial behavior would
serve as a stress buffer for individuals with positive views of
others, and specifically that volunteering hours would buffer the
association between recent stressful events and psychological dis-
tress at high, but not low, levels of benevolence beliefs. The
prospective design of Study 2 allowed for the inclusion of stressful
events experienced between Waves 1 and 2 as well as Wave 1
distress as a control to reduce concerns about confounding due to
unmeasured individual differences. Because the final model, in-
cluding quadratic effects, is quite complex, and because the pro-
spective design should reduce confounding, the model reported in
this article omits other control variables (i.e., demographics, men-
tal health, personality, and social group participation) for clarity.
However, the results of a version of this model that screened these
control variables for entry using a procedure similar to that used in
Study 1 were substantively identical to those of the model reported
here.
As in Study 1, the study hypothesis implied a three-way inter-
action among volunteering, stressful events, and benevolence be-
liefs. To test this hypothesis, a multiple regression analysis for
Wave 2 distress included all three variables, all three two-way
product-term interactions among those variables, and the three-
way product-term interaction. Also as in Study 1, exploratory
models tested for quadratic terms of volunteering, stress, and
hostile cynicism. However, unlike in Study 1, a significant qua-
dratic effect emerged: specifically, a three-way Benevolence ⫻
Volunteering ⫻Stress interaction term containing a quadratic
effect of benevolence beliefs. Thus, the final model contained this
term and its constituent variables along with all hypothesized
terms.
The final model was examined for validity of regression as-
sumptions (e.g., normality and homoscedasticity of residuals) and
for multivariate outliers. Two multivariate outliers (i.e., dfit and
leverage values visually discontinuous from the distributions of
these values; Cohen et al., 2002) were detected and removed from
the analysis. Inspection of univariate data from these individuals
showed that they differed most greatly from others in reporting
more recent stressful events in the past year (19) than did any
others, suggesting that the resulting model may not apply well for
individuals who report extremely high numbers of stressful events.
In addition, visual inspection of the distribution of residuals indi-
cated heteroscedasticity in the form of increasing residual variance
with increasing predicted values of the model. Attempts to address
this issue by transforming the dependent variable using square
root, natural log, and log-10 transformations (Cohen et al., 2002)
did not change the fit of the model or the significance of any
coefficients, indicating that heteroscedasticity was not strongly
influencing the results. Given this pattern of results, data are
presented with respect to original units of the dependent variable
(distress) to maximize interpretability. Nonetheless, in Figure 2,
predicted values of distress are only presented for relatively low
values of stressful events, the key predictor, to minimize model
error.
6POULIN
F2
tapraid5/zg1-hea/zg1-hea/zg100313/zg12786d13z
xppws S⫽1 1/9/13 4:17 Art: 2012-1369
APA NLM

Results of the final model, reported in Table 2, indicated that
quadratic benevolence beliefs and recent stressful events signifi-
cantly interacted with volunteering (B⫽⫺0.03, p⫽.008). This
pattern indicated that the interaction between volunteering and
stress differed across levels of benevolence beliefs, but also that
the moderating role of benevolence beliefs differed across levels of
benevolence beliefs. To examine this effect, I probed the Volun-
teering ⫻Stress interaction with all variables standardized and
benevolence beliefs recentered (at zero) at three different points: at
1 standard deviation below the mean, at the mean itself, and at 1
standard deviation above the mean. These models indicated that
the Volunteering ⫻Stress interaction was not significant at low
levels of benevolence beliefs (B⫽⫺0.01, p⫽.37), nor at the
mean level of benevolence beliefs (B⫽⫺0.01, p⫽.44), but that
it was significant at high levels of benevolence beliefs (B⫽
⫺0.03, p⫽.004). That is, volunteering interacted with stress at
high, but not mean or low, levels of benevolence beliefs.
Examining the Volunteering ⫻Stress interaction at high levels
of benevolence beliefs showed that, as hypothesized, volunteering
buffered the stressful events– distress association at high levels of
benevolence beliefs. The association between recent stressful
events and Wave 2 distress, controlling for Wave 1 distress, was
less than half as strong at high levels of volunteering (B⫽0.04,
p⫽.001) than at low levels of volunteering (B⫽0.11, p⬍.001).
This pattern is illustrated in Figure 2.
Alternative models. As in Study 1, a set of alternative models
explored the robustness of the Study 2 findings. One additional
analysis used benevolence beliefs trichotomized into bottom, mid-
dle, and top thirds of scores and tested the Volunteering ⫻Stress
interaction in the three resulting groups. In the low-benevolence
group (n⫽402), the interaction was not significant (B⫽⫺0.02,
p⫽.11), and in the middle-benevolence group (n⫽495), the
same was true (B⫽⬍– 0.01, p⬎.99). In the high-benevolence
group (n⫽321), however, the Volunteering ⫻Stress interaction
was a significant predictor of distress (B⫽⫺0.03, p⫽.01).
Within the high-benevolence group, volunteering was also dichot-
omized and the association between stress and distress was exam-
ined in the two resulting groups. Among high-benevolence indi-
viduals who engaged in little or no volunteering in the past year
(50 hr or less, n⫽234), stressful events significantly predicted
Predicted Levels of Distress
1
1.1
1.2
1.3
1.4
1.5
1.6
1.7
1.8
1.9
04
Stressful Events
Low Benevolence Beliefs
1
1.1
1.2
1.3
1.4
1.5
1.6
1.7
1.8
1.9
04
Stressful Events
Average Benevolence Beliefs
1
1.1
1.2
1.3
1.4
1.5
1.6
1.7
1.8
1.9
04
Stressful Events
High Benevolence Beliefs
B = 0.07***
B = 0.06***
B = 0.07***
B = 0.06***
B = 0.11***
B = 0.04**
High Volunteering
Low Volunteering
Figure 2. Study 2: Predicted values of Brief Symptom Inventory distress by number of past-year stressful
events and amount of volunteering, graphed separately for low versus high (M⫾1SD) levels of world
benevolence beliefs.
Table 2
Study 2: Multiple Regression Models With Volunteering,
Stressful Life Events, and Benevolence Beliefs as Predictors of
Distress (N⫽1,150)
Variable
Wave 2 distress (R
2
⫽.51)
B95% CI
Wave 1 distress 0.55
ⴱⴱⴱ
[0.51, 0.59]
Stressful events between Waves
1 and 2 0.13
ⴱⴱⴱ
[0.10, 0.16]
Volunteer hours between Waves
1 and 2 ⫺0.03
ⴱ
[⫺0.06, 0.00]
Wave 1 benevolence beliefs ⫺0.01 [⫺0.03, 0.02]
Volunteering ⫻Stressful Events ⫺0.01 [⫺0.05, 0.02]
Benevolence ⫻Stressful Events 0.01 [⫺0.02, 0.04]
Benevolence ⫻Volunteering ⫺0.01 [⫺0.03, 0.02]
Benevolence ⫻Volunteering ⫻
Stressful Events ⫺0.03 [⫺0.06, 0.00]
Benevolence
2
0.01 [⫺0.01, 0.02]
Benevolence
2
⫻Stressful Events 0.02 [0.00, 0.04]
Benevolence
2
⫻Volunteering 0.00 [⫺0.02, 0.01]
Benevolence
2
⫻Volunteering ⫻
Stressful Events ⫺0.03
ⴱⴱ
[⫺0.05, ⫺0.01]
Note. All variables used in interactions are standardized. Interaction
terms represent product terms of standardized constituent variables. Model
fit was highly significant (p⬍.001).
ⴱ
p⬍.05.
ⴱⴱ
p⬍.01.
ⴱⴱⴱ
p⬍.001.
7
VOLUNTEERING AND VALUING OTHERS
T2,AQ:2
tapraid5/zg1-hea/zg1-hea/zg100313/zg12786d13z
xppws S⫽1 1/9/13 4:17 Art: 2012-1369
APA NLM

distress (B⫽0.07, p⬍.001), but among high-benevolence indi-
viduals who volunteered frequently in the past year (more than 50
hr, n⫽87), stressful events predicted distress less strongly (B⫽
0.02, p⫽.09).
Also as in Study 1, a series of models examined associations of
social engagement, in the form of group memberships, with dis-
tress. Social engagement did not predict distress on its own, in
interaction with stressful events, or in the context of three-way
interactions with volunteering and stress or with benevolence
beliefs and stress (ps⬎.05).
Discussion
The results of Study 2 also provided support for the study
hypothesis. Specifically, volunteering buffered the association be-
tween stressful events and psychological distress, but only among
individuals with positive views of others—that is, those high in
benevolence beliefs. In other words, at high, but not low or even
average, levels of benevolence beliefs, volunteering predicted a
diminished association between stressful events and distress.
Again, these findings are consistent with the notion that positive
views of others serve to facilitate stress-buffering prosocial moti-
vations in contexts beyond close relationships.
General Discussion
The present research presents the first evidence that the health
benefits of volunteering depend on having positive views of others,
building on prior research on moderators of the volunteering–
health association (Konrath et al., 2012;Oman, 2007). In Study 1,
volunteering versus not volunteering predicted reduced associa-
tions between life stress and mortality, but only among individuals
relatively low in hostile cynicism. In Study 2, engaging in volun-
teering behavior versus not doing so predicted reduced associa-
tions between stressful life events and psychological distress, but
only among individuals relatively high in world benevolence be-
liefs (including positive views of other people). Helping behavior,
stress, and positive views of others have all previously been linked
to health and well-being. The present studies clarify the conditions
under which each of these variables may do so.
Implications for Volunteering and Health
In the present data, volunteering served as a stress buffer only
for those with positive views of others. This finding, combined
with prior research on conditions under which volunteering has
benefits (Konrath et al., 2012;Oman, 2007), may qualify results
from previous studies indicating that helping or volunteering gen-
erally promotes health and well-being (e.g., Brown et al., 2003,
2008,2009;O’Reilly et al., 2008;Post, 2007;Poulin et al., in
press). Indeed, the results of Study 1 further suggest that these
benefits may accrue specifically to those in relatively poor health.
Taken together, all of these studies are highly consistent with a
model in which the caregiving behavioral system substantially
accounts for volunteering’s salutary role. The caregiving behav-
ioral system promotes efforts to help valued others (Brown &
Brown, 2006;Goetz et al., 2010), and the present research indi-
cates that volunteering is most beneficial for those who value
others the most. Those who value others may also be most likely
to volunteer for other-focused reasons, another moderator of vol-
unteering’s benefits consistent with the role of the caregiving
behavioral system (Konrath et al., 2012). Similarly, it is plausible
that those who value others are also likely to have high levels of
social contact, fitting with another known moderator of volunteer-
ing benefits (Oman, 2007). The present study could not test these
connections directly, but doing so could be a valuable task for
future research.
These findings, together with the findings of prior research, may
have implications for potential health-promotion efforts that seek
to draw on health benefits of volunteering. It appears that volun-
teering predicts health and well-being (e.g., Krause, 2006;Okun et
al., 2010;O’Reilly et al., 2008; see Post, 2007, for a recent
overview), but it may do so only for certain individuals or under
certain conditions. If so, it may be the case that volunteering
should be emphasized as a health behavior for some individuals
more than for others. Further research efforts should clarify the
conditions under which volunteering is most beneficial with an eye
toward making such recommendations.
Implications for Stress, Positive Views of Others, and
Health
In the findings from the present studies, the association between
stressful life events and adjustment were strongly qualified by both
volunteering and views of others. This was especially true for
mortality: Life stress predicted mortality only among cynical in-
dividuals or among low-cynical individuals who did little volun-
teering. The mental and physical implications of stress may be
very different for individuals who feel that they are contributing to
the well-being of others than for individuals who do not make such
contributions. Future research on the effects of stress and health
may need to routinely account for the contributions of helping
behavior alongside other stress moderators.
In interpreting the interactions among views of others, volun-
teering, and stress, it is arguably most straightforward to treat
views of others as a moderator of the stress-buffering effects of
volunteering. However, another way to view these interactions is
to note that the associations between views of others and health are
moderated by stress and by volunteering. That is, having positive
views of others appears to be most relevant for health among
individuals who have experienced recent stress and have engaged
in recent volunteering. This finding has implications for research
on the well-known link between hostile cynicism and health,
including cardiovascular disease (e.g., Pollitt et al., 2005;Ranjit et
al., 2007;Smith, 2006;Uchino, Vaughn, & Matwin, 2008). It is
possible that some of the effects of hostile cynicism can be traced
specifically to reduced stress-buffering effects of helping behavior
among cynical individuals.
Limitations and Future Directions
The findings of the present study represent the first examination
of the interactive role of views of others and helping behavior in
predicting health and well-being. However, they leave several
questions about the relationships among these variables unan-
swered. First, the mechanisms that may explain the associations
between helping and stress-buffering for those with positive views
of others remain unknown. It is plausible that, as noted, helping
8POULIN
tapraid5/zg1-hea/zg1-hea/zg100313/zg12786d13z
xppws S⫽1 1/9/13 4:17 Art: 2012-1369
APA NLM

leads to activation of the caregiving behavioral system, including
feelings of compassion and the release of stress-buffering hor-
mones, but neither study assessed these as potential mediators.
Future research on helping and well-being should do so.
The generalizability of the present findings may also be some-
what limited by the specific way in which volunteering was
assessed in both studies. In those studies, volunteering was mea-
sured as a continuous variable: the number of hours volunteered in
the past year. This approach is arguably more sensitive to potential
dose–response effects of volunteering than a dichotomous measure
asking participants whether they did or did not volunteer. Studies
using a variety of different measurement strategies have largely
converged on the finding that volunteering is beneficial, including
those using dichotomous or subjective frequency measures (e.g.,
Brown et al., 2003;Okun et al., 2010;Poulin et al., in press),
numbers of volunteering organizations (Oman et al., 1999),
amount of time spent volunteering (Brown et al., 2009;O’Reilly et
al., 2008;Poulin et al., 2010), or a combination (Konrath et al.,
2012). Nonetheless, one cannot rule out the possibility that the
specific pattern of findings reported in the present research might
differ depending on volunteering measure used.
An additional limitation of the present studies is that their
correlational design prevents inference about the causal direction
of the associations among views of others, helping, and stress-
buffering. The longitudinal design of both studies demonstrates
that views of others and helping together have predictive power
over stress-buffering with respect to mortality and psychological
adjustment, and controlling for psychological distress assessed
before stress and volunteering in Study 2 helps to make reverse
causation a less likely explanation for the observed results. How-
ever, future work should take an experimental approach to manip-
ulate views of others, helping, and stress to determine whether
trust influences stress-buffering effects of helping behavior. Re-
sults of future research, along with the findings of the present
studies, can further illuminate when and for whom helping is
healthy.
References
Barefoot, J. C., Dodge, K. A., Peterson, B. L., & Dahlstrom, W. (1989).
The Cook–Medley hostility scale: Item content and ability to predict
survival. Psychosomatic Medicine, 51, 46 –57.
Batson, C. D., Eklund, J. H., Chermok, V. L., Hoyt, J. L., & Ortiz, B. G.
(2007). An additional antecedent of empathic concern: Valuing the
welfare of the person in need. Journal of Personality and Social Psy-
chology, 93, 65–74. doi:10.1037/0022-3514.93.1.65
Batson, C. D., Fultz, J., & Schoenrade, P. A. (1987). Distress and empathy:
Two qualitatively distinct vicarious emotions with different motivational
consequences. Journal of Personality, 55, 19 –39. doi:10.1111/j.1467-
6494.1987.tb00426.x
Bradburn, N. (1969). The structure of psychological well-being. Chicago,
IL: Aldine.
Brown, S. L., & Brown, R. M. (2006). Target article: Selective investment
theory: Recasting the functional significance of close relationships.
Psychological Inquiry, 17, 1–29. doi:10.1207/s15327965pli1701_01
Brown, S. L., Brown, R. M., House, J. S., & Smith, D. M. (2008). Coping
with spousal loss: Potential buffering effects of self-reported helping
behavior. Personality and Social Psychology Bulletin, 34, 849 – 861.
doi:10.1177/0146167208314972
Brown, S. L., Nesse, R. M., Vinokur, A. D., & Smith, D. M. (2003).
Providing social support may be more beneficial than receiving it:
Results from a prospective study of mortality. Psychological Science,
14, 320 –327. doi:10.1111/1467-9280.14461
Brown, S. L., Smith, D. M., Schulz, R., Kabeto, M. U., Ubel, P. A., Poulin,
M.,...Langa, K. M. (2009). Caregiving behavior is associated with
decreased mortality risk. Psychological Science, 20, 488 – 494. doi:
10.1111/j.1467-9280.2009.02323.x
Carr, D., House, J. S., Kessler, R. C., Nesse, R. M., Sonnega, J., &
Wortman, C. (2000). Marital quality and psychological adjustment to
widowhood among older adults: A longitudinal analysis. The Journals of
Gerontology, Series B: Psychological Sciences and Social Sciences, 55,
S197–S207. doi:10.1093/geronb/55.4.S197
Cohen, J., Cohen, P., West, S. G., & Aiken, L. S. (2002). Applied multiple
regression/correlation analysis for the behavioral sciences (3rd ed.).
Mahwah, NJ: Erlbaum.
Costa, P. T., & McCrae, R. R. (1992). Normal personality assessment in
clinical practice: The NEO Personality Inventory. Psychological Assess-
ment, 4, 5–13. doi:10.1037/1040-3590.4.1.5
Cox, D. R. (1972). Regression models and life-tables. Journal of the Royal
Statistical Society Series B, 34, 187–220.
Derogatis, L. R., & Savitz, K. L. (2000). The SCL-90-R and Brief Symp-
tom Inventory (BSI) in primary care. In M. E. Maruish (Ed.), Handbook
of psychological assessment in primary care settings (pp. 297–334).
Mahwah, NJ: Erlbaum.
Floyd, K., Mikkelson, A. C., Tafoya, M. A., Farinelli, L., La Valley, A. G.,
Judd, J.,...Wilson, J. (2007). Human affection exchange: XIII.
Affectionate communication accelerates neuroendocrine stress recovery.
Health Communication, 22, 123–132. doi:10.1080/10410230701454015
Gillath, O., Shaver, P. R., Mikulincer, M., Nitzberg, R. E., Erez, A., & Van
IJzendoorn, M. H. (2005). Attachment, caregiving, and volunteering:
Placing volunteerism in an attachment-theoretical framework. Personal
Relationships, 12, 425– 446. doi:10.1111/j.1475-6811.2005.00124.x
Goetz, J. L., Keltner, D., & Simon-Thomas, E. (2010). Compassion: An
evolutionary analysis and empirical review. Psychological Bulletin, 136,
351–374. doi:10.1037/a0018807
Gosling, S., Rentfrow, P., & Swann, W. (2003). A very brief measure of
the Big-Five personality domains. Journal of Research in Personality,
37, 504 –528. doi:10.1016/S0092-6566(03)00046-1
Huber, P. J. (1967). The behavior of maximum likelihood estimates under
nonstandard conditions. Proceedings of the Fifth Berkeley Symposium
on Mathematical Statistics and Probability, 1, 221–233.
Janoff-Bulman, R. (1989). Assumptive worlds and the stress of traumatic
events: Applications of the schema construct. Social Cognition, 7, 113–
136. doi:10.1521/soco.1989.7.2.113
Konrath, S., Fuhrel-Forbis, A., Lou, A., & Brown, S. L. (2012). Motives
for volunteering are associated with mortality risk. Health Psychology,
31, 87–96. doi:10.1037/a0025226
Korn, E. L., Graubard, B. I., & Midthune, D. (1997). Time-to-event
analysis of longitudinal follow-up of a survey: Choice of the time-scale.
American Journal of Epidemiology, 145, 72– 80. doi:10.1093/
oxfordjournals.aje.a009034
Krause, N. (2006). Church-based social support and mortality. The Jour-
nals of Gerontology, Series B: Psychological Sciences and Social Sci-
ences, 61, S140 –S146. doi:10.1093/geronb/61.3.S140
Levenson, H. (1973). Multidimensional locus of control in psychiatric
patients. Journal of Consulting and Clinical Psychology, 41, 397– 404.
doi:10.1037/h0035357
Maner, J. K., & Gailliot, M. T. (2007). Altruism and egoism: Prosocial
motivations for helping depend on relationship context. European Jour-
nal of Social Psychology, 37, 347–358. doi:10.1002/ejsp.364
Midlarsky, E., & Kahana, E. (1994). Altruism in later life (Vol. 196).
Thousand Oaks, CA: Sage.
Okun, M. A., August, K. J., Rook, K. S., & Newsom, J. T. (2010). Does
volunteering moderate the relation between functional limitations and
9
VOLUNTEERING AND VALUING OTHERS
tapraid5/zg1-hea/zg1-hea/zg100313/zg12786d13z
xppws S⫽1 1/9/13 4:17 Art: 2012-1369
APA NLM

mortality? Social Science & Medicine, 71, 1662–1668. doi:10.1016/j
.socscimed.2010.07.034
Oman, D. (2007). Does volunteering foster physical health and longevity?
In S. G. Post (Ed.), Altruism and health (pp. 15–32). New York, NY:
Oxford University Press. doi:10.1093/acprof:oso/9780195182910.003
.0003
Oman, D., Thoresen, C. E., & McMahon, K. (1999). Volunteerism and
mortality among the community-dwelling elderly. Journal of Health
Psychology, 4, 301–316. doi:10.1177/135910539900400301
O’Reilly, D., Connolly, S., Rosato, M., & Patterson, C. (2008). Is caring
associated with an increased risk of mortality? A longitudinal study.
Social Science & Medicine, 67, 1282–1290. doi:10.1016/j.socscimed
.2008.06.025
Pollitt, R. A., Daniel, M., Kaufman, J. S., Lynch, J. W., Salonen, J. T., &
Kaplan, G. A. (2005). Mediation and modification of the association
between hopelessness, hostility, and progression of carotid atheroscle-
rosis. Journal of Behavioral Medicine, 28, 53– 64. doi:10.1007/s10865-
005-2563-y
Post, S. G. (Ed.). (2007). Altruism and health: Perspectives from empirical
research. New York, NY: Oxford University Press.
Poulin, M. J., Brown, S. L., Dillard, A., & Smith, D. M.). (in press). Stress
does not predict increased mortality among those who give to others.
American Journal of Public Health.
Poulin, M. J., Brown, S. L., Ubel, P. A., Smith, D. M., Jankovic, A., &
Langa, K. M. (2010). Does a helping hand mean a heavy heart? Helping
behavior and well-being among spouse caregivers. Psychology and
Aging, 25, 108 –117. doi:10.1037/a0018064
Radloff, L. S. (1977). The CES-D scale: A self-report depression scale for
research in the general population. Applied Psychological Measurement,
1, 385– 401. doi:10.1177/014662167700100306
Ranjit, N., Diez-Roux, A. V., Shea, S., Cushman, M., Seeman, T., Ni, H.,
& Jackson, S. A. (2007). Psychosocial factors and inflammation in the
Multi-Ethnic Study of Atherosclerosis. Archives of Internal Medicine,
167, 174 –181. doi:10.1001/archinte.167.2.174
Rogers, W. H. (1993). Regression standard errors in clustered samples.
Stata Technical Bulletin, 13, 19 –23.
Rosenberg, M. (1965). Society and the adolescent self-image. Princeton,
NJ: Princeton University Press.
Schlenker, B. R., & Britt, T. W. (2001). Strategically controlling informa-
tion to help friends: Effects of empathy and friendship strength on
beneficial impression management. Journal of Experimental Social Psy-
chology, 37, 357–372. doi:10.1006/jesp.2000.1454
Silver, R. C., Holman, E. A., McIntosh, D. N., Poulin, M. J., & Gil-Rivas,
V. (2002). Nationwide longitudinal study of psychological responses to
September 11. Journal of the American Medical Association, 288, 1235–
1244. doi:10.1001/jama.288.10.1235
Smith, T. W. (1994). Concepts and methods in the study of anger, hostility,
and health. In A. W. Siegman & T. W. Smith (Eds.), Anger, hostility,
and the heart (pp. 23– 42). Hillsdale, NJ: Erlbaum.
Smith, T. W. (2006). Personality as risk and resilience in physical health.
Current Directions in Psychological Science, 15, 227–231. doi:10.1111/
j.1467-8721.2006.00441.x
Stürmer, S., Snyder, M., & Omoto, A. M. (2005). Prosocial emotions and
helping: The moderating role of group membership. Journal of Person-
ality and Social Psychology, 88, 532–546. doi:10.1037/0022-3514.88.3
.532
Uchino, B. N., Vaughn, A. A., & Matwin, S. (2008). Social psychological
processes linking personality to physical health: A multilevel analysis
with emphasis on hostility and optimism. In F. Rhodewalt (Ed.), Per-
sonality and social behavior (pp. 251–283). New York, NY: Psychology
Press.
White, H. (1980). A heteroskedasticity-consistent covariance matrix esti-
mator and a direct test for heteroskedasticity. Econometrica, 48, 817–
830. doi:10.2307/1912934
Williams, R. L. (2000). A note on robust variance estimation for cluster-
correlated data. Biometrics, 56, 645– 646. doi:10.1111/j.0006-341X
.2000.00645.x
Received March 29, 2012
Revision received September 4, 2012
Accepted September 20, 2012 䡲
10 POULIN
tapraid5/zg1-hea/zg1-hea/zg100313/zg12786d13z
xppws S⫽1 1/9/13 4:17 Art: 2012-1369
APA NLM