Multiple Roles and Well-being: Sociodemographic
and Psychological Moderators
Christina J. Chrouser Ahrens & Carol D. Ryff
Published online: 1 December 2006
# Springer Science + Business Media, Inc. 2006
Abstract Research on multiple roles has supported the
enhancement hypothesis, but it is unclear if benefits of
multiple role involvement exist across all segments of the
population. This study was designed to examine whether
the role enhancement hypothesis suits both men and
women with varied education levels. A further goal was
to determine if perceived control moderates associations
between multiple role involvement and well-being. This
sample included 2,634 individuals from the Midlife in the
United States (MIDUS) survey who occupied up to eight
roles each. Psychological well-being was measured in six
dimensions (autonomy, environmental mastery, personal
growth, positive relations with others, purpose in life, and
self-acceptance); positive and negative affect were also
measured. Results of hierarchical regression analyses
supported the role enhancement hypothesis, as greater role
involvement was associated with greater well-being; how-
ever, the findings suggest that it was only well educated
women with multiple roles who showed higher levels of
autonomy. Perceived control was also found to moderate
some of the obtained linkages.
Contemporary women and men are frequently invested in
multiple roles. A recent national survey shows that 64% of
women and 74% of men aged 25–74 are married, and only
17% do not have children (Marks, Bumpass, & Jun, 2004).
In addition, 70% of women and 81.9% of men between the
ages of 20 and 64 are in the labor force (Clark &
Weismantle, 2003), and over one-half of women with a
newborn child were employed in 1995 (U.S. Bureau of the
Census, 1998). The role of elder caregiver is also becoming
more common, as women and men in the United States can
currently expect to spend more years of their life with one
or both parents over the age of 65 than with children under
the age of 18 (Watkins, Menken, & Bongaarts, 1987).
The present study was based on a national sample of
American women and men in order to examine the
association between multiple roles and six dimensions of
psychological well-being. Educational standing was a key
variable in the study to clarify how position in the social
structure might be central to understanding how multiple
roles and well-being are linked. In this investigation we also
examined how this association was moderated by perceived
control, which has been a focus of inquiry as a moderator of
social class influences on health and well-being.
Multiple Role Theories
Two theories have been proposed to explain the effects of
multiple roles on well-being. The role strain perspective, or
scarcity hypothesis, proposes that increased numbers of roles
lead to overload and strain, which can translate into negative
effects on physical and psychological well-being (Goode,
1960; Marks, 1977). Many researchers have failed to find
extensive support for this theory (Adelmann, 1994a, b; Hong
& Seltzer, 1995; Jackson, 1997; Menaghan, 1989; Miller,
Moen, & Dempster-McClain, 1991; Moen, Dempster-
McClain, & Williams, 1989, 1992; Pietromonaco, Manis,
& Frohardt-Lane, 1986; Thoits, 1983, 1986).
Sieber (1974) recognized that role accumulation may
lead to rewards that outweigh any possible negative effects.
Sex Roles (2006) 55:801–815
C. J. C. Ahrens (*):C. D. Ryff
University of Wisconsin,
2245 Medical Sciences Center, 1300 University Avenue,
Madison, WI 53706, USA
He theorized that an individual with a wide variety of role
partners may be able to compensate for relationship failures
by falling back on other relationships. Qualitative analyses
have indicated that individuals perceive their role identities
as sources of existential meaning, purpose, and behavioral
guidance (Simon, 1997), and recent reviews indicate that
women and men who fulfill multiple roles, as opposed to
few roles, report lower levels of mental and physical health
problems and greater levels of subjective well-being
(Barnett & Hyde, 2001).
According to the role enhancement perspective, or what
Thoits (1983) referred to as the identity accumulation
hypothesis, increased numbers of roles enhance an individ-
ual’s resources, social connections, power, prestige, and
emotional gratification (Marks, 1977; Sieber, 1974; Thoits,
1983). This theory has garnered considerable empirical
support (Adelmann, 1994a, b; Barnett, Marshall, & Singer,
1992; Hong & Seltzer, 1995; Jackson, 1997; Kikuzawa,
2006; Miller et al., 1991; Moen et al., 1989, 1992;
Pietromonaco et al., 1986; Stephens, Franks, & Townsend,
1994; Thoits, 1983, 1986).
For example, in an early review of multiple roles research,
Thoits (1986) noted that researchers have typically restricted
their analyses to an examination of spouse, parent, and
employee roles and ignored other role involvements such as
friend, church member, and organization member. Thoits
(1983) refined the role enhancement hypothesis by focusing
on a larger number of roles (eight) and found that increased
numbers of roles were associated with lower levels of
psychological distress. A subsequent study provided evi-
dence that increased numbers of roles were associated with
lower anxiety and depression for both women and men
(Thoits, 1986). The number of role identities held did not
account for gender differences in distress.
Jackson (1997) made refinements to the role enhance-
ment hypothesis by providing evidence that the benefits of
role involvement vary for different ethnic groups. Greater
role occupancy was linked with lower depression and
greater happiness among non-Hispanic White and Mexican
American men and women, but evidence suggested no such
association among African Americans and Puerto Ricans
(Jackson, 1997). On the contrary, Cochran, Brown, and
McGregor (1999) reported that having more roles was
linked with lower depression among both White and
African American women. Further refinements were also
made by Kikuzawa (2006), who explored cross-cultural
differences in role occupancy and found that having more
roles was associated with lower depression for American
and Japanese women and men. However, the benefit of
having each additional role was greater in the US than in
Japan (Kikuzawa, 2006).
Adelmann (1994a) sought evidence that the role en-
hancement hypothesis was applicable to older adults. Links
were found between greater role occupancy and better
subjective health, fewer health disabilities, and fewer
chronic conditions for both women and men in later life.
Further analyses revealed that increased roles were also
associated with higher well-being, which was measured in
terms of life satisfaction, depression, and self-efficacy
Many researchers have limited their samples to women,
and greater role occupancy has been linked with greater
self-esteem and job satisfaction for employed women
(Pietromonaco et al., 1986). Moen et al. (1989) examined
the effect of six roles over 30 years and found that women
who occupied more roles tended to live longer than women
who occupied fewer roles; subsequent analyses also
revealed links with better health outcomes 30 years later
(Moen et al., 1992). Based on the same sample, Miller et al.
(1991) found that greater role involvement was related to
higher life satisfaction and self-esteem for married women.
A later study by Hong and Seltzer (1995) illustrated that
greater role occupancy was predictive of lower future
depression after the researchers controlled for initial levels
of depression for caregiving women who occupied up to
eight roles. That study included analyses based on both the
number of roles occupied and different role combinations,
and the evidence suggests that the number of roles held,
rather than the specific roles occupied, seems to have a
stronger influence on psychological well-being.
Scarcity Versus Enhancement for Whom?
There is some evidence to suggest that men may benefit
from involvement in multiple roles as much as women do,
although the effect of particular roles may vary by gender.
There is little evidence as to whether such findings are
applicable to individuals of varied educational status, or if
the association between multiple roles and psychological
well-being may also be influenced by other intervening
processes, such as sense of control.
Gender as a moderator Some studies have shown the
association between greater role occupancy and higher
psychological well-being and better health to be stronger
for men than for women (Adelmann, 1994a, b), whereas
others have shown similar associations for women and men
(Thoits, 1983, 1986). Note that all of these studies
demonstrated that both men and women benefit from
involvement in multiple roles.
Studies focused on the roles of spouse, parent, and
employee have revealed that men and women may
experience different outcomes as a result of combining
these roles. Simon (1997) found that employed women
reported greater exposure to strain from combining work
802Sex Roles (2006) 55:801–815
and parenting roles than employed men did; however,
mothers were not more affected than fathers from this
strain. Janzen and Muhajarine (2003) reported that Canadi-
an women who combined the roles of spouse, parent, and
employee had better self-rated health and a lower likelihood
of having a chronic illness than women with fewer roles,
while for men, it was only older men with three roles who
reported better self-rated health. The results of a cross-
cultural study included that women in Finland, men in
Estonia, and women and men in the United Kingdom
reported the highest levels of life satisfaction when
combining all three roles, while men in Finland and women
in Estonia did not show this pattern (Schoon, Hansson, &
Salmela-Aro, 2005). A study of caregiving found that men
who combined the role of caregiver with other family roles
experienced a greater level of burden than women who
combined these roles (Jutras & Veillux, 1991).
Occupancy of the parental role coupled with the absence
of the spousal role may be particularly difficult for women.
Menaghan (1989) found that men reported more psycho-
logical symptoms when they were unmarried or were
simultaneously married, unemployed, parents. Women
suffered more symptoms when they were unmarried
mothers; employment had more varied effects (Menaghan,
1989). Sachs-Ericsson and Ciarlo (2000) found that having
more roles was associated with lower rates of psychiatric
disorders for men and women, and that single mothers were
particularly vulnerable to psychiatric disorders. Combining
the role of single parenthood with employment has also
been linked with lower life satisfaction among women in
Finland, Estonia, and the UK (Schoon et al., 2005). More
researchers need to consider how a wider array of roles may
be differentially linked with well-being outcomes for
women and men.
Education as a moderator According to Thoits (1982),
individuals with more, rather than fewer, years of education
are less distressed by undesirable events. Other researchers
have reported that educational attainment is negatively
associated with distress (Kessler, 1982; Link, Lennon, &
Dohrenwend, 1993; Ross & Mirowsky, 1989; Ross & Van
Willigen, 1997). It is interesting that some findings suggest
that, although greater education is linked with lower
distress, distress is reduced more for women than men
(Kessler, 1982; Ross & Van Willigen, 1997).
Little research considers how links between role
involvement and well-being may vary based on educational
attainment. According to Thoits (1987), lower class
individuals who have multiple roles may be less likely to
be able to negotiate satisfying relationships with role
partners, and thus, they may experience more role strain
and psychological distress than more highly educated
individuals with the same roles. This suggests that
education may work as a moderator of associations between
role involvement and well-being. In addition, education
may moderate distress differently for women and men.
(Kessler, 1982; Ross & Van Willigen, 1997).
Although previous studies of links between role involve-
ment and well-being have included information about
educational attainment, it has typically been included as a
control variable (Adelmann, 1994a, b; Hong & Seltzer,
1995; Miller et al., 1991; Moen et al., 1989, 1992;
Pietromonaco et al., 1986; Thoits, 1983, 1986). The present
study includes education as a substantive variable of
interest, as individuals with more education may have
enhanced well-being because they have greater access to
resources, employment, and supportive environments (Ross
& Van Willigen, 1997), which allow them to handle
multiple roles more effectively.
Perceived control as a moderator Sense of control may be
a relevant moderator, given that it has been linked with
psychological distress (Ali & Avison, 1997; Mirowsky &
Ross, 1986, 1990; Ross & Mirowsky, 1989; Ross & Van
Willigen, 1997). In a meta-analysis of 45 studies that
focused on cognitive representations of illness, there were
significant links between reporting more control over one’s
illness and having better psychological well-being and
lower psychological distress (Hagger & Orbell, 2003).
An individual’s perceived sense of control may also be
an important link between educational attainment and well-
being. Individuals with more education have higher levels
of personal control than individuals with lower educational
levels (Bird & Ross, 1993; Ross & Mirowsky, 1992).
However, other research suggests that links between
education, income, and depression are largely explained
by the pattern that individuals with a higher sense of control
have fewer depressive symptoms (Ross & Mirowsky,
1989). Later work also showed that individuals with higher
levels of education have higher perceived control, which is
associated with lower levels of emotional and physical
distress (Ross & Van Willigen, 1997).
be a moderator of links between education and well-being. In
analyses of the Midlife in the United States (MIDUS) sample,
Lachman and Weaver (1998) found that having high
perceived control was beneficial for all socioeconomic groups
in terms of links with greater life satisfaction, better perceived
health, and lower depression. However, control was found to
moderate the association such that lower income groups were
afforded a greater benefit by having a high sense of control.
Individuals in the lowest income group who had a high sense
of control showed levels of well-being comparable with the
higher income groups. For higher income groups, well-being
was generally high and did not vary much as a function of the
level of control (Lachman & Weaver, 1998).
Sex Roles (2006) 55:801–815803
Pertinent to the experience of multiple roles, Christensen,
Stephens, and Townsend (1998) found that mastery, or
perceived competence and control, in the roles of mother,
wife, and caregiver was related to higher psychological well-
being, as measured by life satisfaction and depression. They
also determined that mastery for each role seemed to have a
cumulative effect, such that higher levels of mastery across
different roles was associated with higher life satisfaction
(Christensen et al., 1998). Mastery in work and family roles
has also been investigated as a mediator of associations
between stress and rewards in each role and an individual’s
rating of the importance of the role. As work stress
increased, the rated importance of employment decreased
through a weakened sense of mastery in the employee role,
while as rewards increased, the importance of employment
increased through a heightened sense of mastery (Norton,
Gupta, Stephens, Martire, & Townsend, 2005).
These findings suggest that perceived control may act as a
moderator of links between multiple role involvement and
well-being. Furthermore, the results of these studies suggest
that perceived control may be particularly important among
those who lack high socioeconomic standing, although this
possibility has not been assessed in the context of multiple
roles, where prior research has also shown a sense of mastery
to be a significant factor.
Positive Outcomes in the Multiple Role Experience
Well-being, broadly defined, includes hedonic well-being
(evaluations of happiness and life-satisfaction) and eudai-
monic well-being (reaching human potential) (Ryan & Deci,
2001). Early role theorists suggested links between role
involvement and eudaimonic well-being. “Role requirements
give purpose, meaning, direction, and guidance to one’s life.
The greater the number of identities held, the stronger one’s
senseofmeaningful,guidedexistence”(Thoits,1983, p. 175).
Empirical investigations of role involvement, however, have
focused on relations between role involvement and negative
outcomes, such as depression, distress (Adelmann, 1994b;
Cochran et al., 1999; Hong & Seltzer, 1995; Kikuzawa,
2006; Martire, Stephens, & Townsend, 2000; Thoits, 1983,
1986), and poor health (Adelmann, 1994a), although some
researchers have considered the association between multiple
roles and life satisfaction and self-efficacy (Adelmann,
1994b; Martire et al., 2000; Miller et al., 1991; Pietromonaco
et al., 1986). Few investigators have considered how
multiple role involvement may stimulate human develop-
ment by enhancing personal growth, goal attainment, life
purpose, and connections with other people.
There is some evidence to suggest that multiple role
involvement is linked with eudaimonic aspects of well-
being. Although Burton (1998) only included an assess-
ment of three social roles (employee, spouse, and parent),
occupancy of all three roles was linked with the greatest
increase in integrative meaning, which was defined as a
feeling that one’s life had purpose and meaning. Integrative
meaning was also identified as a mediator of links between
role involvement and psychological distress (Burton, 1998).
The present investigation focused on evaluating links
between the cumulative occupancy of up to eight roles
(spouse, parent, employee, churchgoer, friend, volunteer,
social organization member, and caregiver) and six compo-
nents of eudaimonic well-being. The intent was to clarify
the extent to which multiple roles are associated with the
belief that one is growing, developing, and realizing
personal potential (personal growth), has goals to live for
that make life meaningful (purpose in life), feels positive
self-regard about one’s self and past life (self-acceptance),
has the sense that one can create a surrounding environment
suited to personal needs and capacities (environmental
mastery), sees oneself as capable of following one’s own
convictions (autonomy), and has trusting, loving relation-
ships with others (positive relations with others) (Ryff &
Keyes, 1995). In addition to these measures, the study also
included an assessment of positive affect and negative
affect to evaluate comparability with prior findings.
Based on previous research, it was hypothesized that, as the
number of roles increased, eudaimonic well-being and
positive affect would increase for both women and men,
and negative affect would decrease. To the extent that
multiple roles require resources and knowledge to be
maximally beneficial for well-being, it was expected that
the links between increasing roles and increased well-being
would be most evident among those with high education.
The third hypothesis predicted that the relationship
between multiple roles and well-being would be moderated
by perceived control. Thus, it would be among individuals
with high multiple roles who possessed a high sense of
control that the pattern of higher eudaimonic well-being,
higher positive affect, and lower negative affect would be
most strongly evident. Regarding the issue of education, a
corollary hypothesis was that the moderating effects of
control would be particularly consequential for understand-
ing links between multiple roles and well-being among
those with low education. This study was based on a
national sample of adult men and women that afforded
extensive socioeconomic variability, thereby broadening the
focus beyond smaller, less representative samples.
We employed measures of eight roles, thereby allowing
us to address the observation by Thoits (1986) that many
studies were limited to only the roles of spouse, parent, and
804Sex Roles (2006) 55:801–815
employee. A more rigorous operationalization of multiple
roles was also employed. Previously, individuals were often
considered to be occupying a role if they indicated ever
spending any time performing behavior indicative of the
role. This could have led to an inflation of role involve-
ment, as individuals who spent very little time in a role
could have been categorized as fulfilling the role. Finally,
the focus on positive psychological measures that have not
been included in previous multiple roles research also
constituted a unique emphasis in the present study.
Sample and procedures
Data were from the Midlife in the United States (MIDUS)
survey, which included a 45 min telephone interview and a
mailed questionnaire that took approximately 2 h to complete.
Participants were recruited between 1994 and 1995 using
random-digit dialing. The response rate was 70% for
telephone interviews and 87% for the self-administered
questionnaires (Keyes, Shmotkin, & Ryff, 2002). The total
sample consisted of 3,032 noninstitutionalized, English-
speaking women and men between the ages of 25 and
74 years who lived in a household with at least one
telephone and resided within the 48 contiguous US states
(Brim, Ryff, & Kessler, 2004; Keyes et al., 2002). The
present analyses were conducted with 2,634 individuals who
completed all necessary information to calculate the multiple
role variable. This subset of the total sample included 1,299
men (49.3%) and 1,335 women (50.7%). This subset was
also divided by education; 220 participants had less than a
high school education (8%), 768 had a high school degree or
GED (29%), 836 had some college or a 2-year degree (32%),
and 808 had at least a 4-year college degree (31%).
Due to missing data on the multiple role variable, 398
respondents were not included in any analyses. In compar-
ison to the sample used for analyses, the group excluded
due to missing data included more women, t(3,030)=−2.27,
p<0.05, more older respondents, t(3,003)=−8.03, p<0.001,
and more individuals with lower educational levels,
The means and standard deviations for all variables, save
gender, are provided in Table 1. With regard to sociodemo-
graphic characteristics, gender was coded as man=0,
woman=1, and age was included as a continuous variable
with a range of 25–74 years. Education was assessed based
on responses to a question asking the respondent to indicate
their education level from 12 options of increasing
education. After we reversed the order of two responses
(three or more years of college, and graduated from a 2-
year college or vocational school, or associate’s degree) the
education variable was treated as a continuous variable with
increasing values corresponding to increasing education.
Multiple roles The multipleroles variablewas constructed by
from among eight roles, including spouse, parent, employee,
religious participant, friend, volunteer, social organization
member, and caregiver. Although it is difficult to place
restrictions on the roles of spouse and parent, the remaining
roles are open to discussion of what behaviors, or what levels
of behaviors, are adequate to confer upon an individual a role
status. Empirical information can guide decisions as to what
level of a behavior is necessary to constitute a role. Frequency
distributions were examined for questions used to define
involvement in each role, and cutoff values were selected that
would include approximately 30% of respondents in a role.
This restriction narrows the definition of multiple roles so that
only those individuals who spent considerable time in a role
were considered to occupy the role. After we selected the
either occupied a role or did not occupy a role based on the
requirements necessary for a role status.
The role of spouse was determined based on a question
that asked if the respondent was married, separated,
divorced, widowed, or never married (married=1, any
other response=0). Sixty-four percent of respondents
indicated they were married. Respondents with any children
between infancy and 18 years of age (37.6% of sample)
were coded as 1=parent, and respondents with no children
were coded as 0. Parents who only had adult children
(44.9% of sample) were assigned one-half of a role, on the
rationale that their responsibilities were not equivalent to
those individuals who were parenting younger children.
Table 1 Means and standard deviations of variables included in
Positive relations with others
Purpose in life
Sex Roles (2006) 55:801–815 805
The role of employee was based on whether the respondent
was currently working full-time at a job (yes=1, no=0).
The role of full-time employee was occupied by 59.7% of
respondents. Individuals were awarded one-half of a role if
they indicated that they were currently working part-time at
a job (yes=0.5, no=0). The role of part-time employee was
occupied by 17.9% of respondents.
The cutoff values for the remaining roles were chosen by
examining frequency distributions, as explained above.
Those coded as 1=religious participant (35.6% of sample)
reported currently attending religious services at least four
times per month; those coded as 0 reported attending
religious services less than four times per month. The role
of friend was based on how often the respondent was in
contact with any of their friends through visiting, phone
calls, letters, or electronic mail. Those coded as 1 (60.7% of
sample) reported that they had contact with a friend at least
several times per week; all others were coded 0. Although
this cutoff allowed more respondents to occupy the friend
role than was allowed for other roles, it seemed too
restrictive to categorize individuals in the role of friend only
if they had contact with a friend every day, which was the
next available response category. The role of volunteer was
based on four questions about the number of hours per
month the respondents did formal volunteer work for
various organizations, which included: (1) hospital, nursing
home, or other health-care-oriented, (2) school or other
youth-related, (3) political organizations or causes, and
(4) any other organization, cause, or charity. The number of
hours for each service was summed, and respondents were
coded as 1 (27.6% of sample) if they currently participated
in volunteer work for at least five hours per month; those
who volunteered fewer hours were coded as 0. The role of
social organization member was based on the frequency of
attending meetings of four types: (1) religious groups,
(2) unions or other professional groups, (3) sports or social
groups, and (4) other groups (not including any required by
their jobs). The number of times per month for each type
was summed, and respondents were coded as 1 (32.3% of
sample) if they attended meetings at least four times per
month; all others were coded 0. The role of caregiver was
based on the number of hours per month respondents spent
providing unpaid assistance to five different groups of
individuals, including: (1) parents or the people who
raised them, (2) their in-laws, (3) their grandchildren or
grown children, (4) any other family members or close
friends, and (5) anyone else (such as neighbors or people
at church). The total number of hours was summed
across the five groups, and respondents were coded as 1
(28.1% of sample) if they provided such assistance for at
least 20 h/month; all others were coded 0.
The multiple roles variable was created by summing the
total number of roles each respondent occupied; scores
ranged from 0 to 8.5. However, due to the small number of
individuals who occupied eight or more roles, individuals
with eight or more roles were combined with those
individuals who occupied 7.5 roles for analyses. Thus, the
final range was 0 to 7.5.
Perceived control Perceivedcontrolwasassessedwithascale
adapted from Prenda and Lachman (2001). It combined four
items that assess personal mastery (e.g., whether or not I am
assess perceived constraints (e.g., I have little control over the
from 1=“agree strongly”,to 7=“disagree strongly”. Ratings on
both sets of items were reverse scored and the two subscales
were averaged so higher scale scores reflect higher standing in
each dimension. For the combined control scale, scores on the
perceived constraints scale were reversed and then averaged
with the mastery scale. Higher scores on the combined scale
indicate higher perceived control. Internal consistency of the
scale was high (alpha=0.85).
Psychological well-being The six dimensions of well-being
include positive relations with others, self-acceptance,
autonomy, personal growth, environmental mastery, and
purpose in life. Scale definitions are provided in Ryff and
Keyes (1995); factor analyses support the six factor
structure. Scale intercorrelations were weak to modest and
ranged from 0.13 to 0.46. These dimensions of well-being
have been shown to have different age and sex trajectories,
and the patterns obtained for each outcome have been
replicated across different studies with measurement instru-
ments that varied in the number of items used to assess each
dimension (Ryff & Keyes, 1995). In MIDUS, each
dimension was assessed with three items rated on a scale
that ranged from 1=“agree strongly”, to 7=“disagree
strongly”. Scale scores were summed across the items
following the reverse coding of positive items, so that
higher scores reflect greater levels of well-being. These
scales have moderate internal reliability: alpha=0.58 for
positive relations with others; alpha=0.59 for self-accep-
tance; alpha=0.48 for autonomy; alpha=0.55 for personal
growth; alpha=0.52 for environmental mastery; alpha=0.36
for purpose in life. These lower alpha coefficients resulted
from the a priori decision to represent the multidimensional
structure of the longer well-being scales rather than to
maximize internal consistency (see Keyes et al., 2002; Ryff
& Keyes, 1995). The short form scales correlated 0.70 to
0.89 with the longer parent scales (Ryff & Keyes, 1995).
Positive affect was measured with six items on which
respondents rated how often during the past month they had
felt cheerful, in good spirits, extremely happy, calm and
peaceful, satisfied, and full of life on a scale that ranged
from 1=“all of the time” to 5=“none of the time” (see
806 Sex Roles (2006) 55:801–815
Mroczek & Kolarz, 1998). The scale score was constructed
by averaging across the six items and reverse coding so
higher scores reflect higher positive affect. This scale
exhibits good internal reliability, with an alpha of 0.91.
Negative affect was assessed with six items on which
respondents rated how often during the past month they had
felt so sad that nothing could cheer them up, nervous,
restless or fidgety, hopeless, that everything was an effort,
and worthless on a scale that ranged from 1=“all of the
time” to 5=“none of the time” (see Kessler, Mickelson, &
Williams, 1999). This scale score was constructed by
averaging across the items and reverse coding such that
higher scores represent higher negative affect. This scale
exhibits good internal reliability, with an alpha of 0.87.
Hierarchical multiple regression was used to test the core
hypotheses of the study. Separate equations were used for
each of eight dependent variables, including the six dimen-
sions of well-being and positive and negative affect. Socio-
demographic variables (i.e., gender, age, and educational
attainment) were entered in Model 1. The number of roles
occupied was added in Model 2 to test the hypothesis that
involvement in greater numbers of roles would significantly
and positively predict psychological well-being and positive
affect after controlling for sociodemographic variables.
Significant negative predictions were hypothesized for nega-
tive affect. Interactions between the number of roles occupied
and education were inserted in a subsequent step to assess
whether the above prediction was restricted only to those with
higher educational attainment. Interaction terms were also
created with gender to investigate if the association between
multiple roles and well-being varied based on respondents’
gender. Model 3 addressed the hypothesis that perceived
control would moderate associations between the number of
roles occupied and well-being, and the interaction between
control and the number of roles was entered in Model 4. To
test the hypothesis that such moderating effects might be
three-way interactions between education, multiple roles, and
perceived control were also evaluated. Interaction terms were
also created with gender to investigate if perceived control
moderated the interaction between multiple roles and well-
being differently based on a respondent’s gender. Regression
coefficients for significant effects across Models 1–4 are
reported separately in Tables 2, 3, 4, and 5. Lower-order
interactions were included while testing for three-way
interactions, but the coefficients are not provided (coeffi-
Table 2 Hierarchical OLS regression coefficients for environmental mastery, purpose in life, positive relations with others, and positive affect.
VariablesModel 1 Model 2
Purpose in life
aThe R2values presented in Tables 2, 3, and 4 reveal that although multiple role involvement significantly predicts various aspects of
psychological well-being, the variability in well-being accounted for by multiple role involvement is relatively small.
*p<0.05. **p<0.01. ***p<0.001 (two-tailed tests)
Sex Roles (2006) 55:801–815 807
cients available upon request). Significant three-way inter-
actions are graphed according to the procedures described by
Aiken and West (1991).
For each significant interaction, we used the ARC 1.04
statistical package procedures to check for violations of
multiple regression assumptions. For each criterion variable,
the distribution of values was negatively skewed (responses
were more heavily concentrated in the higher ends of the
scale). These skewed distributions resulted in non-constant
variance (which was discovered by examination of the non-
constant variance plots) and non-normal error distributions
(which was discovered by examination of the histograms of
residuals and the q–q plots of the observed residuals as
compared against a normal distribution) for each criterion.
These violations may lead to biased standard errors. For each
significant interaction, a bootstrap of 10,000 samples was
tested. Bootstrapping is a technique in which repeated
samples are taken, with replacement, from the original sample
and then the data are reanalyzed (Pardoe & Weisberg, 2001).
The resulting empirical sampling distribution is used to
determine corrected standard errors. In each case, boot-
strapping did not change the results, as the coefficients
remained virtually the same, the standard errors changed very
little, and the significance of the interactions did not change.
ARC was also used to examine diagnostics and to identify
potential outliers. Cases were examined for their leverage,
distance, and influence by examination of the scatterplots.
Plots of Cooks-D (a measure of influence) were examined
closely, as influence is a measure ofhow muchan observation
changes an analysis and is a function of both leverage and
distance (J. Cohen, P. Cohen, West, & Aiken, 2003). For each
significant outcome, cases with a disproportionate influence
were removed and analyses were rerun to determine if the
significance of the outcome was changed by the high
influence cases. Some significant outcomes were influenced
by outliers, as indicated in the following summary of results.
Multiple Roles and Well-being
Multiple role involvement significantly and positively
predicted environmental mastery, positive relations with
Table 3 Hierarchical OLS regression coefficients for significant interactions for autonomy.
Variables Model 1Model 2Model 3 Model 4
−0.511 *** −0.077
−0.492 *** −0.075
0.020 0.022 ***
−0.511 *** −0.077
−0.492 *** −0.075
0.336 0.108 ***
0.797 0.005 **
0.020 0.022 ***
−0.010 0.000 −0.064
*p<0.05. **p<0.01. ***p<0.001 (two-tailed tests)
Table 4 Hierarchical OLS regression coefficients for significant interactions for personal growth.
Variables Model 1Model 2Model 3 Model 4
−0.024 *** −0.100
−0.022 *** −0.089
0.454 0.197 ***
−0.013 ** −0.053
0.217 0.057 ***0.199
−0.957 0.004 *
p<0.05. **p<0.01. ***p<0.001 (two-tailed tests)
808 Sex Roles (2006) 55:801–815
others, purpose in life, and positive affect, after we
controlled for gender, age, and education (see Table 2).
For these outcomes, multiple roles did not interact
significantly with any other variables included in the
analyses. As role occupancy increased, individuals had a
greater experience of effectively managing their lives and
surroundings; they demonstrated more trusting, loving
relationships with others; they reported a heightened sense
of purpose and meaning in their lives; and they reported
experiencing more positive emotions.
Multiple roles also significantly predicted all remaining
aspects of well-being (excluding autonomy) and negative
affect. However, these effects were qualified by various
moderating influences, which are described below.1
Although not the explicit focus of this study, gender
differences were found for some well-being outcomes.
Women reported more positive relations with others, lower
environmental mastery, and lower positive affect than men
(see Table 2). Women also had lower autonomy (see
Table 3), lower self-acceptance, and more negative affect
(see Table 5). However, responsive to the explicit aims of
the study, gender was also shown to interact with multiple
role occupancy, but only for the outcome of autonomy.
Sociodemographic factors and multiple roles
It was hypothesized that those with more education would
be better able to handle multiple roles, as higher educa-
tional attainment is linked both to the types of roles an
individual occupies as well as the resources and skills
individuals may have for managing multiple role demands.
A significant gender×education×multiple roles interac-
tion was found for autonomy. The upper portion of Table 3
displays the coefficients for the interaction, and the top of
Fig. 1 illustrates the interaction. The data support the
hypothesis for women, such that those with a high number
of multiple roles and higher levels of education had greater
autonomy than those with a higher number of multiple roles
and lower levels of education. For men, the interaction was
in the opposite direction of what was predicted, as men
with a higher number of multiple roles and a lower level of
education had greater autonomy than those with a higher
number of multiple roles and a higher level of education.
Perceived control and multiple roles
It was hypothesized that control would moderate the
association between multiple roles and psychological well-
being such that those individuals with a higher number of
multiple roles and a higher sense of control would report
greater well-being2and less negative affect, and that these
moderating effects would be particularly consequential for
Table 5 Hierarchical OLS regression coefficients for significant interactions for negative affect and self-acceptance.
VariablesModel 1 Model 2Model 3 Model 4
−0.006 *** −0.118
−0.031 *** −0.119 0.039 *** −0.029 *** −0.110
0.101 0.122 ***
−0.006 *** −0.126
0.096 0.068 ***
−0.008 *** −0.159
0.053 0.070 **
−0.008 *** −0.162
−0.192 *** −0.468
−0.048 *** −0.117 0.013 *** −0.037 *** −0.089
−0.298 *** −0.452 0.195 *** −0.401 *** −0.608
0.028 *** 0.427 0.012 ***
−0.444 *** −0.063
0.135 0.018 ***
0.537 0.276 ***
0.184 0.044 ***
*p<0.05. **p<0.01. ***p<0.001 (two-tailed tests)
1A multiple roles squared term was added to the basic models to test
for curvilinearity effects. There were significant quadratic trends for
environmental mastery, positive relations with others, self-acceptance,
and positive affect. In each case, the linear effect remained significant
and was stronger than the curvilinear effect. For the remaining
outcomes, the addition of the squared multiple roles term did not
change the significance of the linear effects.
2The construct of perceived control may overlap conceptually with
the constructs of environmental mastery and autonomy. The correla-
tion between perceived control and environmental mastery is 0.59 and
between perceived control and autonomy is 0.33. Correlations
between other variables are available upon request.
Sex Roles (2006) 55:801–815809
our understanding of links between multiple roles and well-
being among those with lower educational attainment.
A significant gender×multiple roles×control interaction
was found for autonomy. The lower portion of Table 3
displays the coefficients for the interaction, and the bottom
of Fig. 1 illustrates the interaction. The figure provides
support for the guiding hypothesis for women, as the
strongest link between increased control and increased
autonomy was evident among those with a higher number
of multiple roles. For men, the opposite pattern was
obtained. That is, it was among men with a lower, rather
than a higher, number of multiple roles that the association
between control and autonomy was strongest.
A significant education×multiple roles×control interac-
tion was found for personal growth. Table 4 displays the
coefficients for the interaction, and Fig. 2 illustrates the
interaction. One outlier was discovered when we examined
a plot of Cooks-D, and, after removal of the case, the data
were reanalyzed. Figure 2 demonstrates that perceived
control has a large main effect, but there are, nonetheless,
differences in slope. Thus, sense of control does appear to
moderate the link between multiple roles and personal
growth, but such effects are most prominent (as predicted)
among those with high role involvement but low educa-
tional standing, as well as among those with low role
involvement, regardless of educational status. As such, high
sense of control appears to compensate for the disadvan-
tages of low educational status in the face of high role
involvement (as hypothesized), but high sense of control
also appears to compensate for low role involvement,
regardless of educational status.
There was also a significant multiple roles×control
interaction for negative affect. The upper portion of Table 5
displays the coefficients for the interaction, and the top of
High MRs - Men
High MRs - Women
Low MRs - Men
Low MRs - Women
High MRs - Men
High MRs - Women
Low MRs - Men
Low MRs - Women
Fig. 1 Gender×education×
multiple roles interaction for
autonomy and gender×multiple
roles×control interaction for
810 Sex Roles (2006) 55:801–815
High MRs - High Education
High MRs - Low Education
Low MRs - High Education
Low MRs - Low Education
Fig. 2 Education×multiple
roles×control interaction for
High Multiple Roles
Low Multiple Roles
High Multiple Roles
Low Multiple Roles
Fig. 3 Multiple roles×control
interactions for negative affect
Sex Roles (2006) 55:801–815811
Fig. 3 illustrates the interaction. Although having a lower
sense of control is associated with greater negative affect
for those with high or low role involvement, the strongest
negative association between sense of control and negative
affect was found among those with low rather than high
role involvement. Overall, the highest reports of negative
affect came from individuals (both male and female) who
reported both low role involvement and a low sense of
control. Such findings indicate that having a greater sense
of control may offset the possible disadvantages of having
Finally, there was a significant multiple roles×control
interaction for self-acceptance that resembled the multiple
roles×control interaction for negative affect. The lower
portion of Table 5 displays the coefficients for the
interaction, and the bottom of Fig. 3 illustrates the
interaction. Three outliers were discovered when we
examined a plot of Cooks-D, and after removal of these
cases, the data were reanalyzed. As above, the strongest
positive association between sense of control and self-
acceptance was found among those with low role involve-
ment. The increase in self-acceptance associated with
higher control was greater among individuals with low role
involvement. As noted earlier, this pattern may suggest that
having a greater sense of control may offset the possible
disadvantages of having fewer roles.
This study was designed to investigate the association
between multiple roles and psychological well-being in a
national sample of adults in the United States. The
objective was to investigate a wide array of roles and to
assess whether the role enhancement hypothesis is applica-
ble across men and women and for those with varied
educational attainment. Sense of control was also investi-
gated as a potential moderator of the associations between
multiple roles and well-being, which included eudaimonic
measures rather than a focus only on hedonic measures or
negative outcomes. The use of more differentiated out-
comes facilitates identification of specific areas in which
multiple role involvement works together with other
variables, such as education.
The general hypothesis was that an increased number of
roles would be associated with increased psychological
well-being and decreased negative affect. Support for this
hypothesis was obtained for each measure except autono-
my. For women and men of varied educational status, an
increased number of roles was associated with better quality
relations with others, a greater sense of effectively
managing one’s life and surroundings, greater purpose in
life, and greater positive affect. These associations provide
further support for the expansion hypothesis and are
congruent with previous findings that link multiple roles
to increased positive outcomes, such as life satisfaction,
happiness, and self-efficacy (Adelmann, 1994b; Jackson,
1997; Martire et al., 2000; Miller et al., 1991; Pietromonaco
et al., 1986), and decreased negative outcomes, such as
1994b; Hong & Seltzer, 1995; Jackson, 1997; Martire et al.,
2000; Menaghan, 1989; Thoits, 1983, 1986). Increased role
involvement was also associated with more positive self-
regard, increased feelings of personal growth, and lower
negative affect for both women and men, although these
effects varied based on how much control and/or education
It was hypothesized that multiple roles would be
associated with higher well-being particularly among those
with high educational attainment. The rationale was that
those with more education may have not only better roles
(i.e., higher-paying and more autonomous jobs), but also
increased resources (e.g., financial, psychological) for
managing the demands of multiple roles. The results
provided support for this hypothesis for women, although
the effects were targeted on only two aspects of well-being:
autonomy and personal growth. High role occupancy was
linked with greater autonomy for women as educational
level increased. For men, however, the results were in the
opposite direction, such that high role occupancy was
linked with diminished autonomy as educational level
Interpretation of these effects is challenging. It is
possible that women with more education have more
resources and skills for managing multiple roles, and
thereby feel a heightened sense of autonomy in carrying
out their various commitments. Alternatively, women with
fewer roles and less education may maintain higher
autonomy by focusing their resources on a small number
of roles. Women with higher education but fewer roles may,
however, feel they have fallen short of expectations shaped
by their educational experiences. In other words, they may
feel that they should be accomplishing more and this
undermines their sense of autonomy. For men, those who
are highly educated and have many roles may feel less
autonomous because they have diminished flexibility in
their schedules as they become more involved in household
tasks. All of these possibilities need to be verified with
further assessments probing the above ideas.
Education was also implicated in the association be-
tween multiple roles and personal growth. Here, however,
sense of control moderated the association as well. It was
hypothesized that individuals with both high role involve-
ment and high control would have the strongest links to
higher well-being and that this would be particularly
consequential for individuals with lower educational attain-
812 Sex Roles (2006) 55:801–815
ment. The results supported this outcome, but also showed
that high sense of control was also more strongly linked
with the personal growth of those with low role involve-
ment, regardless of educational standing. Sense of control
thus appears to mitigate against the adverse consequences
of multiple situations: managing multiple roles despite
limited educational attainment, and dealing with limited
role involvement (regardless of educational attainment).
There were other outcomes for which control signifi-
cantly moderated the association between multiple roles
and well-being. Women and men showed positive links
between increased control and increased autonomy, but
these effects were strongest among women with high role
involvement, as predicted. It appears that sense of control is
even more important for enhanced autonomy for women
with high role involvement than it is for men or women
with low role involvement. Women with many roles may
find it more difficult to experience self-determination when
they have little control over their roles. Such effects may be
particularly true among women who occupy role combina-
tions that are difficult or stressful, although this interpreta-
tion needs to be evaluated empirically.
Perceived control was also a factor in the relation between
role involvement and negative affect and self-acceptance. The
associations between increased control, decreased negative
affect, and increased self-acceptance were stronger for those
with low role involvement. These findings point to an
unexpected moderating influence: namely, that sense of
control may help mitigate the downside of occupying fewer
roles, especially in terms of having positive self-regard and
less negative affect.
Overall, these findings suggest that, although positive
associations exist between role occupancy and positive well-
being and lower negative affect, some effects depend on an
individual’s gender or educational status, and many out-
comes are specific to particular dimensions of well-being.
Therefore, it is important for future researchers to consider
how links between role involvement and well-being differ
based on an individual’s sociodemographic background and/
or personal characteristics, such as sense of control. These
findings also point to aspects of psychological functioning
that have previously been missing in multiple roles research,
such as a sense of personal growth, positive relationships with
others, and sense of autonomy. These, as the findings
underscore, are differentially related to the experience of
occupying multiple roles.
A limitation of the present study is the cross-sectional
design. It may be that varied levels of psychological well-
being influence the roles in which people engage, or that
multiple roles and well-being have a reciprocal relationship.
For instance, individuals who have better profiles of
positive psychological well-being may pursue more roles
because they believe that they are capable of handling the
increased responsibilities and commitments, whereas indi-
viduals who have more unfavorable profiles of positive
psychological well-being may refrain from the pursuit of
multiple roles because they are already challenged by
attempts to manage their existing roles effectively. Follow
up research is needed with longitudinal data to assess how
these associations vary across time as individuals experi-
ence changes in their role occupancy.
The present study was also limited by the variables that
were available in the MIDUS study. Some of the mediators
and moderators of the links between multiple roles and well-
being that were suggested in previous research (Barnett &
Hyde, 2001) could not be examined because they were not
available. In addition, the MIDUS dataset did not have a
measure of role quality for all of the roles that were included,
so role quality could not be included in the present study. It
has been suggested that role quality is a fundamental
component of psychological well-being (Barnett & Hyde,
2001; Stephens & Franks, 1999), and the positive associa-
tion found between multiple role occupancy and psycholog-
ical well-being may be qualified by links that exist between
the quality of each role and psychological well-being. Future
researchers should explore potential moderators of the links
between multiple role involvement and well-being.
In addition, further studies are needed to examine how
different combinations of roles are linked with well-being
and how these associations vary by gender. Previous studies
have focused on the roles of parent, spouse, and employee,
but little attention has been given to additional roles that may
work in combination with those roles. It might be hypoth-
esized that a nurturing role such as caregiver, which many
individuals assume out of obligation, may more easily be
integrated into a woman’s role repertoire, while for men,
such a role may be incompatible with their current role
expectations. A role such as volunteer is pursued by choice
and may afford an individual greater psychological benefits
than a role such as caregiver, and it might be hypothesized
that volunteering provides psychological and physical
benefits regardless of gender. Future research needs to
investigate roles that provide benefits to individuals as well
as explore the mechanisms through which these roles make
an impact upon psychological and physical health.
project was provided by a grant from the National Institute on Aging
(P01-AG020166) to the University of Wisconsin—Madison. The
authors are grateful to Janet Hyde, Judy Harackiewicz, and Jeremy
Biesanz for helpful comments on an earlier version of this manuscript.
Support for the investigators at the MIDUS
Adelmann, P. K. (1994a). Multiple roles and physical health among
older adults. Research on Aging, 16, 142–167.
Sex Roles (2006) 55:801–815813
Adelmann, P. K. (1994b). Multiple roles and psychological well-being
in a national sample of older adults. Journal of Gerontology, 49,
Aiken, L. S., & West, S. G. (1991). Multiple regression: Testing and
interpreting interactions. Thousand Oaks, CA: Sage.
Ali, J., & Avison, W. R. (1997). Employment transitions and psycholog-
ical distress: The contrasting experiences of single and married
mothers. Journal of Health and Social Behavior, 3, 345–362.
Barnett, R. C., & Hyde, J. S. (2001). Women, men, work, and family:
An expansionist theory. American Psychologist, 56, 781–796.
Barnett, R. C., Marshall, N. L., & Singer, J. D. (1992). Job
experiences over time, multiple roles, and women’s mental
health: A longitudinal study. Journal of Personality and Social
Psychology, 62, 634–644.
Bird, C. E., & Ross, C. E. (1993). Houseworkers and paid workers:
Qualities of the work and effects on personal control. Journal of
Marriage and the Family, 55, 913–925.
Brim, O. G., Ryff, C. D., & Kessler, R. C. (2004). How healthy are
we? A national study of well-being at midlife. Chicago, IL:
University of Chicago Press.
Burton, R. P. D. (1998). Global integrative meaning as a
mediating factor in the relationship between social roles and
psychological distress. Journal of Health and Social Behavior,
Christensen, K. A., Stephens, M. A. P., & Townsend, A. L. (1998).
Mastery in women’s multiple roles and well-being: Adult
daughters providing care to impaired parents. Health Psychology,
Clark, S. L., & Weismantle, M. (2003). Employment status: 2000. U.S.
Census Bureau. Retrieved September 21, 2003 from (http://www.
Cochran, D. L., Brown, D. R., & McGregor, K. C. (1999). Racial
differences in the multiple social roles of older women: Implica-
tions for depressive symptoms. The Gerontologist, 39, 465–472.
Cohen, J., Cohen, P., West, S. G., & Aiken, L. A. (2003). Applied
multiple regression/correlation analysis for the behavioral
sciences (3rd ed.). Hillsdale, NJ: Erlbaum.
Goode, W. J. (1960). A theory of role strain. American Sociological
Review, 25, 483–496.
Hagger, M. S., & Orbell, S. (2003). A meta-analytic review of the
common-sense model of illness representations. Psychology and
Health, 18, 141–184.
Hong, J., & Seltzer, M. M. (1995). The psychological consequences of
multiple roles: The nonnormative case. Journal of Health and
Social Behavior, 36, 386–398.
Jackson, P. B. (1997). Role occupancy and minority mental health.
Journal of Health and Social Behavior, 38, 237–255.
Janzen, B. L., & Muhajarine, N. (2003). Social role occupancy,
gender, income inadequacy, life stage and health: A longitudinal
study of employed Canadian men and women. Social Science &
Medicine, 57, 1491–1503.
Jutras, S., & Veilleux, F. (1991). Gender roles and care giving to the
elderly: An empirical study. Sex Roles, 25, 1–18.
Kessler, R. C. (1982). A disaggregation of the relationship between
socioeconomic status and psychological distress. American
Sociological Review, 47, 752–764.
Kessler, R. C., Mickelson, K. D., & Williams, D. R. (1999). The
prevalence, distribution, and mental health correlates of per-
ceived discrimination in the United States. Journal of Health and
Social Behavior, 40, 208–230.
Keyes, C. L. M., Shmotkin, D., & Ryff, C. D. (2002). Optimizing
well-being: The empirical encounter of two traditions. Journal of
Personality and Social Psychology, 82, 1007–1022.
Kikuzawa, S. (2006). Multiple roles and mental health in cross-
cultural perspective: The elderly in the United States and Japan.
Journal of Health and Social Behavior, 47, 62–76.
Lachman, M. E., & Weaver, S. L. (1998). The sense of control as a
moderator of social class differences in health and well-being.
Journal of Personality and Social Psychology, 74, 763–773.
Link, B. G., Lennon, M. C., & Dohrenwend, B. P. (1993).
Socioeconomic status and depression: The role of occupations
involving direction, control, and planning. American Journal of
Sociology, 98, 1351–1387.
Marks, S. R. (1977). Multiple roles and role strain: Some notes on
human energy, time, and commitment. American Sociological
Review, 42, 921–936.
Marks, N. F., Bumpass, L. L., & Jun, H. J. (2004). Family roles and
well-being during the middle life course. In O. G. Brim, C. D.
Ryff, & R. Kessler (Eds.), How healthy are we? A national study
of well-being at midlife (pp. 514–549). Chicago, IL: University
of Chicago Press.
Martire, L. M., Stephens, M. A. P., & Townsend, A. L. (2000).
Centrality of women’s multiple roles: Beneficial and detrimental
consequences for psychological well-being. Psychology and
Aging, 15, 148–156.
Menaghan, E. G. (1989). Role changes and psychological well-being:
Variations in effects by gender and role repertoire. Social Forces,
Miller, M. L., Moen, P., & Dempster-McClain, D. (1991). Mother-
hood, multiple roles, and maternal well-being: Women of the
1950s. Gender & Society, 5, 565–582.
Mirowsky, J., & Ross, C. E. (1986). Social patterns of distress. Annual
Review of Sociology, 12, 23–45.
Mirowsky, J., & Ross, C. E. (1990). Control or defense? Depression
and the sense of control over good and bad outcomes. Journal of
Health and Social Behavior, 31, 71–86.
Moen, P., Dempster-McClain, D., & Williams, R. M. (1989).
Social integration and longevity: An event history analysis of
women’s roles and resilience. American Sociological Review,
Moen, P., Dempster-McClain, D., & Williams, R. M. (1992).
Successful aging: A life-course perspective on women’s
multiple roles and health. American Journal of Sociology, 97,
Mroczek, D. K., & Kolarz, C. M. (1998). The effect of age on positive
and negative affect: A developmental perspective on happiness.
Journal of Personality and Social Psychology, 75, 1333–1349.
Norton, T. R., Gupta, A., Stephens, M. A. P., Martire, L. M., &
Townsend, A. L. (2005). Stress, rewards, and change in the
centrality of women’s family and work roles: Mastery as a
mediator. Sex Roles, 52, 325–335.
Pardoe, I., & Weisberg, S. (2001). An introduction to bootstrap
methods using Arc. Retrieved January 22, 2003, from the
University of Minnesota, School of Statistics Web site: (http://
Pietromonaco, P. R., Manis, J., & Frohardt-Lane, K. (1986).
Psychological consequences of multiple social roles. Psychology
of Women Quarterly, 10, 373–382.
Prenda, K. M., & Lachman, M. E. (2001). Planning for the future: A
life management strategy for increasing control and life satisfac-
tion in adulthood. Psychology and Aging, 16, 206–216.
Ross, C. E., & Mirowsky, J. (1989). Explaining the social patterns of
depression: Control and problem solving—Or support and
talking? Journal of Health and Social Behavior, 30, 206–219.
Ross, C. E., & Mirowsky, J. (1992). Households, employment, and the
sense of control. Social Psychology Quarterly, 55, 217–235.
Ross, C. E., & Van Willigen, M. (1997). Education and the subjective
quality of life. Journal of Health and Social Behavior, 38, 275–
Ryan, R. M., & Deci, E. L. (2001). On happiness and human
potentials: A review of research on hedonic and eudaimonic well-
being. Annual Review of Psychology, 52, 141–166.
814 Sex Roles (2006) 55:801–815
Ryff, C. D., & Keyes, C. L. M. (1995). The structure of psychological
well-being revisited. Journal of Personality and Social Psychol-
ogy, 69, 719–727.
Sachs-Ericsson, N., & Ciarlo, J. A. (2000). Gender, social roles, and
mental health: An epidemiological perspective. Sex Roles, 43,
Schoon, I., Hansson, L., & Salmela-Aro, K. (2005). Combining work
and family life. Life satisfaction among married and divorced
men and women in Estonia, Finland, and the UK. European
Psychologist, 10, 309–319.
Sieber, S. (1974). Toward a theory of role accumulation. American
Sociological Review, 39, 567–578.
Simon, R. W. (1997). The meanings individuals attach to role
identities and their implications for mental health. Journal of
Health and Social Behavior, 38, 256–274.
Stephens, M. A. P., & Franks, M. M. (1999). Parent care in the context
of women’s multiple roles. Current Directions in Psychological
Science, 8, 149–152.
Stephens, M. A. P., Franks, M. M., & Townsend, A. L. (1994). Stress
and rewards in women’s multiple roles: The case of women in
the middle. Psychology and Aging, 9, 45–52.
Thoits, P. A. (1982). Life stress, social support, and psychological
vulnerability: Epidemiological considerations. Journal of Com-
munity Psychology, 10, 341–362.
Thoits, P. A. (1983). Multiple identities and psychological well-being:
A reformulation and test of the social isolation hypothesis.
American Sociological Review, 48, 174–187.
differences in distress. American Sociological Review, 51, 259–272.
Thoits, P. A. (1987). Gender and marital status differences in control
and distress: Common stress versus unique stress explanations.
Journal of Health and Social Behavior, 28, 7–22.
U.S. Bureau of the Census (1998). Population profile of the United
States: 1997. Washington, DC: U.S. Government Printing Office.
tions of family change. American Sociological Review, 52, 346–358.
Sex Roles (2006) 55:801–815815