Social Indicators Research (2005) 73: 431–458
Springer 2005

DOI 10.1007/s11205-005-0805-6

LINDSEY A. BAKER, LAWRENCE P. CAHALIN,

KERSTIN GERST and JEFFREY A. BURR

PRODUCTIVE ACTIVITIES AND SUBJECTIVE

WELL-BEING AMONG OLDER ADULTS:
THE INFLUENCE OF NUMBER OF ACTIVITIES
AND TIME COMMITMENTw

(Accepted 12 January 2005)

ABSTRACT. This study examines relationships among three measures of subjective

well-being (life satisfaction, happiness and depressive symptoms), and two global
measures of productive activity (number of activities and time commitment). We
argue that participation in multiple productive activities should increase subjective
well-being because these behaviors increase social integration and provide mean-
ingful social roles. Using the first two waves of the Americans’ Changing Lives
survey, we estimate a series of OLS and ordered logistic regression models to
examine this issue among a sample of respondents 60 years old and older. Our
multivariate regression results show that as time committed to productive activities
increases, life satisfaction increases. Both increasing numbers of productive activities
and increasing time commitment predict higher levels of happiness. Also, we find
only modest support for a relationship between productive activities and the number
of and changes in depressive symptoms. Our results provide support for the idea that
engaging in productive activities is beneficial to older persons’ well-being, implying
confirmation of the role enhancement hypothesis and demonstrating the importance
of social integration.

KEY WORDS: older persons, productive activities, subjective well-being

INTRODUCTION

The purpose of this study is to investigate relationships among pro-

ductive activities and subjective well-being in later life. Specifically,
we analyze the relationship between subjective well-being and the

w
This research was partially supported by a grant from the National Institute on
Aging (R03 AG018910). We thank Jan Mutchler for helpful comments on an earlier
version of this paper.
432 LINDSEY A. BAKER ET AL.

volume of productive activity in which persons engage, as charac-

terized by the number of activities and the time commitment to these
activities. We specifically concentrate on self-reports of life satisfac-
tion, a global concept reflecting a cognitive evaluation of one’s life
situation, self-reported happiness, and depressive symptoms.
Gerontologists, feminist scholars, economists, and others have
argued that productivity should be broadly defined to include unpaid,
non-market activities (e.g., volunteering, caregiving; see, Bass and
Caro, 2001; Herzog et al., 1989). Adopting this broad view, it is clear
that older persons contribute significantly to the well-being of other
individuals as well as to the larger society. Using data from the
Commonwealth Productive Aging Study, Bass and Caro (1995) show
that more than 25% of persons over age 55 work for pay, 25%
provide assistance to one or more disabled persons, and 40% help
care for children and grandchildren. Substantial evidence exists that
the contributions of older persons, both in terms of paid and unpaid
work, generate enormous benefits for the nation and for the indi-
viduals who are served. Estimates show that caregivers provided
nearly $200 billion in unpaid service in 1997 (Arno et al., 1999).
In part because of the recognition of these contributions by older
persons, productive activity in later life has emerged in the last decade
as an increasingly important area of gerontological scholarly en-
deavor (Morrow-Howell et al., 2001). Scholars now are not only
asking questions about who engages in productive activities, but also
whether and in what ways participation in a productive activity may
be related to positive and negative outcomes in later life (Jackson
et al., 1993). Our concern in this study is with the link between
productive activities and subjective well-being.
Questions which remain to be addressed include whether the
number of productive activities and the intensity of commitment to
these activities enhance the quality of life of older persons (Jackson
et al., 1993). Relatively few studies examine how participation in a
wide range of productive activities, combined with the time com-
mitted to these activities, may impact an array of indicators of well-
being. We believe it is instructive to acknowledge the fact that older
persons are often involved, to varying degrees of commitment, with
several different productive activities in their daily lives, and that the
degree of involvement and commitment is likely to be important for
conditioning subjective well-being. The examination of potential links
 PRODUCTIVE ACTIVITIES AND SUBJECTIVE WELL-BEING 433

among productive activities and well-being is necessary because as

people age they often lose important social roles, which may be re-
lated to the potential for declining well-being. Participation in mul-
tiple productive roles reflects the possibility that role substitution
(e.g., moving from paid work to volunteering) may ameliorate the
deleterious effects of role loss (see Mutchler et al., 2003).
We develop a rationale for why the number of productive activities
and the intensity of commitment to these activities should be linked
to subjective well-being drawing on the perspectives of role theory
and social integration theory. As Wethington et al. (2000) note ‘‘The
extensive knowledge base about social integration has clear implica-
tions for promoting productive social involvement, as well as health
and well-being, through the later years of life’’ (emphasis added, p.
48). They go on to encourage researchers to look beyond the work
role in terms of older persons’ well-being, including such roles as non-
paid work, organizational and religious involvements, and family
roles (e.g., caregiving). We take up their challenge, addressing our
specific goals with data from the first two waves of the Americans’
Changing Lives (ACL) panel survey.

Productive Activity and Subjective Well-Being

This study focuses on two key concepts: productive activity and
subjective well-being. Although observers have offered a number of
conceptual definitions of productive activity, one key point emanat-
ing from the literature is that no consensus exists on how to define
productive activity. However, a list of the components considered
relevant includes market-based economic activities (paid work), non-
market activities with economic value, such as formal social and civic
contributions (volunteering, informal helping behavior or social
assistance), self-improvement (education, training) and self-care
(Sherraden et al., 2001). We take the position that activities defined
as productive should benefit others (they should not be purely con-
sumptive), should have a social component, and should be mean-
ingful to individuals.
Based on these criteria, our definition of productive activity is
activity that produces goods or services, whether paid or not; we
specifically focus on activity that is embedded in social networks
(Morrow-Howell et al., 2001). We focus on five broad categories of
434 LINDSEY A. BAKER ET AL.

productive activity: (1) paid work, (2) formal volunteering, (3) care-
giving (for persons who have health problems), (4) informal helping
behavior (e.g., providing transportation, running errands for others),
and (5) do-it-yourself activities (e.g., yard work, home repair).
Housework is not included in our measure because it is typically
accomplished in isolation from other persons, whereas activities like
yard work and home repair are more likely to place persons outside
of their homes and in an environment where they may interact with
friends and neighbors.
Compared to the study of productive aging, the study of subjective
well-being in later life has a long history (e.g., Klumb, 2004; Lawton,
1975; Pinquart and Sorensen, 2000; Wheeler et al., 1998). Research
shows that important correlates of subjective well-being include
physical activity, socioeconomic status, social support, establishment
of meaningful roles, and adequate physical health, as well as per-
sonality characteristics such as personal control, self-esteem and self-
efficacy (e.g., Diener, 2000; Okun and Stock, 1984). Despite this
history, the conceptual definition and empirical measurement of
subjective well-being varies across disciplines and even within disci-
plines. Psychologists treat subjective well-being as a super-ordinate
construct within the context of cognitive theories of emotion that
subsume subordinate constructs such as happiness, life satisfaction,
and morale (Stock et al., 1986). As used more broadly by gerontol-
ogists, subjective well-being is a somewhat amorphous concept, pre-
senting researchers with a number of choices and perspectives,
including physical, psychological, clinical, social, and cultural
dimensions (Lawton, 1997).
A central concept in the study of subjective well-being is life sat-
isfaction, which may be defined as ‘‘a cognitive assessment of one’s
progress toward desired goals’’ (George, 1979, p. 210) or as ‘‘an
assessment of overall conditions derived from a comparison of one’s
aspirations and achievements’’ (Campbell et al., 1976: from Stock
et al., 1986, p. 92). Sociologists, psychologists, health researchers,
and others often include measures of depressive symptomology, self-
esteem, self-efficacy and sometimes mastery (Krause et al., 1992)
when conceptualizing subjective well-being. We employ an inclusive
view of subjective well-being in this study, incorporating life satis-
faction and happiness along with symptoms of psychological distress
(depressive symptoms).
 PRODUCTIVE ACTIVITIES AND SUBJECTIVE WELL-BEING 435

Literature Review: Research Linking Productive Activity and

Well-Being
Studies examining links between productive activity and well-being
usually focus on only one or perhaps a few of the broad array of
productive activities in which older persons may be involved. While
we believe this research represents an important contribution to the
field, it does not address the question of how a wide range of pro-
ductive activities in later life impacts well-being because it overlooks
the fact that many older persons engage in numerous productive
activities, sometimes dedicating a considerable amount of time to
these activities. We briefly review the current empirical research
focusing on specific productive activities and well-being.
Wilson (2000) concludes that volunteering activity improves both
physical and mental health (see also Morrow-Howell et al., 2003;
Wheeler et al., 1998; Young and Glasgow, 1998). Volunteering ap-
pears most beneficial for subjective well-being for those who are
moderately to highly active volunteers, especially among those who
report low levels of informal social interaction or who volunteer for
religious organizations. Recent research on volunteering indicates
that being involved in multiple types of volunteer activity is positively
associated with life satisfaction, as well as level of self-reported health
and mortality risk (Musick et al., 1999; Van Willigen, 2000). Musick
et al. (1999) report that time commitment to volunteering is related to
mortality in a curvilinear fashion, whereby those who do not vol-
unteer at all and those who volunteer at higher levels do not receive
the same survival benefits as those volunteering at a moderate level. It
is not clear whether this same relationship holds for other measures
of well-being.
Research findings on the links between labor force participation
and subjective well-being are somewhat equivocal in that some
analysts find that paid work in the later stages of the life course is
not empirically linked to psychological or physical well-being while
others find positive relationships. For example, one cross-sectional
study found that paid work is immaterial to either positive or
negative affect for both older men and older women (McIntosh and
Danigelis, 1995) while results from another cross-sectional study
from the same time period supported a positive relationship be-
tween work and life satisfaction (Aquino et al., 1996). Researchers
436 LINDSEY A. BAKER ET AL.

who have relied on cross-sectional data have come under criticism

because links between productive activity participation and well-
being are likely reciprocal and cannot be easily disentangled with
such data. Although relatively little research using longitudinal data
exists in examining paid work and life satisfaction, the longitudinal
analyses that have been conducted tend to find no clear differences
in well-being between retirement and non-retirement life stages (e.g.,
Palmore et al., 1985; Sterns et al., 1997).
The impact of caregiving for a disabled friend or relative on
subjective well-being has been researched widely, often indicating a
negative effect on the well-being of the caregiver – especially when the
caregiver provides intensive amounts of care (Rose-Rego et al., 1998;
Roth et al., 2001; Strawbridge et al., 1997). A major theme of this
literature is that female caregivers seem to experience more depres-
sion, lower life satisfaction, and overall lower well-being when com-
pared with male caregivers (Rose-Rego et al., 1998). Further,
Pavalko and Woodbury (2000) indicate that extensive caregiving
tends to have more negative effects on psychological well-being than
physical well-being.
While the majority of extant research focuses on a single indi-
cator of productive activity and well-being, there are examples of
research that consider more complex pictures of the productive
activities of older persons. Luoh and Herzog (2002), using three
waves of the Asset and Health Dynamics among the Oldest Old
(AHEAD) study, report finding that older persons who perform
100 or more hours of volunteering and paid work show improved
health and survivorship. When Krause et al. (1992) compared the
independent contributions of informal help and formal volunteering
on well-being, they found that the total effect of informal help was
an important contributor to well-being but that volunteering was
not.
Antonucci et al. (1991) examined volunteer work, helping others,
do-it-yourself activities, and housework to determine what effect
these productive activities had on life satisfaction. Their cross-sec-
tional analysis of ACL data indicated that men’s life satisfaction was
positively affected only by do-it-yourself activity and women’s was
positively affected by do-it-yourself activity and helping others.
Women showed less life satisfaction when hours of volunteering in-
creased. Glass et al. (1999) demonstrated that multiple productive
 PRODUCTIVE ACTIVITIES AND SUBJECTIVE WELL-BEING 437

activities are important predictors of mortality among a group of

healthy seniors employing prospective data from the New Haven
Established Populations for Epidemiological Studies of the Elderly
(EPESE). When comparing sets of productive activity measures to
sets of social and physical activity measures, the productive activity
set performed as well or better than the other two.
McIntosh and Danigelis (1995), using data from the first wave of
the ACL study, compared and contrasted gender and race differences
in subjective well-being for a number of indicators of productive
activity. They found that paid work had no effect on their measures
of subjective well-being for any of the race–gender groups they
analyzed, but that religious participation and formal volunteering in
non-religious organizations had positive effects for some gender–race
groups and that informal helping benefited the well-being of women
only. While each of these studies was path breaking in that they
considered several indicators of productive activities and well-being,
each had limitations including the use of cross-sectional data, lack of
evaluation of time commitment to productive activities and/or
truncated sets of control variables.

Role Theory, Social Integration, Productive Activity and

Well-Being
What might explain the generally positive influence of engagement in
productive activities for subjective well-being? We believe role theory
(Adelmann, 1994; Crosby, 1986) and social integration theory
(Durkheim, 1951; Rosow, 1967) are useful for conceptualizing how
engagement in productive activity (and the social roles inherently tied
to most of these activities) may influence subjective well-being (see
also Klumb, 2004; Van Willigen, 2000). Both approaches stipulate
that being embedded in supportive social networks, engaging in
meaningful social roles, and receiving social support improves the
quality of one’s life (e.g., Pillemer et al., 2000). Observers have long
argued that the more social positions or roles that a person occupies,
the better their mental and physical health (see Moen et al., 1992;
Verbrugge, 1983). The roles that a person occupies affect her/his self-
perception and behavior in part by providing a wider and denser
social network, increasing the possibility of receiving social support
in times of need.
438 LINDSEY A. BAKER ET AL.

A substantial body of research supports the hypothesis that being

embedded in social networks positively impacts well-being (e.g.,
Berkman and Syme, 1979; Glass et al., 1999; Silverstein and Parker,
2002; Thoits, 1983). We argue that productive activity is a behavioral
expression of the commitment to specific social roles (more than
simply a status) that places the individual within a variety of eco-
nomic and non-economic networks (Jackson et al., 1993). Thus,
participating as a volunteer, caregiver or neighborhood helper
represents a productively active role (also consistent with activity
theory). Further, productive activities are different from social
activities, which typically do not have the goal of helping others.
Productive activity is also different from physical activities (e.g.,
exercise) because physical activities are often accomplished in social
isolation and do not have as an objective the provision of assistance
to others.
Engagement in multiple productive activities leaves open the
possibility that well-being may be impaired because a person finds
him/herself stretched too thin with too many demands. This yields
two competing possibilities in terms of the impact of the number of
and time commitment to productive activities and subjective well-
being. One possibility is that as engagement in productive activities
increases, a role enhancement effect will occur, yielding higher levels
of subjective well-being. The other possibility is that as the number of
and commitment to productive activity roles increases beyond an
unspecified point, role strain or role conflict will occur whereby
subjective well-being will be negatively impacted. Further, as we
noted above, research is emerging that suggests moderate involve-
ment in specific productive activities in later life has more positive
benefits than involvement at low or high levels. Our goal is to address
the research question: How does the number of productive activities
in which an older person engages and the time commitment to these
activities impact subjective well-being?
Based on the weight of the empirical evidence and the perspectives
of role theory and social integration theory, we hypothesize that:
1. As the number of productive activities increases, subjective well-
being will increase.
2. As time commitment to productive activities increases, subjective
well-being will increase.
 PRODUCTIVE ACTIVITIES AND SUBJECTIVE WELL-BEING 439

3. A curvilinear relationship exists between productive activity and

subjective well-being, such that the benefits of participation will
accrue to persons who participate at moderate levels.

RESEARCH DESIGN

We use panel data from the ACL survey to investigate the relation-
ship between productive activities and subjective well-being. The
ACL survey is a nationally representative sample of 3617 persons age
25 and older living in the U.S. first interviewed in 1986 (Wave 1) with
2867 respondents re-interviewed in 1989 (Wave 2). The ACL over-
samples Blacks and persons age 60 and over. We combine both waves
of data for this analysis, selecting for analysis respondents who are
age 60 or older at Wave 1 and who were interviewed in both waves
(N ¼ 1279). These data are particularly useful for this study for a
number of reasons, including (1) the breadth and depth of informa-
tion collected on productive activities and subjective well-being, (2)
the collection of data at two time points, allowing us to accomplish a
longitudinal analysis and helping us to disentangle the temporal
relationship between productive activities and measures of subjective
well-being, and (3) the data are nationally representative.
Sample weights are provided with the data file. Centered weights
are estimated for the study sample and applied to adjust for the over-
sampling characteristics of the original survey. Also, because the ACL
data were collected using a complex sample design, it is necessary to
move beyond a ‘‘naı̈ve’’ weighting scheme to one that adjusts the data
for the sampling error introduced at each level of sample selection.
Although a naı̈ve weighting strategy and a strategy that incorporates
the clustering effects of the sample design yield similar regression
coefficients, the standard errors of the coefficients are often different.
The standard errors of the naı̈ve weighting approach tend to be biased
downward, sometimes yielding inappropriate estimates of statistical
significance of effects. We employ the SVYREG and SVYOLOG
procedures in STATA to estimate our regression models. We handle
missing values due to item non-response by assigning the mean,
median or modal response, as appropriate.
Because sample attrition occurs between the first and second
waves of the panel, we compared survey respondents who were
440 LINDSEY A. BAKER ET AL.

interviewed at both time points (the study sample) with those

respondents who were interviewed only at time one along a number
of dimensions (see Appendix A). This analysis shows that respon-
dents in our study sample were more likely to be female, younger,
more highly educated, and report fewer health conditions than those
who were not interviewed in both waves. In addition, study sample
respondents participated in more productive activities, devoted more
hours to productive activity, were more physically and socially active,
and reported a larger social support network than respondents who
were not re-interviewed. Finally, those who remained in the panel
reported being more satisfied with life and having fewer depressive
symptoms than those who left the study. Based on these differences,
readers should use appropriate caution with respect to the general-
izability of the results.
Dependent variables. We examine three measures of subjective
well-being from the second wave of the ACL. The first two include
global measures of life satisfaction and happiness. Life satisfaction is
measured on a seven-point scale (1=completely dissatisfied to
7=completely satisfied). Happiness includes three response catego-
ries (not too happy, pretty happy and very happy). The response sets
for both indicators are ordered. In preliminary analyses, we estimated
ordered logistic regression models in an attempt to more effectively
capture the rank-order information in these variables. However, for
life satisfaction, the proportional odds test indicated that it was
inappropriate to employ ordered logistic regression with this variable
(see Long, 1997). Therefore, we treat the life satisfaction measure as if
it were a continuous variable, estimating life satisfaction models using
OLS regression techniques. The proportional odds test for the hap-
piness variable indicated that using ordered logistic regression tech-
niques was appropriate. We also include a count of the number of
depressive symptoms, using a modified measure of the CESD scale
(CESD; Radloff, 1977; Wallace and O’Hara, 1992). This measure is
created by summing the eleven indicators of depressive symptoms
found in the ACL (range 0–11; alpha = 0.78): OLS regression
techniques are employed for this dependent variable.
Independent variables. Two indexes of productive activity are
generated; one represents the total number of productive activities in
which respondents are involved and the other estimates the amount
of time commitment respondents give to all productive activities
 PRODUCTIVE ACTIVITIES AND SUBJECTIVE WELL-BEING 441

combined. The first composite index measuring number of productive

activities includes regular and irregular paid work, five types of for-
mal volunteering, acute and chronic caregiving for a disabled friend,
relative, or neighbor, four types of informal help provided to friends,
relatives or neighbors and four types of do-it-yourself (domestic)
activities (range 0–16: alpha ¼ 0.72). Approximately 97% of
respondents in our study sample reported participating in housework
(defined as work inside the home as compared to domestic activities
engaged in outside the home). Thus adding housework to our mea-
sure of productive activity would not greatly affect the variability of
our productive activity index.
The second composite index measures respondents’ time com-
mitment to productive activities. The ACL does not provide time
commitment for each of the specific activities listed above, but rather
it groups the more specific activities into five broad categories: paid
work, volunteering, caregiving, informal help, and do-it-yourself
activity. Hours of paid work are reported in continuous hours
annually, whereas for the other productive activity variables, time
commitment is reported in six broad categories. We group the hours
reported for each activity into four categories: no time committed=0,
low commitment=1, medium commitment =2, and high commit-
ment =3 (see Appendix B for details on this grouping strategy). This
strategy allows us to have a common metric of time commitment
across the productive activities. We create a composite index (range
0–20) by summing respondents’ level of commitment on each of the
five dimensions of productive activity (alpha ¼ 0.48).
We also include measures of other types of activity likely to
influence an older person’s subjective well-being. First, a composite
measure is created to capture level of involvement in physical activ-
ities: gardening, taking walks, and exercising/playing sports
(0=never, 1=rarely, 2=sometimes, 3=often). Level of involvement
across types of physical activity is summed to form a measure ranging
from 0 to 9. Social activity is measured in a similar way by summing
across three activities: attending meetings, talking on the phone, and
visiting friends (0=never, 1=less than once a month, 2=about once
a month, 3=two or three times a month, 4=once a week, 5=more
than once a week: range = 0–15).
Additional control variables include respondent’s age (years), sex
(1=female), race (1=Black), marital status (1=married), number of
442 LINDSEY A. BAKER ET AL.

persons who could provide social support in times of need (0–40),

education (years), self-reported number of health conditions (0–7),
and religious participation (religious service attendance, read religious
books, watch or listen to religious programming: range=1–18). More
detail on the definitions and coding of the variables is presented in
Appendix B.
We make use of both waves of the ACL panel data, estimating
models where the predictor variables are measured at Wave 1 and
subjective well-being is measured at Wave 2. This strategy increases
the likelihood that any relationships we find among subjective well-
being and productive activity in later life are not confounded by
simultaneous measurement of the variables in the model (this strategy
does not prove causality).
For life satisfaction and depressive symptoms, we take a longitu-
dinal multivariate regression approach that allows us to predict levels
of subjective well-being at Wave 2, as well as changes in subjective
well-being between Wave 1 and Wave 2 as a function of the number
of productive activities and the time commitment to these activities.
For the analysis of levels of subjective well-being, we include the set
of control variables identified above. For the change analysis, we
include a baseline (Wave 1) measure of either life satisfaction or
depressive symptoms along with the set of control variables. Because
the ACL survey does not ask a question about happiness at Wave 1,
we are only able to estimate models of level of happiness at Wave 2.
Finally, because some research shows that caregiving is negatively
related to subjective well-being whereas our other indicators are
positively related to subjective well-being, we estimated our multi-
variate regression models with modified productive activity variables.
The modification involved dropping caregiving from the indexes. The
results of these regressions are consistent with the results based on the
full index (results available from authors upon request). Therefore,
we report the results from the full productive activity indexes.

RESULTS

Descriptive statistics and zero-order correlations are shown in

Table I. The mean global life satisfaction score for this sample of
respondents age 60 and above is 5.6 (s.d.=1.5) and the mean
 TABLE I
Zero-order correlations and descriptive statistics

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15

1 PA# 1.0
2 PA h 0.835* 1.0
3 Age )0.314* )0.305* 1.0
4 Female )0.216* )0.162* 0.070* 1.0
5 Black )0.132* )0.120* 0.014 0.023 1.0
6 Married 0.302* 0.233* )0.238* )0.271* )0.113* 1.0
7 Education 0.367* 0.368* )0.183* )0.005 )0.227* 0.122* 1.0
8 Health )0.171* )0.182* 0.114* 0.151* 0.113* )0.104* )0.200* 1.0
9 PhyAct 0.482* 0.419* )0.195* )0.195* )0.130 0.159* 0.255* )0.207* 1.0
10 SocAct 0.331* 0.319* 0.015 0.184* )0.047 0.011 0.245* )0.081* 0.198* 1.0
11 Religious Part 0.073* 0.055* 0.107* 0.150* 0.182* )0.027 )0.127* 0.130* 0.053 0.250* 1.0
12 Support 0.195* 0.179* )0.054 )0.062* )0.075* 0.104* 0.018 )0.023 0.111* 0.095* 0.141* 1.0
13 Life Satisfaction2 0.088* 0.117* 0.002 )0.005 0.003 0.072* )0.045 )0.098* 0.113* 0.105* 0.110* 0.146* 1.0
14 CESD2 )0.238* )0.230* 0.094* 0.140* 0.091* )0.113* )0.229* 0.250* )0.247* )0.103* 0.047 )0.111* )0.364* 1.0
15 Happiness2 0.188* 0.187* 0.008 )0.081 )0.045 0.142* 0.076 )0.138* 0.187* 0.152* 0.082* 0.176* 0.525* )0.307* 1.0
Mean 4.7 5.0 69.6 0.60 0.09 0.62 10.9 1.9 5.0 8.9 7.8 8.7 5.6 3.4 1.2
StdDev 2.9 2.9 6.9 0.49 0.28 0.49 3.4 1.4 2.6 3.3 4.3 8.5 1.5 2.7 0.63

Source: 1986 and 1989 Americans’ Changing Lives Survey. See text and Appendix B for variable definitions.
Notes: N (unweighted) = 1279. Statistics based on weighted data. Variables 1–11 from Wave 1 and variables 13–15 from Wave 2. * p £ 0.05.
1 PA = Productive activities index 1 (number of activities); 2 PA h = Productive activities index 2 (annual hours); 3 Age=Respondent’s Age in years; 4
Female (versus male); 5 Black (versus non-black); 6 Married (versus non-married); 7 Education (years); 8 Health (number of health conditions); 9 PhyAct
PRODUCTIVE ACTIVITIES AND SUBJECTIVE WELL-BEING

(physical activities index); 10 SocAct (social activities index); 11 ReligiousPart (religious participation index); 12 Support (social support index); 13 Life
Satisfaction2 (global question); 14 CESD2 (depressive symptoms index); 15 Happiness2 (global question).
443
444 LINDSEY A. BAKER ET AL.

happiness score is 1.2 (s.d.=0.6). The mean depressive symptom index

score is 3.4 (s.d.=2.7). The mean number of productive activities re-
ported is 4.7 (s.d.=2.9). The mean age of the sample is 69.6
(s.d.=6.9); 60% of the sample is female and 9% of the sample is
African American. Sixty-two percent of respondents are married and
the mean number of years of education is 10.9 (s.d.=3.4). The mean
number of health conditions is 1.9 (s.d.=1.4) and the score for the
physical and social activities measures are 5 and 8.9, respectively. The
mean number of persons in respondents’ social support network is 8.7
(s.d.=8.5) and the mean religious participation score is 7.8 (s.d.=4.3).
Bivariate relationships among the variables are given in the form
of correlations. The number of productive activities reported and the
commitment of time to these activities are positively correlated with
both life satisfaction and happiness and negatively correlated with
number of depressive symptoms. Respondents reporting more phys-
ical and social activities, as well as those who are more engaged in
religious activities and who report larger social support networks,
report higher levels of life satisfaction and happiness and fewer
depressive symptoms. Further, being married is positively associated
with both of these measures of subjective well-being while poor health
is negatively correlated. These findings are consistent with findings
reported in the existing research literature.
Table II describes the mean number of productive activities
measured at Wave 1 for specific categories of the dependent variables
measured at Wave 2 – life satisfaction, happiness, and depressive
symptoms. Panel A (Life Satisfaction) demonstrates that the mean
number of productive activities for older persons reporting being
dissatisfied on the life satisfaction measure is 3.1 (N ¼ 42) while the
mean number of activities for those who report being satisfied is 4.9
(N ¼ 773). Panel B (Happiness) shows that those older persons who
report they are ‘‘not too happy’’ participate in an average of 3.1
productive activities (N ¼ 147). Those who report being ‘‘very hap-
py’’ participate in an average of 5.2 productive activities (N ¼ 444).
Panel C (Depressive Symptoms) shows that the majority of individ-
uals (N ¼ 807) report no depressive symptoms; these respondents
also report participating in an average of 5.2 productive activities.
Persons reporting five or more depressive symptoms (N ¼ 38) report
participating in an average of 2.7 activities, slightly more than half as
many as those reporting no symptoms.
 PRODUCTIVE ACTIVITIES AND SUBJECTIVE WELL-BEING 445
TABLE II

Mean productive activity levels for categories of life satisfaction, happiness, and
depressive symptoms

Panel A
Life satisfaction Wave 2 (range 1–7)
Productive Dissatisfied Somewhat Neutral Somewhat Satisfied
activities dissatisfied satisfied
Mean # 3.1 3.7 4.3 5.0 4.9
(0–15)
N 42 54 243 168 773

Panel B
Happiness Wave 2 (range 0–2)
Productive Not too Pretty Very
activities happy happy happy
Mean # 3.1 4.7 5.2
(0-15)
N 147 688 444

Panel C
Depressive symptoms Wave 2 (range 0–11)
Productive 0 1 2 3 4 5+
activities
Mean # 5.2 4.0 3.6 4.0 3.9 2.7
(0–15)
N 807 255 119 37 23 38

Thus a general pattern exists showing that as the mean number of

productive activities increases, so too does subjective well-being.
Also, for life satisfaction and depressive symptoms, the advantage of
participating in productive activities is highest among those reporting
a moderate number of productive activities, suggesting the possibility
of a curvilinear effect. At this more simple level of analysis, the data
indicate that low and high levels of participation in productive
activity may yield lower levels of subjective well-being; while more
moderate activity levels may be optimum for enhancing subjective
well-being. Note that for the happiness measure, the relationship with
number of productive activities appears to be linear.
The results of the regression analyses of life satisfaction, happi-
ness, and depressive symptoms are presented in Tables III through V.
446 LINDSEY A. BAKER ET AL.

For life satisfaction and number of depressive symptoms, we estimate

six models (two bivariate and four multivariate regression models),
which include separately our two measures of productive activity
(total number and time commitment). One set of models estimates
levels of subjective well-being measured at Wave 2 as a function of
number of activities and time commitment, as well as the full slate of
independent variables measured at Wave 1. Another set of models
estimates changes in subjective well-being as a function of number of
activities and time commitment, as well as the control variables.
These models include a baseline (Wave 1) estimate of the dependent
variable. In these models the coefficient for a given predictor variable
may be interpreted as follows: a one unit change in the independent
variable yields a given amount of change (value of beta coefficient) in
the dependent variable between Waves 1 and 2, net of the set of
controls.
Note also that in preliminary analyses we found a curvilinear
relationship between life satisfaction and depressive symptoms
dependent variables and our two measures of productive activity (see
also Table II). However, after introducing our set of covariates into
the models, the analyses showed only a linear relationship between
subjective well-being and productive activities. Consequently, the
models reported here include only linear terms for the two productive
activity measures. Thus, only at a simple level of analysis is our third
hypothesis supported.
We estimate models of life satisfaction using OLS regression
techniques; these results are reported in Table III. The cumulative
index capturing the number of productive activities is strongly related
to level of life satisfaction in the bivariate model, but after controlling
for the relevant covariates, number of productive activities is not
related to the level of life satisfaction in a statistically significant
manner. This result does not support our first hypothesis. Further,
size of social support network and participation in physical activities
are positively related to the level of life satisfaction, while number of
health conditions is negatively related to the level of life satisfaction.
The impact of the number of productive activities is not statistically
significant in the model that predicts change in life satisfaction
between Waves 1 and 2. The results show that life satisfaction is quite
stable over time, following from the model that includes an indicator
of life satisfaction at Wave 1. Further, having a larger network of
 TABLE III
OLS regression results for Life Satisfaction (Wave 2) regressed on productive activities indexes and selected variables (Wave 1)

Variables b (se) (1) b (se) (2) b (se) (3) b (se) (4) b (se) (5) b (se) (6)
PA index 1 (#) 0.044* (0.019) 0.008 (0.024) 0.006 (0.020)
PA index 2 (h) 0.059* (0.018) 0.039* (0.021) 0.038* (0.019)
Physical activities 0.048* (0.022) 0.028 (0.021) 0.041* (0.021) 0.020 (0.020)
Social Activities 0.030* (0.015) 0.016 (0.015) 0.024  (0.015) 0.010 (0.015)
Social support 0.020** (0.005) 0.013* (0.005) 0.019** (0.005) 0.012* (0.005)
Religious 0.022* (0.013) 0.007 (0.011) 0.021* (0.013) 0.006 (0.011)
participation
Health conditions )0.110* (0.045) )0.087* (0.044) )0.108* (0.045) )0.085* (0.044)
Education )0.045* (0.017) )0.036* (0.017) )0.050* (0.018) )0.041** (0.018)
Age 0.006 (0.007) )0.001 (0.006) 0.008 (0.007) 0.001 (0.006)
Female 0.088 (0.110) 0.041 (0.101) 0.108 (0.108) 0.063 (0.100)
Black 0.051 (0.144) 0.096 (0.145) 0.056 (0.144) 0.101 (0.145)
Married 0.183* (0.125) 0.032 (0.120) 0.172  (0.122) 0.020 (0.117)
Life satisfaction 0.556** (0.060) 0.556** (0.060)
Intercept 5.406 4.858 3.531 5.323 4.678 3.349
2
R 0.008 0.061 0.166 0.014 0.064 0.169
Adj. R2 0.007 0.053 0.158 20.013 0.056 0.161
Notes: Unstandardized coefficients (b) and standard errors (se).
PRODUCTIVE ACTIVITIES AND SUBJECTIVE WELL-BEING

* p £ 0.05; ** p £ 0.001;  0.10 ‡ p ‡ 0.05, one-tailed test of statistical significance.

447
448 LINDSEY A. BAKER ET AL.

social support to call on in times of need predicts an increase in the

level of life satisfaction over time, while poorer health and having a
greater amount of education predicts a decrease in life satisfaction
over time.
Our results indicate that the amount of time contributed to pro-
ductive activities is positively related to the level of life satisfaction
and change in life satisfaction from Wave 1 to Wave 2. This is evident
in both the bivariate and multivariate models. Thus while the number
of activities does not appear to predict the level of or change in
subjective well-being as measured by life satisfaction, the amount of
time commitment is statistically related to both level and change in
life satisfaction. These results provide some support for our second
hypothesis.
The results of our ordered logistic regression analysis of produc-
tive activity and the level of happiness at Wave 2 are reported in
Table IV. The results from the first model indicate that as the number
of productive activities in which an older person engages increases,
there is an increase in the log-odds of happiness; however, the rela-
tionship is only statistically significant in the bivariate model: the
effect is eliminated when controls are added. As participation in
physical and social activities increases and the size of respondents’
social support network increases, the level of reported happiness in-
creases. Higher levels of happiness are reported for those who are
married versus those who are not married. Finally, as the number of
health conditions reported increases, the level of happiness decreases.
The results of the models including time commitment to productive
activities show that as the hours of commitment to productive
activities increases, the level of happiness also increases. In the mul-
tivariate analyses of happiness, our first hypothesis is not supported
but some support for the second hypothesis is attained.
The OLS regression results for the analysis of depressive
symptoms and productive activity are presented in Table V. For
the analysis of the number of depressive symptoms, both the
number of productive activities engaged in and the time commitment
to those activities are negatively related to the number of depres-
sive symptoms reported by older persons, net of our set of con-
trols. However, the effect is only statistically significant in the
bivariate models. For the change analysis, number of and com-
mitment to productive activities is related to depressive symptoms
 PRODUCTIVE ACTIVITIES AND SUBJECTIVE WELL-BEING 449

TABLE IV
Ordered logistic regression results for Happiness Scale (Wave 2) regressed on
productive activities indexes and selected variables (Wave 1)

Variables b (se) (1) b (se) (2) b (se) (3) b (se) (4)

PA index 1 (#) 0.122** 0.029 (0.031)
(0.026)
PA index 2 (h) 0.122** 0.049* (0.028)
(0.025)
Physical 0.082* (0.025) 0.079* (0.023)
activities
Social activities 0.056* (0.024) 0.052* (0.025)
Social support 0.032** (0.007) 0.032** (0.007)
Religious 0.021 (0.022) 0.021 (0.022)
participation
Health )0.149* (0.075) )0.146* (0.076)
conditions
Education )0.004 (0.020) )0.008 (0.020)
Age 0.019* (0.010) 0.021* (0.010)
Female )0.127 (0.140) )0.119 (0.140)
Black 0.0003 (0.150) 0.0006 (0.149)
Married 0.402* (0.124) 0.406* (0.117)
Threshold 1 )1.505* 0.494 )1.472* 0.636
Threshold 2 1.230* 3.372* 1.265* 3.520*
Model X2 24.4** 91.7** 24.3** 94.3**
DF 2 12 2 12
Notes: Log-likelihood coefficients (b) and standard errors (se) * p £ 0.05;
** p £ 0.001;  0.10 ‡ p ‡ 0.05, one-tailed test of statistical significance.
Happiness scale (0 = not too happy, 1 = pretty happy, 2 = very happy).

in a negative direction: however, the relationship is statistically

significant in the bivariate case and is only marginally related to
the number of depressive symptoms reported. In terms of the
relationship between productive activities and depressive symptoms,
our results do not provide strong support for either of our first
two hypotheses.
Our results also show that participation in physical activity, size of
social support network, and education level are negatively related to
the number of depressive symptoms reported. The results also indi-
cate that as the number of health conditions reported increases, so
too does the number of depressive symptoms reported. Finally, older
women report a higher number of depressive symptoms than men, net
 TABLE V
450
OLS regression results for number of Depressive Symptoms (Wave 2) regressed on productive activities indexes and selected variables
(Wave 1)

Variables b (se) (1) b (se) (2) b (se) (3) b (se) (4) b (se) (5) b (se) (6)
PA index 1 (#) )0.222** (0.034) )0.061 (0.050) )0.040 (0.041)
PA index 2 (h) )0.212** (0.035) )0.061  (0.045) )0.040 (0.036)
Physical activities )0.128* (0.049) )0.067* (0.043) )0.132* (0.044) )0.069* (0.037)
Social activities )0.019 (0.025) 0.0001 (0.021) )0.020 (0.025) )0.0003 (0.022)
Social support )0.024* (0.010) 0.0007 (0.009) )0.024* (0.010) 0.0007 (0.008)
Religious 0.016 (0.020) 0.016 (0.017) 0.016 (0.019) 0.016 (0.016)
participation
Health conditions 0.339** (0.053) 0.091* (0.039) 0.335** (0.053) 0.088* (0.038)
Education )0.102* (0.031) )0.072* (0.026) )0.101* (0.030) )0.071** (0.026)
Age )0.003 (0.014) 0.002 (0.013) )0.003 (0.015) 0.002 (0.013)
Female 0.382* (0.190) 0.280* (0.187) 0.394* (0.180) 0.289* (0.178)
Black 0.031 (0.215) 0.036 (0.196) 0.035 (0.214) 0.039 (0.197)
Married )0.084 (0.199) 0.280  (0.187) )0.104 (0.196) 0.103 (0.176)
LINDSEY A. BAKER ET AL.

Depressive 0.492** (0.037) 0.491** (0.037)

symptoms
Wave 1 (log)
Intercept 4.426 5.077 2.370 4.441 5.151 2.414
2
R 0.057 0.140 0.345 0.053 0.141 0.345
Adj. R2 0.056 0.133 0.338 0.052 0.133 0.339
 
Notes: Unstandardized coefficients (b) and standard errors (se). * p £ 0.05; ** p £ 0.001; 0.10 ‡ p ‡ 0.05, one-tailed test of statistical
significance.
 PRODUCTIVE ACTIVITIES AND SUBJECTIVE WELL-BEING 451

of the influence of productive activity and the control variables in the

model. In the following section we summarize our results and discuss
them in light of current research and theory.

DISCUSSION

Our primary goal for this study was to examine the impact of a wide
range of productive activities on several measures of subjective well-
being for a sample of older persons in the U.S. While the amount of
attention to the study of productive activity in later life has acceler-
ated recently and some researchers have begun to establish empirical
links between productive activity and well-being, we contend it is
appropriate to begin to examine how the full constellation of activ-
ities and the time committed to these activities impacts subjective
well-being. This is pertinent because older persons typically engage in
more than one activity during their later years. To our knowledge,
this paper is among the first to address these issues.
The results of our analyses provide partial support for two of our
three hypotheses. Our first hypothesis, that participation in a greater
number of productive activities would be positively related to
subjective well-being, was confirmed only at the bivariate level. After
controlling for other domains of activity, social support, health and
demographic characteristics, the impact of number of productive
activities on subject well-being was reduced to statistical insignifi-
cance (although the effects were in the expected direction). We find
support for the second hypothesis in that the greater the time
commitment to productive activities, the greater the level of and
improvement (change) in life satisfaction over time. Also, higher
levels of time commitment to productive activities are associated
with higher levels of happiness. Finally, we find a moderately sta-
tistically significant relationship between hours committed to pro-
ductive activity and level of depressive symptoms. While our third
hypothesis, that there would be a curvilinear effect of productive
activities on subjective well-being, received some support when we
cross-tabulated number of productive activities with life satisfaction
and depressive symptoms, a more rigorous evaluation of this
hypothesis within a multivariate regression framework did not
support this expectation.
452 LINDSEY A. BAKER ET AL.

In general, we find that engagement in and commitment to pro-

ductive activities later in life is somewhat beneficial to subjective well-
being within the context of our research design. These findings are
consistent with the existing research that suggests multiple roles, as
reflected by participation productive activities, leads to a better
quality of life, possibly through increases in social integration and
improvements in self-esteem and self-efficacy (see Adelmann, 1994;
Jackson et al., 1993; Morrow-Howell et al., 2003). The results also
seem to support the role enhancement perspective and do not provide
support for the role strain or role conflict perspectives.
Moreover, the results of our change analyses suggest that life
satisfaction and self-report of depressive symptoms are fairly stable
conditions. Such stability may represent actual consistency in these
social psychological domains, or it may represent an underreporting
of negative conditions – a hesitation by individuals to admit a sense
of failure or problems with mental health. Further, we find that our
measures of subjective well-being are related to physical and social
activities and number of health conditions, findings that have been
observed in other research (Glass et al., 1999). Our finding that in-
creased social support increases subjective well-being is also consis-
tent with social integration theory (Durkheim, 1951; Pillemer and
Glasgow 2000; Rosow, 1967). Social integration theory suggests that
social networks generally improve quality of life, including mental
and physical health (Moen et al., 1992; Verbrugge, 1983).
Our definition of productive activity specifically focused on activity
that is embedded in social networks which follows our central thesis
that productive activity that is social has a positive effect on well-being.
This definition means that we did not include housework in our mea-
sure because it is typically accomplished in isolation as a non-social
activity. Also, as we noted above about 97% of the sample reported
participating in some level of housework. Had we been able to include
housework in our analysis, and if housework does have positive effects
on well-being, then we might expect even stronger relationships among
our measures of productive activity and well-being. In future research,
analysts may want to pursue this issue more thoroughly.
A better understanding of the central issues in this study may also
be enhanced in future research efforts by attention to some additional
issues. First, a wider range of indicators of subjective well-being
should be considered (e.g., morale, self-esteem, self-efficacy). After
 PRODUCTIVE ACTIVITIES AND SUBJECTIVE WELL-BEING 453

this is accomplished, more confidence may be generated for the

conclusions drawn from this analysis. Second, the indicators of life
satisfaction and happiness in the ACL are less than optimal because
they are based on global questions in the survey. It would be helpful
to have multiple indicators of life satisfaction and happiness con-
structs so that the complex nuances of the conceptualization of these
measures could be attained. In addition, researchers will want to
extend this analysis by examining a range of specific domains of life
satisfaction (e.g., health, marriage, family, neighborhood; see
Michalos et al., 2001). Third, when more recent data sources become
available that include high quality measures of the range and com-
mitment to productive activities as well as measures of subjective
well-being with stronger psychometric qualities, it will be useful to
replicate the analyses in this study to compare results. Finally, the
influence of engagement in multiple productive activities and the
intensity of that engagement for well-being should be extended to
analyses of physical health and mortality.

APPENDIX A
Comparison of characteristics of respondents in study sample (both waves) with
Wave 2 non-respondents (attritors)

Mean Percent
Wave 1 Both Wave 1 Both
only waves only waves
Married 59% 63%
Black 11% 9%
Female* 52% 61%
Age* 73 69
Education* 10 11
Productive activity index 1* 3.10 4.78
Productive activity index 2* 3.31 5.06
Social activity index* 7.68 9.04
Physical activity index* 4.08 5.06
Religious participation index 6.87 6.83
Social support* 7.39 8.81
Health conditions* 2.22 1.86
Completely or very satisfied w/life* 71% 73%
CES-D* 7.15 6.50
*p < 0.05; continuous variable comparisons evaluated with t-tests, categorical
variable comparisons evaluated with v2
454 LINDSEY A. BAKER ET AL.

APPENDIX B
Coding of dependent and independent variables

Variable Coding/question wording/index

components
Dependent variables (Wave 2)
Life satisfaction ‘‘How satisfied are you with your life as a
whole these days?’’ 1 = completely dis-
satisfied; 4 = neutral; 7 = completely
satisfied
Happiness ‘‘Taking all things together how would
you say things are these days…?’’
0 = not too happy; 1 = pretty happy;
2 = very happy
Depressive symptoms 11-item CES-D score (range: 0–11)
Independent variables (Wave 1)
Productive activity index 1 Total # of productive activities partici-
(number of activities) pated in (range: 0–16)
Paid Work: regular/irregular employ-
ment Volunteering: religious/political/
educational/senior group/other
Caregiving: acute or chronic care for
friend/relative/neighbor
Informal Helping: errands/housework/
childcare/other
Do-it-yourself: home improvement/can-
ning/yard work/car repair
Productive activity index 2 Level of commitment in hours (grouped)
(amount of time commitment) to productive activities (range: 0–20)
Hours of paid work over last 12 months
(‘‘none’’ 0 = no hours; ‘‘low’’ 1 = 1–
1000 h; ‘‘medium’’ 2 = 1001–1999 h;
‘‘high’’ 3 = 2000 or more hours)
Caregiving, volunteering, informal help-
ing, and do-it-yourself hours over last
12 months (‘‘none’’ 0 = no hours;
‘‘low’’ 1 = 10 or 30 h; ‘‘medium’’
2 = 60, 80 or 120 h; ‘‘high’’ 3 = 200
or more hours)
Social activity Level of participation in social activities
(range: 0–15) Talk on Phone/Visit with
Friends/Attend Meetings
 PRODUCTIVE ACTIVITIES AND SUBJECTIVE WELL-BEING 455

APPENDIX B
Continued

Variable Coding/question wording/index

components

Physical activity Level of participation in physical activ-

ities (range: 0–9)
Work in Garden/Active Sports/Take
Walks
Social support Size of social support network (range:
0–40)
‘‘About how many friends or other
relatives do you have whom you could
call on for advice or help if you needed
it?’’
Control variables
Health conditions Total # of health conditions self-re-
ported (range: 0–7)
arthritis/rheumatism/lung disease/hyper-
tension/heart attack/heart trouble/ dia-
betes/cancer/problems with circulation,
corns, calluses/stroke/broken bones/ur-
inary incontinence
Age Age in years
Education Highest year of education
Religious participation Sum of three indicators: religious service
attendance, read religious books, and
watch or listen to religious programming
(1 never. . .6 more than once a week)
(range: 0–15)
Married 1 = married; 0 = other
Black 1 = black; 0 = other
Female 1 = female; 0 = male

REFERENCES

Adelmann, P.: 1994, ‘Multiple roles and psychological well being in a national sample
of older adults’, Journal of Gerontology 49, pp. S277–S285.
Antonucci, T.C., J.S. Jackson, R.C. Gibson and A.R. Herzog: 1991, ‘Sex differences
in age and racial influences on involvement in productive activities’, in M. Ste-
venson, M. Kite and B. Whitley (eds.), Gender Roles in Later Adulthood (Ball
State University Press, Muncie, IN).
456 LINDSEY A. BAKER ET AL.

Aquino, J.A., D.W. Russell, C.E. Cutrona and E.M. Altmaier: 1996, ‘Employment
status, social support and life satisfaction among the elderly’, Journal of Coun-
seling Psychology, 43(4), pp. 480–489.
Arno, P.S., C. Levine and M.M. Memmott: 1999, ‘Economic value of informal
caregiving’, Health Affairs 18(2), pp. 182–188.
Bass, S.A. and F.G. Caro: 1995, ‘Dimensions of productive engagement’, in S. Bass
(ed.), Older and Active: How Americans over 55 Are Contributing to Society,
(Yale University Press, New Haven), pp. 71–96.
Bass, S.A. and F.G. Caro: 2001, ‘Productive aging: A conceptual framework’, in N.
Morrow-Howell, J. Hinterlong and M. Sherraden (eds.), Productive Aging:
Concepts and Challenges (The Johns Hopkins University Press, Baltimore), pp.
37–78.
Berkman, L.F. and S.L. Syme: 1979, ‘Social networks, host resistance and mortality:
A nine-year follow-up study of Alameda County residents’, American Journal of
Epidemiology 100, pp. 186–204.
Campbell, A., P.E. Converse and W.L. Rodgers: 1976, The Quality of American
Life: Perceptions, Evaluations, and Satisfactions (Russell Sage Foundation, New
York).
Crosby, F.J. (ed.): 1986, Spouse, Parent, Worker: On Gender and Multiple roles,
(Yale University Press, New Haven).
Diener, E.: 2000, ‘Subjective well-being: The science of happiness and a proposal for
a national index’, American Psychologist 55(1), pp. 34–43.
Durkheim, E.: 1951, ‘Suicide: A Study in Sociology; translated by John A. Spaulding
and George Simpson (Free Press, Glencoe, Illinois).
George, L.K.: 1979, ‘The happiness syndrome: Methodological and substantive is-
sues in the study of social-psychological well-being in adulthood’, Gerontologist,
19(2), pp. 210–216.
Glass, T.C., R. Mendes De Leon, R.A. Marottoli and L. Berkman: 1999, ‘Popula-
tion based study of social and productive activities as predictors of survival among
elderly Americans’, British Medical Journal 319, pp. 478–483.
Herzog, A.R., R.L. Kahn and J.N. Morgan: 1989, ‘Age differences in productive
activities’, Journals of Gerontology 44(4), pp. S129–S138.
Jackson, J.S., T.C. Antonucci and R. Gibson: 1993, ‘Cultural and economic contexts
of aging productively over the life course: An economic network framework’, in S.
Bass, F. Caro and Y. Chen (eds.), Achieving a Productive Aging Society (Auburn
House, CT Westport), pp. 247–268.
Klumb, P.: 2004, ‘Benefits from productive and consumptive activities: Results from
the Berlin Aging Study’, Social Indicators Research 67, pp. 107–127.
Krause, N., A. R. Herzog and E. Baker: 1992, ‘Providing support to others and well-
being in later life’, Journal of Gerontology: Psychological Sciences 47, pp. 300–311.
Lawton, M.P.: 1975, ‘The Philadelphia Geriatric Center Morale Scale: A revision’,
Journal of Gerontology 30, pp. 80–95.
Lawton, M.P.: 1997. ‘Measures of quality of life and subjective well-being’, Gener-
ations 21, pp. 45–47.
Long, J. Scott: 1997, Regression Models for Categorical and Limited Dependent
variables (Sage Publications, Beverly Hills, CA).
Luoh, M.C. and A.R. Herzog: 2002, ‘Health and mortality as individual conse-
quences of volunteer and paid work in old age’, Journal of Health and Social
Behavior 43, pp. 490–509.
 PRODUCTIVE ACTIVITIES AND SUBJECTIVE WELL-BEING 457

McIntosh, B. and N. Danigelis: 1995, ‘Race, gender, and the relevance of pro-
ductive activity for elder’s affect’, Journal of Gerontology 50, p. S229.
Michalos, A.C., A.M. Hubley, B.D. Zumbo and D. Hemingway: 2001, ‘Health and
other aspects of the quality of life of older people’, Social Indicators Research 54,
pp. 239–274.
Moen, P., D. Dempster-McClain and R.M. Williams, Jr.: 1992, ‘Successful aging: A
life course perspective on women’s multiple roles and health’, American Journal of
Sociology 97, pp. 1612–1638.
Morrow-Howell, N., J. Hinterlong, P.A. Rozario and F. Tang: 2003, ‘Effects of
volunteering on the well-being of older adults’, Journal of Gerontology: Social
Sciences 58B, pp. S137–S145.
Morrow-Howell, N., J. Hinterlong and M.W. Sherraden: 2001, ‘Productive Aging:
Concepts and Challenges (Johns Hopkins University Press, Baltimore, MD).
Musick, M.A., A.R. Herzog and J.S. House: 1999, ‘Volunteering and mortality
among older adults: Findings from a national survey’, Journal of Gerontology
54B(3), pp. S173–S180.
Mutchler, J., J. Burr and F. Caro: 2003, ‘From paid worker to volunteer: Leaving the
paid workforce and volunteering in later life’, Social Forces 81(4), pp. 1267–1293.
Okun, M.A. and W.A. Stock: 1984, ‘Quality of research on correlates of subjective
well-being in adulthood’, Experimental Aging Research 10(3), pp. 161–163.
Palmore, E.B., B.M. Burchett, G.G. Fillenbaum, L.K. George and L.M. Wallman:
1985, Retirement: Causes and Consequences (Springer Publishing Company, New
York).
Pavalko, E.K. and S. Woodbury: 2000, ‘Social roles as process: Caregiving careers
and women’s health’, Journal of Health and Social Behavior 41, pp. 91–105.
Pillemer, K. and N. Glasgow: 2000, ‘Social integration and aging: Background and
trends’, in K. Pillemer, P. Moen, E. Wethington and N. Glasgow (eds.), Social
Integration in the Second Half of Life (Johns Hopkins University Press, Balti-
more, MD).
Pillemer, K.A., E. Moen, E. Wethington and N. Glasgow (eds.): 2000, Social Inte-
gration in the Second Half of Life (Johns Hopkins University Press, Baltimore,
MD).
Pinquart, M. and S. Sorensen: 2000, ‘Influences of socioeconomic status, social
network, and competence on subjective well-being in later-life: A meta-analysis’,
Psychology and Aging 15(2), pp. 187–224.
Radloff, L.S.: 1977, ‘The CES-D scale: A self-report depression scale for research in
the general population’, Applied Psychological Measurement 1, pp. 385–401.
Rose-Rego, S.K., M.E. Strauss and K.A. Smyth: 1998, ‘Differences in the perceived
well-being of wives and husbands caring for persons with Alzheimer’s disease’, The
Gerontologist 38, pp. 224–230.
Rosow, I.: 1967, Social Integration of the Aged (Free Press, New York).
Roth, D.L., W.E. Haley, J.E. Owen, O.J. Clay and K.T. Goode: 2001, ‘Latent
growth models of the longitudinal effects of dementia caregiving: A comparison of
African American and White family caregivers’, Psychology and Aging 16, pp.
427–436.
Sherraden, M., N. Morrow-Howell, J. Hinterlong and P. Rozario: 2001, ‘Produc-
tive aging: Theoretical choices and directions’, in N. Morrow-Howell, J. Hinter-
long and M. Sherraden (eds.), Productive Aging: Concepts and Challenges (Johns
Hopkins University Press, Baltimore, MD), pp. 260–284.
458 LINDSEY A. BAKER ET AL.

Silverstein, M. and M.G. Parker: 2002, ‘Leisure activities and quality of life among
the oldest old in Sweden’, Research on Aging 24, pp. 528–547.
Sterns, H.L., N.K. Matheson and L.S. Park: 1997, ‘Work and retirement’, in K.F.
Ferraro (ed.), Gerontology: Perspectives and Issues, 2nd editon (Springer, New
York), pp. 171–192.
Stock, W.A., M.A. Okun and M. Benin: 1986, ‘Structure of subjective well-being
among the elderly’, Psychology and Aging 1(2), 91–102.
Strawbridge, W.J., M.I. Wallhagen, S.J. Shema and G.A. Kaplan: 1997, ‘New
burdens or more of the same? Comparing grandparent, spouse, and adult-child
caregivers’, The Gerontologist 37, pp. 505–510.
Thoits, P.A.: 1983, ‘Multiple identities and psychological well-being: A reformula-
tion and test of the social isolation hypothesis’, American Sociological Review 48,
pp. 174–187.
Van Willigen, M.: 2000, ‘Differential benefits of volunteering across the life course’,
Journals of Gerontology 55B(5), S308–S318.
Verbrugge, L.M.: 1983, ‘Women and men: Mortality and health of older people’, in
M.W. Riley, B.B. Hess and K. Bond (eds.), Aging in Society: Selected Reviews of
Recent Research (Lawrence Erlbaum Associates, Hillsdale, NJ), pp. 139–174.
Wallace, J. and M. W. O’Hara: 1992, ‘Increases in depressive symptomatology in
rural elderly: Results from a cross-sectional and longitudinal study’, Journal of
Abnormal Psychology 101, pp. 398–404.
Wethington, E., P. Moen, N. Glasgow and K. Pillemer: 2000, ‘Multiple roles, social
integration and health’, in K. Pillemer, P. Moen, E. Wethington and N. Glasgow
(eds.), Social Integration in the Second Half of Life (Johns Hopkins Press, Bal-
timore, MD).
Wheeler, J.A., K.M. Gorey and B. Greenblatt: 1998, ‘The beneficial effects of vol-
unteering for older volunteers and the people they serve: A meta-analysis’,
International Journal on Aging and Human Development 47(1), pp. 69–79.
J. Wilson: 2000, ‘Volunteering’, Annual Review of Sociology 26, pp. 215–240.
Young, F.W. and N. Glasgow: 1998, ‘Voluntary social participation and health’,
Research on Aging 20(3), pp. 339–362.

Gerontology Institute Lindsey A. Baker

University of Massachusetts
100 Morrissey Blvd.
Boston, MA 02125-3393
USA
E-mail: lindseybaker@excite.com

Northeastern University Lawrence P. Cahalin

USA

University of Massachusetts Kerstin Gerst

Boston, MA Jeffery A. Burr
USA