Generativity in multiple roles

Abstract

Interest in the Eriksonian notion of generativity and its role in the lives of mature adults has recently increased. In the
present study, we examined generativity separately in the roles of wife, worker, and mother, and examined the utility of our
strategy relative to more global measurement strategies in explaining variation in well-being. Two samples of employed mothers
were studied, one sample employed in private industry and the other in a university setting. Statistical analyses demonstrated
that measurement equivalence existed across the two samples (i.e., that the patterns and magnitudes of factor loadings did
not differ significantly). For 8 of 11 indices of well-being examined across the two samples, role-specific measures of generativity
explained significantly greater variation than did global measures.

Full text

R
e
pr
i
nt
ed
from
Margi
e
E.
L
ac
hm
a
n
a
nd
Ja
c
qu
e
lyn
Boon
e
J
a
m
es
,
e
d
s.
MULTIPLE PATHS
OF
MIDLIFE
DEVELOPMENT
S
TUDI
ES
ON SUCCESSFUL
MIDLI
FE
DE
V
E
LO
P
M
E
N
T
Th
e
John
D
.
a
nd
C
a
t
h
e
rin
e
T.
MacArthu
r F
ounda
t
ion
Se
ri
e
s
on
M
e
nta
l
H
e
a
l
th and
D
e
ve
l
opm
e
nt
© 1997 by
Th
e
Univ
e
r
s
ity of
C
h
i
cago
A
ll
ri
g
ht
s
r
ese
rv
e
d

E I G H T
Th
e
G
en
e
rativity
of
Employ
e
d
Moth
e
r
s
In
Multipl
e
Rol
e
s
:
19
7
9
and
1991
S
h
e
ll
ey
M
.
M
acDermid
,
Ga
b
r
i
ela
H
e
il
brun,
a
nd
L
a
ura
G
ill
esp
i
e
D
e
H
aa
n
Although
wo
m
e
n
's
ro
l
es
h
a
v
e
rece
i
ve
d
in
c
r
eas
in
g
a
tt
ention in
r
ece
nt
yea
r
s
,
partly
b
eca
u
se
of
s
t
ea
dil
y
ri
s
in
g
p
ar
t
i
cip
a
tion
b
y
wome
n in t
h
e
pa
id labor
f
orce,
wom
e
n
's
d
e
v
e
lopm
e
nt
du
ri
n
g
middl
e
ad
ulthood
st
ill
i
s
not w
e
ll
und
ers
tood.
Ou
r
rese
arch
fo
c
u
ses
on
g
e
n
e
r
a
ti
v
ity
(
Er
ik
so
n
,
1
95
0) in
th
e c
onte
x
t of
multipl
e
r
ol
es as
a
m
a
jo
r
c
o
mpo
n
en
t
o
f em-
plo
ye
d
moth
ers'
m
idlif
e
ex
p
er
ie
n
ces.
W
e
a
tt
emp
t to
m
e
a
s
u
re
ge
n
e
r
a
ti
vity separa
t
e
l
y in
th
ree
ma
jo
r
ro
l
es
a
nd
th
e
n to link rol
e-
sp
ec
ifi
c
ge
ne
r
ativity to
wom
e
n
's
eva
lu
a
tion
s
o
f
their role
involv
e
ments and
th
e
ir
well
-
b
e
in
g
.
W
e
e
xa
m
i
n
e
ge
n
e
r
a
tivit
y
in two
sa
mpl
es
of
wom
e
n who
a
ll
are
moth
ers
,
wor
k
ers,
a
nd
wi
ves
.
D
a
ta were
c
oll
ec
ted from
th
e
fir
s
t
sa
mpl
e
in
19
78
-7
9
(
Ba
ru
c
h,
B
ar
-
n
e
tt
,
&
Riv
e
r
s,
1983)
,
a
nd
fr
om
th
e
seco
n
d
sa
mpl
e
in
199
1
(M
ac
-
Der
mid
&
G
ill
es
pi
e,
199
2
)
.
During
th
e i
n
t
erveni
ng
y
ea
r
s,
b
o
th
th
e
co
n
-
tex
t and
th
e
l
e
vel of
wom
e
n'
s
p
a
rticip
a
tion in
multipl
e
r
ol
es
c
han
ge
d
c
on
si
de
ra
bl
y
:
e
mploym
e
nt
b
eca
m
e
th
e
s
tati
s
ti
ca
l
norm
f
or
wom
e
n,
f
er-
tili
ty
r
a
t
es
fe
ll,
a
nd
divo
rce
r
a
t
es
s
t
a
bili
ze
d. I
n
thi
s
ch
a
pt
er,
w
e
exa
min
e
diff
e
ren
ces
b
e
tw
ee
n
th
e
se
two
s
ampl
es
in
th
e
int
e
rrelat
ion
s
hip
s
of
ge
n
-
er
ativity, rol
e
p
e
rform
a
n
ce
,
a
nd
well
-
b
e
in
g.
P
A
R
T
ICI
PATION
I
N
M
ULTIPLE
R
O
LES:
TH
E
C
O
NTEXTS
OF
GENERATIV
I
TY
Ge
n
e
r
a
ti
v
it
y,
th
e
s
e
v
en
th
a
nd long
es
t
o
f
E
rikson'
s
ei
g
ht
s
t
a
g
es
o
f
p
s
ychosoci
a
l
d
e
velopment, was originally
d
e
fin
e
d
a
s
"
th
e
c
oncern
i
n
es
tablishing and
guidin
g
th
e
n
ex
t
gen
era
tion
...
th
e
con
ce
pt
o
f
ge
n
-
er
ativity
is
m
e
ant to
includ
e
s
u
c
h
mor
e
po
p
ul
ar
s
ynon
yms
as
p
r
odu
c
-
ti
v
ity
a
nd
cr
ea
ti
v
ity,
which
,
how
e
v
e
r,
ca
nn
o
t
re
pl
ace
it
"
(
Er
ik
s
on,
19
5
0
,
p.
26
7).
Er
ikson
a
s
sociat
e
d with each
sta
g
e
a
ba
s
ic
s
t
reng
th or
virtu
e
.
In
th
e
ca
se
o
f
ge
n
e
rativity,
thi
s
w
as
"
care,"
wh
e
r
e
by
"
th
e
vit
a
lity of
a
n
order of care
[is
a
s
sured] to
thos
e
wid
e
a
r
eas of adult
in
v
olv
e
m
e
nt
s
which,
a
cc
o
r
din
g
to
a
Hindu
ex
pr
ess
ion,
g
u
ara
nte
e
th
e
'
maint
e
n
a
n
ce
o
f the
world
»'
(
E
rik
s
on,
E
rik
s
on,
&
Ki
v
nick, 1986).
F
in
a
ll
y,
E
rik
s
on
2
0
7

MacD
e
rmid,
H
e
ilbrun, and
D
e
Haan
viewed individual development
as
occurring in the context of role in-
volvements: "matured adulthood, then, means
a
set of vital involve-
ments in life's generative activities
...
participation in areas of
involve
-
ment in which one can learn to take care of what one truly cares for"
(Erikson, Erikson,
&
Kivnick,
1986,
p. 50).
Interest in adult development during midlife surged during the
1980s
after
a
lull during the previous decade. New evidence showed
that among adults approaching midlife, generativity was positively re-
lated to
age
(Darling
-
Fisher
&
Leidy,
1988;
Ochse
&
Plug,
1986;
Ryff
&
Heincke, 1983). And it
is
related to successful functioning in public and
private life (Vaillant
&
Milofsky, 1980). Extending Erikson's theory,
McAdams, Ruetzel, and Foley (1986, p. 802) demonstrated that "adults
whose biographical scripts for the future emphasize generativity
...
[scored] high on both power and intimacy motivation combined"
(though results were stronger for men than for women). The nature
of postmidlife relationships between generativity and age
is
less clear
(McAdams,
St
.
Aubin,
&
Logan, 1993).
Although frequently acknowledged
as
important, links between
generativity and role involvements have seldom been studied empiri-
cally, although Sherman (1987), for example, suggests that "problems
of identity in midlife are frequently those arising from
a
mismatch
between the person's sense of self and his or her role(s) in life" (p.
102). Conceptualizing generativity
as a
global personality trait, some
researchers have implicitly assumed that (1) individuals who perceive
themselves
as
generative are more likely to behave in generative ways,
and (2) generative individuals will be consistently generative across
role domains. Possibly
as
a
result of these assumptions, assessments of
generativity have focused on
either
thoughts
or
role involvements more
often than on both. In some cases, respondents have been judged by
researchers to be more or less generative on the basis of their general
perceptions of themselves (Ryff
&
Heincke, 1983) or on expert observ-
ers' ratings of generative themes in their plans or goals for the future
(McAdams, Ruetzel,
&
Foley, 1986), regardless of the degree to which
they were involved in roles where generative behavior might occur.
In other cases, respondents have been classified
as
generative or not
generative
primarily
on the basis of their participation in particular
roles (Vaillant
&
Milofsky, 1980): "the overriding factor governing the
clinical decision to put
a
man in this stage was his assumption of re-
sponsibility for other adults" (p. 1355).
20S

Employed
Moth
e
rs
in
Multip
l
e
Rol
e
s
An important contribution of recent studies by McAdams and his
colleagues
is
the measurement of both perceptual and
behaviora
l
aspects of generativity. For
e
x
amp
l
e,
V
a
n
d
e
Water and McAdams
(1989)
e
x
amined respondents'
paper-and-p
e
nci
l
se
l
f
-
reports of
gen
-
erativity and coders'
rat
i
ngs of generative themes in respondents' de-
s
c
r
iption
s
of their four
mo
s
t
import
a
nt
curr
e
nt
commitment
s
to
r
e
veal
a
correlation of
r
=
.
32.
Th
e
y concluded that "general cognitive atti
-
tudes about
g
enerativity,
..
.
are modestly
r
e
l
ated to generative
behav
-
io
r in
th
e
a
re
a
of adults' main life commitments (p.
447)
.
In
a
study
of
23
m
a
l
e
s
and
56
fem
a
l
es,
McAdam
s
and
St.
Aubin (1992) found
that scores on
th
e
Loyo
l
a
Generativity
Sca
l
e
(LGS) correlated
.59
with
a
49
-
item beh
a
vior checklist. More recently, McAdams,
Sto
Aubin, and
Log
a
n (1993) observed
a
co
rr
e
lation of
.53
b
e
tween respondents'
re
-
po
r
ts of generative concern (via the
LG
S
)
and
gen
e
r
a
tive
ac
tion (via a
40
-
item
behaviora
l
ch
e
cklist) and concluded that "there are meaning-
ful conceptual and
e
mpirical differences among generative concern,
gener
a
tive commitment, generative action, and generative narration"
(p.
227)
.
We found
thes
e
investigations intriguing. Since they demonstrate
that measures of self-perceived
globa
l
generative concern and
genera
-
tive action may be at least somewhat distinct, we became interested in
trying to pull
apa
r
t
g
enerativity to examine it
s
e
p
ar
ate
l
y in
specifi
c
role
s
.
.
The notion of studying development in the
conte
x
t of roles
is
not
new. Researchers of adolescence have been especially active in pointing
out that different identity domains (e.g., occupational, civic, interper-
sonal) may
dev
e
lop
a
t
diff
e
rent
rat
es a
nd thus should be attended to
separate
l
y (Archer,
1992;
Waterman, 1985). Indeed, Grotevant (1987)
argu
e
s
th
a
t
th
e
management of investments in roles and the forces
comp
e
ting with those inv
e
stments constitute the process through
which individuals form their identities. Regarding adults, Hornstein
(1986) calls for
a
"dynamic model of identity that inco
r
porates this
notion of multiple ro
le
commitments" (p. 552), and Juhasz (1989)
proposes
a
tripl
e-
h
e
li
x model of
d
e
v
e
lopm
e
nt
e
mb
e
dded in the roles
of family, work, and
s
e
lf.
Empiri
c
al
suppo
r
ting evidence comes from
Kroger and Haslett (1991), who found significant variability within
individuals across
five
identity domains (vocation, religion, politics,
general role values, and relationships) in adults' recollections from ad-
olescence to adulthood. Thus, "developments across identity compo-
209

MacD
e
rmid,
H
e
ilbrun
,
and
D
e
Haan
nents are not
paralle
l
,
even for people choosing the same
li
fe-sty
l
e"
(p.
323)
.
Women's generativity has so long been associated with mothering
that it can be difficult to conceive of generativity in other contexts such
as
employment and marriage. Erikson's inclusion of productivity and
care (among others)
as
generative
expr
e
ssions encourages us, however,
to broaden our conception.
As
adult role systems become more
com
-
p
l
ex and diverse, and
as
parenthood becomes
l
ess
predominant,
a
broad view of the loci and expressions of generativity increases in
im
-
portance
.
But what does generativity in
workp
l
aces or in marriage look
like?
In workplaces, individuals who choose to take
less
experienced
workers under their wing or to serve
as
mentors are demonstrating
procreative
g
e
nerative behavior intended to maintain the world of the
workplace and to invest in
a
younger generation. In
marriag
e
,
such
investments might focus on supporting spouses in their efforts to
become better
parents
.
Productivity, another key Eriksonian genera-
tive endeavor, might take the form of traditional workplace accom-
p
l
ishments like career advancement or the construction of
a
good mar-
riage
.
Interrelationships among Women's Roles
It may be particularly important to understand women's generativ
-
ity in the context of multiple ro
l
es
.
Women's
l
ife
courses are especially
likely to include discontinuities (Long
&
Porter,
1984)
such
as
moving
in and out of the roles of worker, student, and caregiver many times.
The role patterns displayed by women vary considerably, both across
individuals and over time (Barnett
&
Baruch,
1978;
Giele, 1982). In
the past, some theorists focused primarily on men's development, pro-
posing incremental and orderly developmental sequences that raised
questions about the app
l
icabi
l
ity of their work to women's lives
(Bar
-
nett
&
Baruch,
1978;
Hornstein, 1986). In the future, however, both
women's and men's role patterns are likely to become increasingly fluid
(Juhasz, 1989). Men's careers have become less orderly as the likelihood
of having only one employer has fallen and
as
the need for periodic
retraining has risen.
As
women have become increasingly involved in
the
l
abor force, pressure for men to become more
invo
l
ved in domestic
work has
risen
.
Longevity has risen and
ferti
li
ty has fallen, increasing
the likelihood of caring for both dependent children and
elders
.
Conse
-
quently, both role demands and the complexity of interdependence
210

Employ
e
d
Moth
e
rs
in
Multipl
e
Rol
e
s
among roles may increase (Barnett
&
Baruch, 1987), increasing the
utility of role-specific approaches.
Women may be particularly likely to experience conflicting de-
mands from their various roles (Long
&
Porter, 1984). Since women
occupy lower status jobs, on
ave
r
age, than men do, they have less power
to modify their conditions of work (Barnett & Baruch, 1987); and since
wives tend to have less power at home than husbands do, they also
have less power to modify their conditions of domestic work. Thus,
whil
e
men often may be able to substitute workplace labor for
unap
-
pealing family tasks, particularly when the workplace labor
i
s
lucrative,
women
les
s
frequently
hav
e
this option (Barnett,
1991;
Barnett
&
Baruch,
19
7
8).
While Barnett and Baruch (1987) characterize the
tradi
-
tional
hu
s
band-father role
a
s
"
low
-
strain
" because it includes both low
dem
a
nd
(i.
e
.,
minimal involvement in domestic work) and high con
-
trol (i.e., marital power) relative to their wives, "being
a
mother
is
rarely
a
ssociated with psychological
well
-
being and
is
often associated
with psychological distress" (Barnett, 1991,
p
.
11).
Studies of the interconnectedness of women's roles also show,
how
-
e
ver
,
that rol
e
demand
s
do not always conflict. In some instances,
po
s
itiv
e
ex
periences in one role
m
ay
compen
s
ate for negativ
e
one
s
in
another: stress in the parental role may be moderated by positive expe-
riences in the worker role, for
e
x
ample (Barnett,
1991;
Baruch
&
Bar-
n
e
tt, 1986); it does not appear, however, that dissatisfaction in the
mar
-
ital role can be compensated for (Barnett
&
Baruch, 1987). Such data
support the role enhancement perspective which proposes that
multi
-
ple
role
s
g
e
nerate energy for one another,
as
opposed to the conflict
and tension predicted by the role scarcity perspective.
Although it might be tempting to expect, on the basis of the
interde
-
p
e
ndence of
women
'
s
roles, that women would display generativity
consistently across roles, the negotiation and juggling carried out by
many women in order to participate in multiple roles may invalidate
such an
e
xpectation. The role
pattern
s
displayed by women are ex-
tremely diverse, and the
natur
e
of the interconnections among roles
varies
as a
function of both the roles and the individuals involved.
F
urth
e
r,
v
a
ri
a
tions in
me
a
ning and importance among roles may
re
s
ult
in individual differences in the consequences of involvement in the
same number of roles across individuals (Barnett
&
Baruch, 1985).
The implication for the present research
is
that misleading
ob
-
s
erv
a
tions could result from
as
s
uming that global and
role
-
specific
2
11

MacD
e
r
m
id,
H
e
ilbrun, and
D
e
Haan
generativity are fully redundant. We expected women to be generative
in many different domains of their lives and the degree to which they
were generative to vary across domains. Given the possibility of
consid
-
erable variation even among individuals displaying the same role
pat
-
terns, it seemed useful to control role pattern variation by studying
a
sample of people who are
all
involved in the same roles.
Limited existing evidence of links between generativity and specific
roles focuses primarily on the role of parent. In
a
longitudinal examina-
tion of the Radcliffe class of
1964
at ages 37, 44, and 48, Peterson and
Stewart (1992) found that women whose scores on the California
Q
-
Sort most closely conformed to those of
a
generative prototype defined
by expert judges were also more likely to have had children (r
=
.42
on average over time). These women were also "less tense, depressed,
angry and fatigued" (p. 8), and had higher
well
-
being scores on the
California Psychological Inventory. McAdams and
51.
Aubin (1992)
observed in
a
study of
66
male and
83
female respondents that "among
men especially, having been
a
parent was positively associated with
scores on the
LGS
(Loyola Generativity Scale)" (p. 1012). In
a
study
of adult men, Vaillant and Milofsky (1980, p. 1358) reported that gen-
erativity was correlated with experiences in the roles of parent (i.e.,
closeness to adolescent children), worker (i.e., assumption of manage-
rial responsibilities), spouse
(i
.
e.,
enjoyment of first marriage), and citi-
zen (i.e., voluntary public
service)
.
Although these studies focus on
interindividual rather than intraindividual variations, they do suggest
that particular kinds of role involvements may be more strongly related
than others to generativity and in turn to
well
-
being.
Looking beyond Role Occupancy
Rather than simply considering whether or not an individual
occu
-
pies
a
particular role, existing research suggests that it may be impor-
tant to consider respondents'
evaluations
of their experiences in partic-
ular roles. Barnett and Baruch (1985) found significant relationships
among the quality of experience in the work and parental roles and
role overload, anxiety, and role conflict. Later research with the same
sample (Baruch
&
Barnett, 1986) showed that the quality of experience
in particular roles was differentially related to aspects of well-being.
Giele (1982) argued that
a
theme in recent studies of depression and
well
-
being was the importance of "a sense of competence" (p. 21).
In this study we were more interested in respondents' assessments
of their experiences than in whether or not they occupied particular
212

Employ
e
d
Moth
e
rs
in
Multipl
e
Rol
e
s
roles. Only women who occupied all three roles of parent, worker, and
wife were selected for analysis. Extending the premises of the
role
-
quality research to generativity, we presumed that women who
evalu
-
ated their involvement in
a
particular role positively would be more
likely to report perceiving generativity in that role. We thus focused
on respondents' evaluations of their involvements, specifically on how
competent and satisfied they felt in each role, expecting perceptions
of generativity to be
a
partial function of feelings of competence and
satisfaction.
MULTIPLE EXPRESSIONS
OF
GENERATIVITY
In addition to broadening the range of roles within which generativ
-
ity
is
examined, it may also be useful to consider
a
diverse array of
generative expressions, or ways in which individuals might be genera-
tive. In the past, the study of generativity has been plagued by subtle
gender biases. Despite the breadth of Erikson's
conception,
it has often
been
operationalized
to emphasize expression by means of traditionally
masculine qualities like dominance rather than the equally important
quality of caring (see Hulsizer, Murphy, Noam, Taylor, Erikson,
&
Erikson, 1981), even in studies of women (e.g., Ryff & Migdal,
1984)
.
Several scholars have argued that considerations of adult development
have devalued or ignored the importance of nurturance and
relation
-
ships, thus systematically excluding individuals (most often women)
who are socialized to concentrate on those domains (Stewart
&
Gold-
Steinberg, 1990). For example, Gergen (1990) points out that the
"studies of Vaillant
...
and Gould
...
emphasize
a
male-oriented,
individualistic, rationalistic, and egocentric orientation to life,"
sug
-
gesting that "to judge from the major studies on lifespan development
at midlife, one would think only men survived the third decade of life"
(p.
475)
.
Stewart and Gold-Steinberg (1990) argue that "much of the
theory and research about midlife adults has been "universal" in prin-
ciple, but limited to men in terms of examples
...
,
populations,
sam
-
pIes
...
,
and theorists (Becker, Erikson, Gould, Guttman, Kotre, Levin-
son, McAdams, Vaillant; see Neugarten, 1968, and Fiske, 1980, for
female exceptions)" (p. 544). A concrete example
is
evident in Ryff
and Migdal's (1984) selection of measures focusing on dominance,
breadth of interest, and innovation to operationalize generativity.
At the other extreme, motherhood and nurturance sometimes have
been assumed to be women's only possible generative outlets,
exclud
-
ing the possibility of leadership (Barnett, 1991; Long
&
Porter, 1984).
213

MacD
e
rmid,
H
e
ilbrun,
and
D
e
Haan
Gergen (1990) argues that "women's adult development
is
staged
around the core notion of woman
as
reproducer"
(p
.
474), even though
"women are very responsible for t
h
e
maintenance of all life forms
within the culture, at least from the standpoint of their moral
responsi
-
bil
i
ty' their contributions to social cohesion, their support of the lives
of the competitively unsuccessful (e.g., the aged,
ill,
and young), and
their position
as
spiritual leaders, educators and
as a
yet
-
unfathorned
political force. Women might be viewed
as
those who create most of
the art,
contro
l
much of the money, support most of the cultural
in
-
stitutions, lead most of the volunteer groups, and influence most of
the decision-making processes in some manner" (p. 483).
(See
also
Franz
&
White,
1985;
Giele,
1982;
Gilligan,
1982;
Stewart
&
Go
l
d
-
Steinberg, 1990).
In recent years, however,
a
wider array of generative expressions
has been acknow
l
edged. For example, Kotre (1984) proposed the
exis
-
tence of four types of generativity (i.e., biological,
parenta
l
,
technical,
and cultural) and argued that historical changes
powerfu
ll
y influence
the contexts within which generative behavior occurs and the type of
generative behavior that might occur. Perhaps because his
conceptual
-
ization merged generative activity with the context or ro
le
within which
it might occur
(e.g.
,
biologica
l
generativity was narrowly defined
as
begetting, bearing, and nursing children and thus would occur
on
l
y in
the
parenta
l
ro
l
e), and because
h
i
s
detai
l
ed case studies focused exclu-
sively on eight
sub
j
ects (four of them women in their midthirties or
older), however, Kotre was able to observe only parental and cultural
generativity.
Additiona
l
discussions of diversity in generative themes focus on
the notions of agency and communion. Kotre (1984)
a
r
gues that
gen
-
erativity may be Agentie or Communal, with agency springing from
the
"se
l
f-asserting, self-protecting, self-expanding existence of the indi-
vidua
l
" (p. 16), and communion "representing the participation of the
individual in
a
mutual, interpersonal reality or in some larger organ-
ism"
(p
.
16
)
.
McAdams (1988) suggests that generativity involves both
agency and communion, "challeng[ingJ us
as
adults to be both power-
ful and intimate, expanding the self and surrendering to others in
th
e
same generative act" (p. 274), and cites significant correlations between
generativity and the sum of power and intimacy motivations
as
evi-
dence supporting his view (McAdams,
1988;
McAdams,
Ruetze
l,
&
Fo-
ley,
1986). Echoing these findings, in recent analyses of California
Q-
Sort items forming
a
generative prototype, Peterson and Stewart (1992)
214

Employe
d
Mot
hers in
Multipl
e
Rol
e
s
obs
e
rved
a
factor pattern suggesting what the
r
esearchers called Nur-
turant (corresponding to communal) and Prosocial
(
c
orresponding to
agentie) generativity.
Several researchers have used the fiction, letters, and diaries of
Brit
-
ish feminist, pacifist, and author Vera Brittain to identify diverse gener-
ative themes and to understand them in the role
conte
x
ts within which
they occur (e.g., Peterson
&
Stewart,
1990;
Stewart, Franz,
&
Layton,
1988;
Stewart, Franz, Paul,
&
Peterson, 1991). Generative themes
iden
-
tified by this psychobiographical research include productivity,
parent
-
ing, caring for others, and
a
need to be needed. Peterson and Stewart
(1990)
obs
e
rv
e
d the ebbing and flowing of generative themes
as
Brit-
tain's context and commitments
we
r
e
altered by World War
II.
For
example, while "caring
g
e
nerativity" rose
sh
a
rply with the onset of the
war and the departure of her children for safety, Brittain's "productive
generativity" reached its highest level in
a
decade when her children
returned in
1943
.
In this research, we focus on the two generative themes mentioned
by Erikson that seem to us to be most closely related to existing studies:
productivity and
procre
a
tivity. According to Erikson, "the productive
[aspect of generativity]
...
int
e
g
ra
t
e
s
work
lif
e
with
f
a
mily
lif
e
within
the political and technical framework" (Hulsizer et
aI.,
1981, p.
269)
.
He acknowledges that productivity may occur in "business" (p. 255),
but leaves open the possibility of productivity in other settings. Thus,
"products" could include children of whom one
is
proud,
a
good mar-
riage, or
a
successful career.
Procreativity
is
"primarily
a
concern in establishing and guiding the
next generation" (but not necessarily
one
's
own offspring;
see
Erikson,
1950, p.
26
7
).
The "procreative [aspect of generativity
]
..
.
gives birth
and responds to
th
e
needs of the next generation" (Hulsizer et
aI.,
1981, p. 269). Mentoring colleagues or otherwise "bringing others
along" thus may be procreative activities.
McAdams and
St.
Aubin (1992) contend that "generativity cannot
be understood from
a
single personal or social standpoint, but that it
must instead be viewed
as a
psychosocial patterning of demand, desire,
concern, belief,
commitm
e
nt, action, and
narr
a
tion
...
situat
e
d in
a
particular
soci
a
l
and historical context"
(p
.
1013). Although we cannot
examine here all of the psychosocial components of McAdams's
theo
-
retical vision, we do attempt to understand patterns of women's gen-
erativity across roles,
as
related to role perceptions and well-being, in
social and historical
conte
x
t.
215

MacD
e
rmid,
H
e
ilbrun, and DeHaan
RESEARCH QUESTIONS
We explored answers to the following questions:
(l)
Does generativ
-
ity vary across roles? (2) How does generativity expressed in particular
roles relate to women's assessments of their competence and
satisfac
-
tion in those roles and to their well-being? (3) How do relationships
among
role
-
specific generativity, role performance, and
well
-
being dif-
fer
as
a
partial function of sociohistorical time (i.e., data collected in
1978
-
1979 vs
.
1991)?
PROCEDURES
Sample
1978
-
1979 subsample. Procedures used to collect data from the
1978
-
1979
subsample are described fully in Baruch, Barnett, and Riv-
ers
(1983)
.
Data were collected in
a
northeastern city near Boston
whose inhabitants worked in jobs with
a
wide variety of occupational
prestige levels. Community voting lists were used to identify all women
aged 35-55; the 6,000 women identified were then contacted in ran-
dom order to determine if their family and employment circumstances
qualified them for the study (e.g., women were considered to be
em
-
ployed only if they had worked for pay at least
17.5
hours per week
for the three months prior to data collection). Certain groups of
women (i.e., those in high
-
prestige jobs) were oversampled to ensure
their representation in sufficiently large numbers for analysis. Over
76%
(n
= 238) of the women invited to participate completed
struc
-
tured
face-to
-
face interviews. Analyses in the present study focus on
the
45
women who were mothers, workers,
and
wives.
1991
subsample. Procedures used to gather data from the
1991
sub-
sample are described fully in MacDermid and Gillespie (1992). All re-
spondents were employed at
a
medium
-
sized bank in
a
midwestern
community of 50,000 near
a
large urban
center
.
Because bank
employ
-
ees are predominantly women who work at
a
wide variety of
occupa
-
tional prestige levels, they were considered a suitable comparison group
for the
1978
-
1979
sample
.
All employees
(n
=
367) were invited to
participate in
a
study of families and jobs in their workplace during
work
time
.
Screening questionnaires administered to the 257
employ
-
ees
(70%) who volunteered for the study identified women who had
worked at the bank for at least six months, who lived with
a
spouse
or partner, and who had children no younger than
six.
Of the
68
216

Employed Mothers in Multiple Roles
women that satisfied these conditions, 90% completed longer ques-
tionnaires, in the presence of research staff, tapping well-being, gen-
erativity, and aspects of role involvement. The 45 mothers aged
35-
55 constitute the sample of interest for this study.
Measures
Unless otherwise indicated, measures were identical in the 1978-
1979 and the
1991
data sets.
Role
performance variables
Competence.
Respondents' perceived competence in each of three
roles (parent, worker, spouse) was assessed using
a
single item: "In
general, how good or competent would you say you are
[as
a
parent,
at work,
as
a
spouse]?" with four answer options, ranging from
"Excel
-
lent" to "Not very good."
In
spite of the limited number of answer
options, distributions of the competence variables were well shaped
with minimal skewness
(
-
.1
on average across roles) and kurtosis
(
-
.3
on average). The average intercorrelation of competence scores across
roles was
.31.
Satisfaction.
Respondents' satisfaction with each role was assessed
using
a
single item: "All things considered, how satisfied or dissatisfied
have you been with your [parenting, work, relationship with spouse]
over the last two months?" The seven answer options ranged from
"Completely satisfied" to "Completely dissatisfied." Although the dis-
tribution was somewhat skewed
(
-
Ion
average across roles;
.6
kurto-
sis
on average), this single-item measure correlated well in each role
(r
=
.
58
on average) with eight-item semantic differential measures of
satisfaction in each role (available only in the 1991 data)
.
The average
intercorrelation of satisfaction scores across roles was
.32.
The competence and satisfaction items were widely dispersed
throughout the data collection
instruments
.
Means, standard
devia
-
tions, and ranges for competence and satisfaction in the roles of parent,
worker, and spouse are presented in Table
8.1.
To examine the validity of our measures, we first examined inter-
correlations between competence and satisfaction, expecting stronger
correlations
within
than
across
roles. The average correlation between
competence and satisfaction within roles was more than twice
as
large
(r
=
.25) than the correlation across roles
(r
=
.12)
.
We
a
l
so examined
inter correlations with indicators of
well
-
being, focusing specifically on
life satisfaction and self-esteem, because we expected these aspects of
217

MacD
er
mid
,
H
e
ilb
r
un
,
a
n
d
D
e
Ha
a
n
TABL
E
8.1
Characteristics of
M
eas
ur
es
Numb
e
r Standard
Po
ss
ible
Obtain
e
d
of Items Mean
Devi
a
tion
R
a
n
ge
Rang
e
P
are
nt role
G
en
e
rativity
4
13.4
1.9
0
-
16 7
-
16
Competence
2.9
.
63
1
-
4
1
-
4
Satisfaction
5
.
6
1.2
1
-
4
2
-
7
Worker role
Generativity
4
12.
2
1.9
0
-
16
7
-
16
C
ompetence
3.
3
.59
1
-
4 2
-
4
S
atisfaction
5.2
1.4
1
-7
1
-7
S
pouse role
G
e
ner
a
tivity
4
1
2
.9
2
.9
0
-
1
6
4
-16
C
ompet
e
n
c
e
2.8
.6
4
1
-
4
1
-
4
Satisfaction
5
.5
1.
4
1
-
7
2
-
7
well-being to relate differently to satisfaction
a
nd competence.
Consis
-
tent with our
pr
e
diction
s
,
lif
e
sa
tisfaction
wa
s
mo
re
s
t
r
ongly
r
e
lat
e
d to
rol
e
sa
tisfaction
(
r
=
.2
8)
than to
c
ompeten
ce
(r
=
.
19)
,
whil
e
self
-
est
e
em was more
strongl
y related to competence
(r
=
.5
2
)
than to role
satisfaction
(r
=
.25).
Role-specific gen
e
rativity.
Measures of generativity in the parent,
worker, and spouse roles comprised items
develop
e
d by Baruch,
Bar
-
nett, and Rivers (1983) to indicate
respondent
s'
p
e
rceived rewards and
concerns in each
role
.
Re
s
pondents were asked to
i
ndic
a
t
e
the
e
x
tent
to which they perceived certain features of their involvement in
a
par
-
ticular role
as
rewarding or of concern
(e
.
g.,
"
too
m
a
ny
conflict
s
with
children" was
a
potential concern associated with
th
e
parental role).
On the basis of descriptions gleaned from the generativity literature
(
E
rikson,
1950;
E
rik
s
on, Erik
s
on
,
&
Kivnick
,
1986
;
Hul
s
izer
,
Murphy,
No
a
m, Taylor,
E
rikson,
&
Erikson,
1981;
McAdams, Ruetzel,
&
Foley,
1986;
Ryff
&
Heincke, 1983), one reward item and one concern item
were chosen to
r
e
pres
e
nt the best
(i
.
e
.,
most consistent with
Erikson'
s
views on generativity) example of each of
five
possible ways of being
generative within
e
ach rol
e
(productivity,
procr
e
ativity, creativity, care,
and mastery)
.
Producti
v
ity, procreativity and
cr
e
ativity
a
re
e
x
plicitly
mentioned in Erikson's earliest definitions of
gen
e
rativity
(e
.
g
.
,
Erik
-
son, 1950). Care also
is
mentioned explicitly
a
s
a
strength or virtue
218

Employ
e
d
Mot
h
e
rs
in
Mul
t
ipl
e
Rol
e
s
a
s
sociat
e
d with generativity
(
E
rikson, Erikson,
&
Kivnick, 1986).
Fi
-
nally
,
mastery
is
implicated in
th
e
work of generative
individu
a
ls
"
to
d
e
v
e
lop
a
nd maintain those
s
oci
e
tal
in
s
titutions and natural
r
es
ource
s
without which successive
gen
e
ration
s
will not be able to
s
urvive"
(
E
rik
-
son,
E
rikson,
&
Kivnick, 1986, pp.
7
3
-
7
4).
A
total of 30 items was
chos
e
n,
5
reward items and
5
concern
it
e
ms
for
e
ac
h of the ro
l
es
of parent,
work
e
r, and spouse. Three
ex
pert judges
then
replic
a
ted the item
s
e
lection
proces
s
.
When the judge
s
c
onverg
e
d
on an
it
e
m that was theoretically
se
nsibl
e
but
dive
r
g
e
d from our
a
priori
choic
e
,
we altered our
c
hoice
.
Final percentage
agre
e
m
e
nt
s
b
e-
tw
e
en our
choic
es
and
tho
se
of the judges ranged from
7
6
% to 92%,
with
a
m
e
an of
84%
;
final Kappa
co
e
fficients
wer
e
adequat
e
,
ranging
from .32 to
.
77
,
with
a
mean of .59 (Bakeman
&
Gottman, 1986;
C
oh
e
n,
1960)
.
It
i
s
important to note
th
a
t different items were chosen to rep
rese
nt
each generative theme in each role. The
poo
l
of items used to construct
th
e
gener
a
tivity measures in both samples did not permit
th
e
use of
the
sa
me items for every
role
.
However, even if we had been able to
do
s
o,
w
e
w
e
r
e
not convinced
t
h
at
th
e
appar
e
nt comparability of
s
u
c
h
m
e
a
s
ure
s
would
neces
s
arily
en
s
ur
e
con
s
truct validity.
A
s
a
r
es
ult, w
e
relied upon theory and agreement by
ex
pert judges to
d
e
t
e
rmine our
final
m
e
asures. Although the
me
r
it of our strategy for dealing with this
thorny problem certainly can be
d
e
bated, we
feel
th
a
t our
re
s
ults at
least justify further
study
.
E
st
a
blishing
measurem
e
nt
e
quival
e
n
ce.
In order to establish
th
e
c
om
-
parability of our generativity
mea
s
ur
e
s
across the 1978-1979
a
nd 1991
samples, we used
LISREL
VII software (Joreskog
&
Sorbom, 1988) and
a
ma
x
imum likelihood nested hypothesis testing procedure that com-
pares the relative fit to the
e
x
isting data of factor
loading
s
set equal
across samples with unique loadings for each sample (Joreskog,
19
7
9).
In
thi
s
procedure, measures are usually considered equivalent if the
difference between chi-square tests of the fit of equivalent and unique
loading
s
(called
a
"chang
e
in
c
hi-
s
quar
e
"
t
e
st)
is
not significant.
On
ce
equivalence
is
established, the
qu
a
lity of the fit of the factor structure
to the data can be assessed using
se
v
e
ral criteria:
a
chi
-
squar
e
/
degr
e
es
of freedom ratio of less than
2
.
0
(Joreskog, 1979; Tanaka
&
Huba,
1984)
;
a
goodness of fit
inde
x of .85 or better
(Barnett
,
Marshall,
&
Sayer, 1991); root mean square residuals less than
.
10
(Rupp
&
Segal,
1989); and
t
tests of the
s
ignificance of each
indiv
i
dual factor
loading.
2
19

Ma
c
D
e
rmid,
H
e
il
brun, and
D
e
Haan
TABL
E
8.2
Meas
ur
e
m
e
nt
E
qui
va
le
n
ce
Tes
tin
g
1
99
1
Sa
mpl
e
1978
-
1
979
S
a
mp
l
e
Goo
d
ne
ss
Goo
d
nes
s
C
h
ange
x
2
/
d
f
X
2
di
ff
of
Fit
RM
S
R
of
Fi
t
R
MSR
in
X2
Pare
nt
R
ol
e
1.
71
64.89
(38)
**
.
85
.08
.
86
.
0
8
1.
62
7
1.34
(44
)
**
.
82
.0
7
.
84
.
09
6.45
(6
)
Work
er
Rol
e
1.0
5 3
9.80
(3
8
)
.
94
.
02
.
89
.
0
8
1.
23
54
.1
5
(4
4
)
.
90
.
0
4
.
87
.1
3
1
4.35
(6)
*
S
p
o
u
se
Rol
e
1.66
26.5
9
(
1
6)
*
.94
.0
5
.
88
.06
1.
58
3
1.
5
9
(
2
0
)
*
.
93
.07
.87
.0
8 5
.
00 (4)
*
p <
.
05
.
**p <
.
01.
T
h
e
199
1
d
a
t
a
we
r
e
u
se
d
t
o
i
d
e
nti
fy
th
e
nu
m
b
er
offac
tors
a
nd
th
eir
re
l
a
t
e
d
it
e
m
s;
fin
din
gs
th
en were
cr
o
ss-va
lid
a
t
e
d on
th
e
1978-
1
97
9
d
a
t
a
.
There
w
e
r
e
t
w
o
c
orr
el
a
t
e
d
fa
ctor
s
fo
r
eac
h
r
ol
e,
a
re
w
ar
d
f
a
cto
r
a
nd
a
concern
fac
tor
,
w
i
th
eac
h
rew
a
rd
o
r
c
on
cern item
lo
a
di
ng on
onl
y
o
ne
fac
tor.
S
in
ce
sa
mpl
e
s
i
ze
did not
p
er
mit
s
imult
a
n
e
ou
s
es
timation of
th
e
applic
a
bility
o
f
th
e e
quiv
a
le
nt
mod
el
to both samp
l
es
for
a
ll
thr
ee
rol
es,
eac
h
r
ol
e
w
as
a
n
a
l
yze
d
se
p
ara
t
e
ly.
A
s
Tabl
e
8.2
s
hows,
th
e c
h
a
n
ge
in
c
hi
-s
qu
are
w
as
n
on
sig
nifi
ca
n
t in
th
e
pa
re
nt and
s
pou
se
rol
es,
indicatin
g
m
eas
ur
e
m
e
nt
e
qui
va
l
e
n
ce
acr
o
ss
s
ampl
es
.
Althou
g
h
t
h
e c
h
a
n
ge
in
chi-
s
qu
are w
as
s
i
gn
i
fic
a
nt in
th
e
work
role
,
th
e c
hi
-s
qu
a
r
e
t
es
t of the
e
qu
iva
le
nt model
w
as
non
s
i
g-
nifi
c
ant
(indi
ca
tin
g
g
ood fit),
a
nd
th
e
oth
er
cr
i
ter
i
a
i
ndi
cate
d
g
ood fit.
F
or
th
ese
r
e
a
s
on
s
a
nd on
th
e
o
re
t
i
cal
g
round
s
,
w
e
did
no
t
re
j
e
ct
th
e
e
quiv
a
le
nt
mod
e
l.
On
e
it
e
m in
th
e
w
or
k
r
ol
e
did
n
ot
h
ave
a
lo
a
din
g
s
i
gn
ific
a
ntly
g
r
ea
ter t
h
an zero but
w
as
r
e
t
a
i
ne
d for
th
e
pu
r
po
ses
of
es
timating the
mod
e
l.
Th
e fi
t of the
equiv
a
l
e
nt
mod
e
l
to
th
e
data
fo
r
eac
h
sa
mpl
e
a
l
s
o
w
as a
d
e
qu
ate in
th
e
p
are
nt
an
d
spo
u
se
ro
l
es
.
Scal
e
scor
e
s
an
d
va
li
d
ity.
Althou
g
h
w
e
es
t
a
bli
s
h
e
d
m
eas
ur
e
m
e
nt
e
quival
e
n
ce
u
si
n
g
a
ll 23
items
c
ommon to both
th
e
19
78-
19
79
a
nd
1991
d
a
ta
s
et
s
(only
8
,
9
,
a
nd
6
i
t
e
m
s
w
e
r
e
av
a
il
a
bl
e
i
n
th
e
19
78-
1979
d
a
ta
fo
r the
p
are
nt
,
wor
k
er,
a
n
d
s
pou
se
r
ol
es
,
r
espec
ti
ve
l
y),
w
e
se
l
ecte
d
22
0

Employed
Moth
ers
in
Mu
l
tip
l
e
Roles
on
l
y
t
h
e
produc
ti
v
i
ty and procreativity items for further
ana
l
ysis
b
e-
cause these
i
tems were present in both samples for
a
ll
ro
l
es.
These items
are listed in
Tab
l
e
8.3. Items associated with
pro
du
ctivity, for
examp
l
e,
focus on opportunities for advancement (as a worker), good
commu
-
nication (as a wife), and
feel
i
ng proud of how children are
t
u
rn
i
ng
out (as
a
mother). We created total generativity scores in each role
by summing the raw reward and concern scores (with concern
ite
m
s
reversed). Raw scores were
u
sed instead of weighted scores beca
u
se
raw scores are easier to
interpret
,
because raw and weighted scores were
corre
l
ated .96 on average, and because weight
i
ng
is
"se
l
dom
wor
th
-
w
h
ile" (Gorsuch, 1983, p. 269). We summed rewards and concerns
because the factors were
corre
l
ated
.
Final generativity scores in the
par
-
ent, worker, and spouse roles were moderately
intercorre
l
ated (average
r
=
.
31)
.
Table
8
.
1
describes the final measures.
Va
l
idity of the
ro
l
e-specific generativity measures was supported by
their anticipated moderate
corre
l
ations with the 10-item
Dar
li
ng-
Fisher and Leidy (1988) generativity
subscale
,
which measures gen-
erativityas
a gl
oba
l
persona
li
ty trait
(availab
l
e
on
l
y in
t
h
e
1
991
data):
r
=
.26, .36, and
.35
(p
<
.
0
5), respectively, for
parent
,
worker, and
spouse generativity.
TABLE
8.3
Ro
l
e-specific Generativity Items
Reward Items Concern Items
Worker Role
Procreativity
Helping others develop Job doesn't fit my skills or interests
Productivity
Opportunity for advancement Job conflicts with other
responsi
b
ilities
Spouse or Partner Role
Procreativity
Spouse being a good parent Conflicts about children
Productivity
Poor communication
Parent
Ro
l
e
Procreativity
Helping them (children) develop Worry about the teenage years
Productivity
Feeling proud of how they are turning out Disappointment in what they are like
Good communication
221

MacD
e
rmid,
H
e
i
l
brun
,
and DeHaan
Outc
o
me
measu
r
es
Locus
of
contr
ol.
Perce
ptio
ns of
th
e
d
egree to
w
hi
c
h
c
i
rc
ums
t
a
n
ces
in
li
fe
are
u
nde
r
on
e's
p
ers
onal
c
ontrol
w
e
r
e
m
eas
ur
e
d with
a
se
v
e
n
-
it
e
m
sca
le
d
eve
l
ope
d
b
y
Pea
rlin
a
nd
Sc
hool
er
(19
78
)
.
Resp
ond
en
t
s
w
ere
asked to
stro
n
g
l
y agree, agree,
d
isag
r
ee, o
r
s
t
r
on
g
l
y
di
sagree
w
ith
d
e-
scriptive
statemen
t
s
li
ke, "I
h
ave
littl
e
contro
l
over t
h
e
t
hi
ngs t
h
a
t
h
ap-
pen to
m
e
.
"
Sc
o
res
range
d
f
rom
1
4
to
28,
w
ith
a
me
an of
22
.
2
(
SD
=
3
.
7)
.
Cro
nb
ac
h
's
a
l
p
h
a
was
.
77
.
1
2.
Depression.
D
efine
d
as
"sig
n
s
o
f
w
i
thd
ra
w
a
l
o
f
li
fe
int
eres
t
,
la
ck
of
moti
v
a
tio
n
,
a
nd
lo
ss
of
v
it
a
l e
n
ergy
...
feel
in
gs
of
hopel
ess
n
ess a
nd
f
u
t
ili
ty
.
..
"
(
De
ro
ga
t
is,
L
ipm
an,
Ric
kel
s,
U
hl
e
nhu
t
h
,
&
C
o
v
i
, 1
974
,
p. 4),
d
e
p
ressi
on was
meas
ur
e
d
w
i
t
h
lO
it
e
m
s
fr
om t
h
e
H
opkin
s
S
ymp
-
tom
C
h
ec
kli
s
t
(
De
ro
ga
ti
s
e
t
al.
,
1
9
7
4
).
Respo
nd
e
nt
s
w
e
r
e
as
k
e
d
"
Ho
w
muc
h
h
as
eac
h of t
h
e
fo
ll
ow
i
ng
sym
pto
ms
b
o
th
ered or
d
is
t
resse
d
y
ou
d
ur
i
ng
th
e
p
ast week
i
n
cludin
g
tod
a
y?
"
u
s
in
g
f
o
ur
a
n
s
w
er
op
tio
ns
rangi
n
g
f
ro
m
"n
ot
a
t
a
ll
" to
"quit
e
a
bit.
"
Sa
mpl
e
items
includ
e
d
"
f
e
el
-
in
g
l
one
l
y" and
"
f
ee
lin
g
tra
pp
e
d or
c
a
u
g
ht
."
C
ronb
ac
h'
s
a
lph
a
w
as
.86
i
n a
l
arge
va
lid
a
tion
sam
pl
e
(D
e
ro
ga
ti
s
e
t
a
l.
,
19
74
)
,
qu
i
t
e
c
omp
ara
bl
e
to
th
e
.8
7
va
lu
e
obt
ai
n
e
d in
thi
s
s
tud
y
.
'
Scores
ran
ge
d from lO to
24
(
o
ut of
a
possi
bl
e
4
0
;
hig
h
er scores
indi
ca
t
e
mo
re
depress
i
o
n)
,
w
ith
a
m
ean
o
f
1
4.4
(
SD
=
3.6
)
.
Se
lf
-esteem.
The
R
o
sen
b
e
r
g Se
lf-
Es
t
eem
sca
l
e
(Ro
se
nb
erg
,
19
7
9)
i
s
a
l
O
-
ite
m
sca
l
e
t
h
a
t asks
r
es
pond
e
n
ts to
s
t
r
o
ng
l
y
agree
,
agree
,
d
i
sagree,
o
r
strong
l
y
di
sagree
wi
t
h
a
se
t of
s
tat
e
m
e
nt
s
th
a
t
d
escr
ib
e
th
emse
lv
es.
Exa
m
p
l
e
of
i
tems in
th
e
sca
l
e
includ
e
"
On
th
e
whol
e
I
am
sa
ti
s
fie
d
with
m
y
se
l
f
'
a
n
d
"A
t
t
i
mes
I
think
I
a
m no
go
od
a
t
a
ll."
Sc
o
res
ra
n
ge
d
from
20
to
4
0
(
hi
g
h
er
sc
o
res
i
nd
ica
tin
g
hi
g
h
er
se
l
f-es
t
ee
m)
,
with
a
mean of 35.4
(
S
D
=
4
.
5
)
.
Cr
o
n
b
ac
h
's
a
lph
a
w
as
.86.
3
Life
s
a
tisfaction
.
W
e
u
se
d
th
e
Ca
mpb
e
ll
,
C
onv
erse, and
Rodg
ers
(1
976)
O
vera
ll
L
if
e
Sa
t
isfac
tio
n
i
tem to
meas
u
re
li
fe
sa
ti
sfaction
.
Thi
s
it
em
is
a
7-point
sema
n
t
ic
di
ff
ere
nti
a
l;
sc
or
es
r
a
n
ge
d
fr
om
2
to
7
(h
ig
h
er
sc
o
res
indi
cat
i
ng
g
r
ea
ter
satis
f
a
cti
on
),
with a mean
o
f
5
.
3
(
SD
=
1.2)
.
T
hi
s s
i
ng
l
e-
it
em
m
eas
ur
e
c
orr
el
a
t
e
d w
e
ll
(r
=
.8
1
,
P
<
.001)
with an
eig
ht
-item
seman
ti
c
di
fferen
t
ia
l
measure of
sat
i
sfac
ti
on
(a
v
a
il
-
abl
e
on
l
y in the
1
9
91
da
t
a
).
Beca
u
se
depress
ion
,
l
oc
u
s
of
c
ont
r
ol,
se
l
f-es
t
ee
m,
an
d
li
fe
sat
i
s
f
ac-
tio
n were
s
t
ro
n
gly
i
n
t
ercorre
lat
e
d
(
.
59
on
average
),
a
w
e
ll
-
b
e
in
g
f
ac
to
r
w
as
crea
t
e
d to
re
du
ce
th
e
numb
e
r
o
f var
ia
bl
es
for
a
n
a
l
ys
is
.
A
scree
te
s
t
in
d
i
cated
th
e
ex
i
ste
n
ce
of
a
si
n
g
l
e
fac
to
r,
a
nd
t
h
e c
hi
-s
qu
are
t
es
t
o
f
th
e
s
u
ffic
ie
n
cy of
a
s
in
g
le
fac
to
r
i
nd
icate
d
ade
qu
a
t
e
fit
(
5.
1
(
2
)
P
= .08).
222

Emp
l
oy
e
d
Moth
e
rs
in
Mu
l
tip
l
e
Ro
l
e
s
F
ac
tor
loadin
gs
r
a
n
ge
d
fr
om
.
66
to
.83
(M
=
.76;
hi
g
h
er
s
cor
es
indicat
e
g
r
e
at
er
w
e
ll
-
b
e
in
g
).
A
N
ALY
SES
AN
D
R
ESULTS
D
e
mo
gra
phic
C
omp
a
ri
s
on of
th
e
19
78
-
1
97
9
a
nd
1991
S
a
m
ple
s
Ta
bl
e
8.4
indi
ca
t
es
th
at
th
e
p
re
do
m
i
na
n
tly
E
urop
ea
n
A
m
er
i
c
an
sa
mple
s
did not diff
e
r
s
i
g
nifi
ca
ntl
y in age, duration of
m
arr
i
age
,
or
numb
e
r of
c
hildr
e
n
.
T
h
e
sa
mpl
es,
of
cour
se
,
ca
m
e
fr
om
diff
ere
nt
bi
r
th
c
oho
r
t
s
:
th
e
1
978-
19
7
9
sa
mpl
e
w
as
born
b
e
tw
ee
n
19
24
a
nd
19
44
,
whil
e
th
e
1991
sa
mpl
e
wa
s
bo
rn
b
e
tw
ee
n
19
3
6
an
d
19
5
6.
All
me
mb
ers of
th
e
1991
s
ampl
e
h
a
d at
l
eas
t
on
e
c
hild
o
ve
r
6li
v
in
g
a
t
hom
e,
wh
il
e
9
m
e
m
-
b
e
r
s
of
th
e
19
78-
19
79
sa
mpl
e
h
a
d
c
hild
re
n
und
er
7
,
a
nd
4
m
e
mb
ers
h
a
d
c
hild
r
en who had
all le
ft
ho
rne'
(
c
hi
-s
qu
are
(
2
)
=
15
.2
,
P
<
.001).
Not
s
urpr
is
in
g
l
y,
g
i
ven
hi
s
torica
l
trends, the
1
991
sa
mpl
e
rep
ort
e
d
lon
ger work
wee
k
s
th
a
n
th
e
19
78-
19
7
9
sa
mpl
e:
w
hil
e
th
e
1
978
-
19
7
9
s
ampl
e
was
a
lmost
e
v
e
nl
y
s
plit
b
e
tw
ee
n
pa
rt-tim
e
an
d
full
-
t
i
me
e
m
-
ployees
(46.
7%
a
nd
53.3%
)
,
th
e
1
991
s
ampl
e
a
lmo
s
t
ex
clu
s
iv
e
l
y
work
e
d
full
-
tim
e
(1
3
.
3%
a
nd
86
.7%
;
c
h
i
-
s
qu
a
r
e
(2)
=
11.
9,
P
<
.001).
H
u
s-
ba
nd
s
of
r
es
pond
e
nts in
th
e
19
9
1
sa
mpl
e
a
l
s
o
mor
e
s
trongl
y
pr
e
ferr
e
d
that
th
e
i
r
wiv
es
b
e e
mplo
ye
d.
Whil
e
no
hu
s
band in
197
8
-
19
7
9
e
x-
p
resse
d a
pr
e
f
e
r
e
nc
e
f
o
r
hi
s
wif
e
to
b
e e
mplo
ye
d, 72% of
th
e
hu
s
band
s
of
th
e
1991
res
pondent
s
pr
e
ferre
d
th
a
t
th
e
ir
wiv
es
b
e
em
plo
ye
d
"
(chi
-
square (3)
=
45.7
,
P
<
.
001).
T
h
e
r
e
w
as a
clear
s
oci
a
l
-
cl
ass
di
ffe
r
e
n
ce
b
e
tw
ee
n t
h
e
s
ampl
es,
indi
-
cat
e
d by
di
ffe
r
e
n
ces in
e
duc
a
tio
n, occup
ati
on
,
and
i
nc
om
e.
M
e
mber
s
of
th
e
19
78
-
19
7
9
s
ampl
e
were
mor
e e
du
ca
t
e
d than
th
e
1991
sa
mpl
e,
p
er
h
aps
a
p
art
i
a
l
fu
nct
io
n of t
he
i
r
geo
grap
hic
p
roxim
ity to
a
la
rge
numb
e
r of
univ
er
siti
es.
Whil
e
23
of
th
e
19
78
-
19
7
9
r
es
pond
e
nt
s
r
e-
ported
c
ompl
e
tin
g
a
dv
a
n
ce
d
gra
du
a
t
e
d
egree
s,
none of
th
e
1991
r
e-
s
pond
e
nt
s
h
a
d
a
dvan
ce
d
b
ey
ond
a
n
und
erg
r
ad
u
a
t
e
d
eg
r
ee.
Whil
e
24
o
f
th
e
19
78
-
19
7
9
re
spond
e
nt
s'
hu
s
b
a
nd
s
h
a
d
prof
essi
on
a
l
d
e
gr
ees
,
only
2
o
f
th
e
1991
res
pond
e
nt
s'
hu
s
band
s
h
a
d
s
u
c
h
d
e
gr
ees
.
T
h
ese
diff
ere
n
ces
were
c
on
ce
nt
ra
t
e
d in
th
e
upp
er
tw
o o
cc
u
patio
n
a
l
p
rest
i
ge
l
eve
l
s;
th
e
r
e
w
ere no
e
du
ca
ti
ona
l
di
ff
ere
n
ces
b
etween
th
e
sam
pl
es
i
n
th
e
l
owes
t
o
c-
cup
a
tional
p
res
ti
g
e
gr
ou
p
.
Wh
e
rea
s
th
e
1978
-
19
79
sampl
e
wa
s
a
lmo
s
t
e
v
e
nl
y
di
v
id
e
d
a
mon
g
wom
e
n in
hi
g
h
,
m
e
dium,
a
nd low
p
r
e
s
t
ige
job
s
,
th
e
19
9
1
sa
mpl
e
wa
s
con
ce
ntrat
e
d in medium
p
res
tig
e
jobs
(
c
hi
-s
quar
e
(
2
)
=
1
0.1
,
P
<
.01)
.
F
am
i
ly
incom
e
w
as
s
i
g
nifi
c
antl
y
hi
g
h
e
r in
th
e
19
78-
1
97
9
sa
m
p
l
e
than
in
th
e
1991
sam
pl
e
,
but
t
h
ere
w
ere no
s
i
g
ni
fica
nt
di
ffe
r
e
n
ces in
r
es
pon
-
223

T
A
BLE
8.4
Demo
g
raphi
c
Charact
e
risti
cs
of Samples
1975
-
1979
Samp
l
e
1
991 Sample
Ob
s
erved Observed
M
SV
Ran
g
e
M
SV
Range
A
g
e 44.4 6.5
35
-5
5
42.9
b
5.7
35
-
55
L
e
ngth of
marriage 21.2
b
S
.5
3-33
19.6
7
.
1
3
-
35
Numb
e
r
of
c
hildren
2.6 1.0
1
-
6 2
.
9 1.6
1
-
7
O
c
cup
a
tio
n
al prestige'
56
.4 5.6
1
6
-
Sl 47.S 7.9
26
-
72
High
(
%) 31.1
(
n = 14)
13
.
5
(n = 6)
Medium
(
%
)
35
.
6
(
n = 16)
6S
.
9
(
n =
31
)
Low (%)
33
.
3
(
n
= 15) 17.S (n =
S
)
Work hours/week
32.6 9.S
17
-
50
40
.
3
7.2
16
-
55
Annual
income
"
21.6
'
14.S
2.6
-
4
5.5
16f
7
.
9
3
.
0
-
40.0
Family
income
"
70
.
S
g
2S.7
25.S
-
103.
1
45
.
S
h
15
.
S
11.0
-
S0.0
Edu
c
ation
'
4.9
b
1.9
2
-
7
2.7
0
.
9 2
-
5
Spouse
'
s
edu
c
ation
'
4
.
0 1.2
2
-5
2
.
S
1.0
1
-
5
Significance
"
1.1
(
S7
)
1.0
(
S7
)
1.1
(SS
)
3.3
(
65
)**
4
.
1
(7
5
.
1)***
1.9
(62.6
)
+
4
.
7
(
62.5)***
7.1
(60
.
3
)***
5
.4
(
S6.
1)***
"
Comparisons
w
ere
c
onducted using
t
tests; noninteger degrees of freedom are reported
w
hen comparisons were
ba
s
ed on harmonic
means
.
b
Ba
s
ed on an
n
of 44 because l case has missin
g
data
.
c
Classified as high, medium, or low prestige using 1970 National Opinion
Resear
c
h
Cen
t
er
codes
.
High
-
prest
i
ge
j
obs
:
bank officers and
systems analysts;
medium
-
prestige jobs:
techni
c
ians and tellers;
low
-
prestige
jobs
:
clerks and proofreaders.
d
In
thou
s
ands.
I
ncomes
b
y the
1975
-
1979 sample
w
ere
ad
j
usted for inflation and
ge
o
grap
h
ic variations
i
n cost
o
f
li
vi
ng using tables
f
o
r
Annual Percentage Change in Earnings and Compensation
(
Bureau of the Census, 1980, 1987, and
1991
)
and the
Consu
m
e
r Price Index for
Se
l
e
cte
d Areas (Bureau of Labor
Sta
t
isti
c
s,
1979,
1990
)
.
'
Based on an n of
41
due to
2
cases missing data and the exclusion of
2
outlier
s
who
rep
o
rt
e
d
$
60,000
(
unadjusted
)
a
nnual
income
.
'
Bas
e
d on an n of 36 due to
9
r
e
fusals to report
income
.
g
Based on an
n
o
f
40 due to
3
cases
missin
g
data and
e
x
clusion of
2
outliers
,
"
Ba
s
e
d on an n of 34
du
e
to
11
refu
s
als to report
in
c
ome
.
i
l
= some high school or less;
2
= high
scho
ol diploma;
3
= some college or trade
scho
o
l
wi
thout
degr
ee;
4 =
a
ssoc
ia
te or
tr
a
d
e
sch
o
o
l
degree;
5
=
B
.
A
.
;
6
=
M
.
A
.
;
7
=
postgraduat
e
w
ork beyond
M
.
A
.
i
l = some high school or
less
;
2
= high schoo
l
diploma;
3
= some college or trade school
wi
th
o
ut
degr
ee
;
4
=
assoc
ia
te
o
r
t
rade
s
ch
oo
l
deg
r
ee;
5
=
professional
.

Employed Mothers
in
Multiple
Roles
dents' individual income across samples within occupational prestige
levels (footnotes for
Tab
l
e
8.4 explain how incomes reported in
1978
-
79 were adjusted for inflation and geographic disparities in cost of
living).
Because the demographic characteristics were strongly
intercorre
-
lated (average
r
=
.57), a socioeconomic status factor was
creat
e
d to
use
as
a
control variable in subsequent
analyses
.
This factor comprised
respondent's education, occupational prestige (continuous score), and
income, respondent's husband's education, and family income;
a
scree
plot supported the
e
x
istence of
a
single factor. Factor
l
oadings ranged
from
.
69
to
.82,
with
a
mean loading of
.76.
Do
Levels
of Generativity Vary across Roles?
To answer our first
resear
c
h question, we used
a
repeated-measures
analysis of variance in which the three
role
-
specific generativity scores
were treated
as
a
repeated measure (we
ca
ll
ed this
wit
h
in
-
subjects
fac
-
tor Role).
A
significant effect for Role (F
[
2
,
178)
=
8
.
5
,
P
<
.
001)
confirmed that there was significant variation across roles; we then
conducted
t
tests of the differences between each pair of roles. Results
showed that
av
e
rag
e le
vels of generativity
r
e
ported in the parent role
were higher than those in either the worker or the spouse roles
(parent
-
work:
t
= 4.8,
P
<
.
001;
parent
-
spouse
:
t
=
2
.
0,
P
<
.05)
.
Somewhat
higher generativity was reported in the role of spouse than in the role
of worker
(t
=
1.9,
P
<
.07).
Since the variation we observed in generativity across roles cou
l
d
simply be an artifact of measurement noise, we performed median
splits on each generativity score (creating high and low groups) and
examined the likelihood of falling into the high group in multiple roles.
Presumably if generativity
is
displayed consistently across roles
(i.e
.
,
does not vary),
a
person who
is
above the median in one role
shou
l
d
be above the median in the
others
.
This was not borne out by our chi-
square tests (average
chi
-
square (1)
=
1.7,
P
= n.s.).
How
Is
Rol
e
-Specific Generativity Related to
Role Performance and
Well
-
Being?
We next used path analysis to
e
x
amine the links among respon-
dents' perceptions of competence and satisfaction in each role, their
generativity in each role, and their well-being. Intercorrelations among
all analysis variables are shown in Table
8.5.
Four
hierarchica
l
stepwise
regressions were conducted: each of the measures of
role
-
specific gen-
225

T
A
BL
E
8
.
5
Pearson
C
or
relations
am
o
ng
V
ariables
U
sed in
Pat
h
An
al
y
ses
SES
PC PS
W
C
W
S S
C
SS
P W
S
S
o
c
i
oeconom
i
c
s
t
atus
(
S
E
S
)
Role performance
P
arent
compet
e
n
ce
(
P
C)
2
8**
Paren
t
sat
i
sfaction
(
PS
)
18
+
29
**
W
or
k
compe
t
ence (WC)
4
1
*** 2
4*
1
2
Work
satis
f
ac
ti
on (W
S
)
23
*
0
7
23
*
16
Spouse
co
m
petence
(
S
C)
18+
46
**
23
*
22
*
09
Spouse
sa
t
isfaction
(
SS
)
24
*
02 35
***
1
7
37
**
*
31
**
Generat
ivity
P
arent
(
P
) 17
32
**
51
**
*
16
17
1
7
+
15
W
orker (W)
22
*
24
*
2
7**
13
35
**
*
lO
14
27
**
Spouse
(
S
)
23
*
1
8+
52
*** 15
24
*
28
**
61***
44
*
**
21
*
W
ell-be
i
ng (W
B
)
2
7**
2
9**
38
**
29
**
28
*
*
3
6
**
47
**
32
**
3
0
**
56
**
+
p
<
.
10
.
*
p
<
.
0
5
.
*
*p
<
.01.
*
**
p
<
.
0
01.

Employ
e
d
Moth
e
rs
in
Mu
l
t
i
p
l
e
R
oles
erativ
i
ty was first regressed
o
n competence and satisfaction in
a
ll
t
h
ree
ro
l
es;
t
h
e
we
ll
-
being factor was then regressed on the three ro
le
-
specific
genera
tivity
scores
.
Socioeconom
ic
status was entered first
as a
control
variab
l
e
into
eac
h regression.
As
Figure
8.1
shows, t
h
e
path
mode
l
accounted for an average of
42% of t
h
e
variance in the dependent
variab
l
es
(ranging from 22% for
worker generativity to 50% for
spo
u
se
generativity). Generativity in
each role was
re
l
ated to satisfaction
b
ut not to perceived competence
in that
role
.
In
addition
,
generativity in the spouse role was
re
l
ated
to
b
ot
h
spo
u
sa
l
and parental
satisfact
i
on
.
We
ll
-
b
eing was
significan
tl
y
re
l
ated to generativity in the worker and spouse roles; it was not sig-
n
i
ficantly
re
l
ated to
generativi
ty in the
parenta
l
ro
l
e.
We next conducted ana
l
yses to compare the relative explanatory
power of our
pat
h
mode
l
(w
h
ich included
on
l
y indirect
re
l
ationships
between competence, satisfaction, and
we
ll
-
being)
wit
h (1) path
analy
-
ses
including
on
l
y the competence and
satisfac
t
ion varia
bl
es
(t
h
us
omitting the generativ
i
ty variables)
as
predictors of
we
ll
-being; and
(2) ana
l
yses
incl
u
ding
all
competence,
sat
i
sfact
i
on, and generativity
variab
l
es
as
direct predictors. The generativity variables
exp
l
ained sig-
nificantly more var
i
ability in well-being than the satisfaction and com-
petence variables (difference in
R
2
=
.0
4, F
[
I
,
89]
= 5.0
,
P
<
.05),
and if we included the six satisfaction and competence
variab
l
es
as
direct predictors, did not significantly increase the
exp
l
ained variability
beyond what was accounted for by generativity (difference in
R
2
=
.
03,
F [I, 89]
=
3.0,
P
>
.05). Thus, our or
i
g
i
na
l
mode
l,
which included
on
l
y indirect relationships between
well
-
being and competence and
satisfactio
n
,
provided
a
better fit to our data
t
h
an any of the alternatives
we
tested
.
Do Relationships among
Ro
le
Performance
,
Generativity, and
We
ll
-Being Differ by the Timing of Data
Co
ll
ection?
We first conducted an ana
l
ysis of varia
n
ce
to compare means across
the two samples. There were two significant differences:
1
978-1979
respondents perceived themselves
as
more competent at their jobs (3.5
vs.
3.0,
F
=
18.48,
P
<
.001) and
as
more generative in the spousal
ro
l
e
(13.6
vs.
12.1, F
=
6.4,
P
<
.
01) than did 1991 respondents. Consis-
tent with our earlier demographic comparisons, 1978-1979
respon
-
dents
a
l
so reported
significan
tl
y hig
h
er
soc
i
oeconom
ic
status than did
1991 respondents (.52
vs
.
-
.51
on the
SES
factor,
F
=
38.1,
P
<
.
00
1);
t
h
is
variab
l
e
was
contro
ll
ed in subsequent path ana
l
yses but was never
significant.
227

FIG
U
RE
R
.
I
.
Pat
h
Arialvsis "f P"ll
.
'
~m"l
p
P
are
nt
Ge
n
e
r
a
tlvlty
11
.
6
**
*
(3
.
83
)
.
2
9
W
o
rk
e
r
G
e
n
e
r
a
tlvlty
~
5
.
6*
**(4.
82
)
.
22
S
po
u
s
e
G
e
neratlvlty
27.
2
***(3.8
3
)
.
50
,
.
1
0
*
(
.05
)
.
2
1
[
Well
-
being
J
1
6
.
6
***(
3
.
85
)
.
37
I
n
f
orm
a
ti
on
r
e
por
te
d on
arrows
:
b
,
p,
(se)
.
B
.
I
n
f
orma
tion
r
e
po
rte
d
u
nde
r
b
ox
e
s
:
F
.
p
.
d
f
.
R-
squ
a
red
+
p<
.
10:
·
p<
.
0
5;
··
p<
.
0
1;
···p
<
.
0
01

Emp
l
oy
e
d
Mo
t
h
ers
i
n
M
ul
tip
l
e
Ro
l
es
Nex
t
,
we rep
li
cated
t
he
pa
th
ana
l
yses
separa
t
e
l
y within
t
h
e
19
78
-
1
979
a
n
d
1
99
1
s
ubs
a
mp
l
e
s
,
Res
ult
s
for
th
e
1978
-
1
979
s
u
b
samp
le
are
s
h
own in
F
i
gure
8
.
2
.
T
h
e
exp
l
ai
n
e
d
v
ar
i
a
n
ce
ra
n
ge
d
fr
o
m
26
%
f
o
r
w
e
ll
-
b
eing
t
o
6
1
% for spouse generativity,
wit
h a
mea
n
o
f
4
1
%
.
As
i
n
th
e
ana
l
yses of
t
h
e
fu
ll
samp
le
,
generati
vi
ty in
eac
h
ro
le
was
relate
d
t
o
sa
ti
s
-
faction
,
but not to
compet
e
nce,
th
e
on
e
exception
b
e
i
ng a positive rela-
tion
s
h
ip
b
e
tween competence and
ge
n
era
tivity
i
n the worker
ro
le.
Gen-
erativity in
t
h
e
spousa
l
ro
l
e
was
relate
d to
b
o
t
h
spou
sa
l
a
nd
parenta
l
satisfaction
.
I
n contrast to
t
h
e
ana
l
yses
of
t
h
e
fu
ll
samp
le,
we
ll
-
b
eing
w
as
re
l
at
e
d
e
x
clusive
l
y to
generativ
i
ty in
t
he
parenta
l
ro
l
e,
whic
h
ex
-
p
l
aine
d
2
6% of its
variation
.
Figure
8
.
3
shows
tha
t the variance
ex
pl
a
i
ne
d
b
y
th
e
pa
th
ana
l
ysis
of
t
h
e
1
991
sub
s
ample ranged from 16% for worker
generativ
i
ty to
54% for
we
ll
-
being
,
wit
h
a
mean of 33%. Generativity in both the
par
-
ent
and
th
e
worker
ro
l
es
was
re
l
ated to
s
atisfacti
o
n in
t
h
e
parenta
l
ro
le.
Generat
i
vity
i
n
t
h
e
spo
u
sa
l
ro
l
e
was
re
l
ate
d to satisfaction
i
n
t
h
at
ro
l
e
.
We
ll
-
being was
significan
tly
re
l
ate
d
on
l
y
t
o
ge
n
erativ
i
ty in
th
e
sp
ou
sa
l
ro
l
e,
w
hi
c
h accounted for 56% of
i
t
s
variation
.
A
s
a
fina
l
e
xp
l
oration of the
ut
i
li
ty
of
t
h
e
ro
l
e-specific
appr
o
ac
h
,
we
r
e
plicat
e
d our
regr
e
ssions of
we
ll
-
b
eing using
g
lob
a
l
i
ndicat
o
rs of
gen
e
rativity so
that
we
co
ul
d compare the
pr
o
port
io
ns of
exp
l
ai
n
e
d
variance. In each
s
ubsamp
le,
we averaged
generativ
i
ty scores across
ro
l
es to represent
g
l
oba
l
generativity and then regressed
we
ll
-bei
n
g
on
the new
measure
.
We
a
l
so
regresse
d
we
ll
-
b
eing
i
n
t
h
e
199
1
samp
le
on
the Darling-Fisher and Leidy (1988)
gl
ob
a
l
generativi
t
y
sca
le
(ava
il
a
bl
e
on
l
y in
th
e
1
991
data)
.
In the
1978
-
1979 data,
t
h
e gl
ob
a
l
approach
acco
u
nted for virtually
t
h
e
same percentage of variance as
t
h
e
ro
le
-
specific
appr
o
ach (25% and 26%,
respective
l
y). In
t
h
e
1
99
1
d
ata,
h
ow-
ever,
t
h
e
ro
le
-
spec
i
fic
approac
h
acc
o
unted for sign
i
ficantly more
var
i
-
a
nce (54%)
t
h
an
eith
e
r of the
gl
oba
l
approac
h
es
(47%
a
nd
37%, re-
spective
l
y,
for
generativ
i
ty averaged across
ro
l
es
a
n
d
t
h
e
Da
rlin
g-F
i
s
h
er
and Leidy scale).
DIS
C
USSION
We found support for the va
l
ue of examining
generat
i
v
i
ty as it
is
perceive
d in specific
roles
.
We first
propose
d
t
h
a
t
o
b
servat
io
n of
varia
-
ti
o
n
acro
s
s
ro
l
es
would
supp
o
rt
t
h
e
usefulness of
c
on
si
deri
n
g
gen
-
era
t
ivity in the context of
ro
le
invo
l
veme
n
ts
,
rat
h
er than as a
g
l
o
b
a
l
persona
l
ity trait.
Respon
d
ents did report significa
ntl
y
d
i
fferen
t
l
eve
l
s
of generativity across
r
o
l
es,
t
h
ough one
of
the differences was
on
l
y at
2
29

Parent
Competence
'"
,
Parent
Satisfaction
Worker
Competence
Worker
Satisfaction
Spouse
Competence
Spouse
Satisfaction
-I.J.:::I
.
2
4l
R
n
,
~
7
'
7 •
.
.
Parent
Generatlvlty
13
.
4***(
2
,
39)
.41
,
-
w
-
o
-
rk
-
e
-
r
---
12
(well-being
J
Generatlvlty
. .
6
.
4***(3
,
38) .34
7.1**(
2
,41)
.
26
Spouse
Generatlvlty
,
1
9
.
9
***(
3
,
3
8
)
.
61
In
f
orma
t
ion reported on
arrows
:
b
.
p
,
(se), B
.
Inform
a
tion
r
e
port
e
d
und
e
r
box
es
:
F
,
p
,
d
t
.
A
-
squ
a
red
+
p<
.
1
0
;
·p
<
.
0
5
;
··p
<
.
0
1;
·
··p
<
.
001

Parent
Competence
Parent
Satisfaction
Worker
Competence
Worker
Satisfact
i
on
Spouse
Competence
I
Spouse
Satisfact
i
on
Parent
Genera
t
lv
l
ty
'"
"
4.4
*
(
2
.42)
.
17
/
""'
W
orker
W
e
ll
-
b
e
i
ng
\.
Genera
t
lvlty
2
.
7+(3.4
1
)
.
16
1
6
.
3***(3
,
4
1)
.
54
/
""'
Spouse
Generatlvlty
\.
.
7
.
9***(4.40)
.
44
FIGU
R
E
8
.
3.
Pa
t
h
Anal
y
s
i
s
o
f
1
991
Subsamp
i
e
In
fo
r
ma
ti
on
r
epor
t
ed on
arrows
:
b.
p
,
(se)
,
B
.
I
n
f
orma
t
ion
repor
t
ed under
boxes
:
F.
p
.
df
.
R
-
squared
+p<
.1
0
;
·p
<
.
05;
··
p<
.
O
l;
·
··
p
<
.
OO
l

MacD
e
rmid,
H
e
ilbrun, and
D
e
Haan
the level of
a
trend. In addition, women who reported high generativity
in one role were no more likely than other women to report high
gen
-
erativity in other
roles
.
We interpret these findings
as
providing modest
evidence that there
is
significant variation across roles.
We next explored connections between role-specific generativity,
women's feelings of competence and satisfaction in each role, and their
well
-
being. We proposed significant interconnections that would vary
from role to role. We showed that generativity in particular roles was
generally related to satisfaction but not to competence, and that
satis
-
faction in the parental role was also related to generativity in the worker
or spouse roles, depending on which subsample was being examined.
The relationships between satisfaction in the parental role and gen-
erativity in other roles are not surprising given the prominence of the
parenting role in social expectations of women, the amount of time
women devote to the role, and the degree to which women adjust other
role involvements to accommodate parenting demands.
Generativity in particular roles was differentially related to women's
well-being. Specifically, in the sample
as a
whole, generativity in the
spouse and worker roles, but not the parent role, was significantly re-
lated to
well
-
being
.
The magnitude of the standardized regression coef-
ficient for the spouse role was more than twice
as
large
as
that for the
worker role (see
Fig.
8.1),
a
further indication that the strength of the
relationships between generativity and well-being varies across roles.
Finally, we were interested in how the relationships among
role
-
specific generativity, role performance, and well-being might differ
as a
partial function of sociohistorical factors. We found that while worker
generativity was related to both competence and satisfaction in the
1978
-
1979 sample, it was related to neither in the 1991 sample. Well-
being was linked to parental generativity in the
1978
-
1979 sample but
to spouse generativity in the 1991 sample. Caution
is
necessary, how-
ever, in interpreting these findings because of the large differences be-
tween our samples, although it
is
worth noting that the control variable
for socioeconomic status was not significant in any analysis.
Thus, our findings provide some evidence in favor of
all
three of
our propositions, supporting the value of considering generativity in
the context of multiple roles. We do not, however, suggest that global
approaches have no utility, since our
role
-
specific measures were corre-
lated with
a
global measure of generativity, and
a
global approach ac-
counted for
as
much variance
as
the role-specific approach in one sub-
sample.
232

Employed Mothers in
Mu
l
t
i
p
l
e
Ro
l
e
s
Clearly, t
h
e
parental
ro
l
e
was important in the lives of the women
in t
h
is
research regardless of when data were
co
ll
ected. Respondents
reported higher generativity (on average) in the
parenta
l
ro
l
e
than in
the
ro
l
es
of either worker or spouse.
Parenta
l
satisfaction was related
to generativity in more roles than satisfaction or competence in any
other role. Among 1978-1979 respondents, parental satisfaction was
related to both
parenta
l
and spousal generativity,
whi
l
e
among
199
1
respondents, parental satisfaction was related to
parenta
l
and worker
generativity. These patterns are
c
onsistent with some earlier views that
women's major generative
e
x
pression
is
the nurturing and caring they
provide in the family
context
.
Our path ana
l
yses
s
h
owed, however, that
whi
l
e
the well-being of the
1978
-
1979 subsample was tied
primar
il
y
to parental generativity,
we
ll
-
being in the
199
1
subsample was most
strong
l
y related to generativity in the
spouse
role
.
While the parental
role
is
clearly important, it
is
not the
on
l
y significant role context for
generativity.
Although we expected generativity to be related to perceptions of
both competence and
sat
i
sfaction, in general on
l
y satisfaction proved
to be related to generativity
.
Whi
le
the merit of the competence mea-
sures can
b
e
debated, it
a
l
so may be the case that women choose to
be more or less generative independent of how competent they
fee
l
in
a
particu
l
ar rol
e
.
For
e
x
ample, one might choose to invest time and
energy in socializing
a
less experienced
co
ll
eague regardless of how
competent one
fee
l
s
in one's own position.
Differences between the
1978
-
1979 and 1991 subsamples raise
questions about the power of the social context. The 1978-1979 re-
spondents had higher-level, more
l
ucrative positions than did t
h
e
1991
respondents
.
On average, they represented
a
group of
unusua
ll
y power-
ful and
we
ll
-off women in their social context, in which only about l
in
20
emp
l
oyed women was an executive (Bureau of Labor Statistics,
1991). Among these women, both competence and satisfaction in the
worker
ro
le
predicted worker
generat
i
vity, but
on
l
y generativity in t
h
e
parental ro
l
e
was related to well-being.
By
1990, the historical context
had changed considerably: the percentage of lawyers who were women
had quintupled from 4% to 20.6%; the percentage of computer
pro
-
grammer
s
who were women had
a
l
most
doub
l
ed, from 19.9% to 36%;
and t
h
e
percentage of physicians who were women had jumped from
10
.
1
% to
19
.
3% (Bureau of Labor Statistics, 1991). More than l in
I
D
employed women was an
executive
.
Among the 1991 respondents,
generativity in the worker role was independent of worker competence
233

MacD
e
rmid
,
H
e
ilbrun, and
D
e
Haan
and satisfaction, and well-being was related to generativity in the
spousal role and marginally related to generativity in the worker role.
Given their generally distinguished work roles, it was somewhat sur-
prising that the
well
-
being of the 1978-1979 respondents was exclu-
sively
a
function of generativity in the parental role. Conversely, it was
surprising that the worker generativity of women in the
1991
subsam-
pie, who spent more time working than women in the earlier sample,
was unrelated to competence and satisfaction.
One possible explanation
is
that links between generativity and well-
being conform to social expectations. Thus, even though the 1978-
1979 women worked in higher-level positions, on average, than the
1991
women, they lived in
a
social and historical context within which
parenthood was expected to be their primary source of
identity
.
Con-
sistent with these expectations, the well-being of women in the 1978-
1979 sample was related only to generativity in the parental role. At
the same time, the powerful work positions occupied by these women
may have afforded them more opportunities to be generative; hence
the positive relationships we observed between work competence, satis-
faction, and generativity. Since 1979, social expectations regarding the
prominence of the parental role for women may have loosened with
patterns of reduced fertility and delayed marriage, leading to the
greater importance of generativity in the role of spouse for well-being
among the 1991 respondents. Given that employment for women had
become the modal pattern, the women in the 1991 sample were much
less
unusual than their earlier counterparts; they also occupied
lower
-
level positions, which may have limited their opportunities to be gener-
ative; hence the lack of connection between worker competence, satis-
faction, and generativity. Of course, differences between findings for
the two samples might simply be an artifact of differences between the
samples independent of any changes in the sociohistorical context.
It
is
also important to note that well-being was related to generativity in
the worker role when the two subs am ples were analyzed together.
It
is
interesting to consider the findings of this study in light of
recent research concerning the role of spouse. Recall findings by Bar-
nett and Baruch (1987) reported earlier that women's negative evalua-
tions of the spousal role do not appear to be compensated for by posi-
tive experiences in other roles. They also suggest (1987, p. 139) that
the rising prevalence of dual-earner families
is
affecting the role of
husband by increasing its demands, with increased marital strain
as
one consequence. McKinlay, Triant, McKinlay, Brambilla, and Ferdock
234

Employ
e
d Mothers
in
Multip
l
e
Roles
(199
0
)
reinforce
th
i
s
suggestion
:
"Per
h
aps the most
i
mpor
t
ant finding
regarding the impact of nurturing
ro
l
es
i
s
the persistent effect of stress
from a spouse" (p.
133)
.
Th
u
s,
the greater
i
mp
o
rtance of
generativ
i
ty
in the spouse role for well-being in the more recently
co
ll
ected data
is
consistent with observations from other research
a
bout
possibl
e
shifts in
t
h
e
implications of
particu
l
ar roles.
Fina
ll
y, the differences in the
re
l
ative
ut
il
ity of
glo
b
a
l
and
ro
l
e-
specific approaches at different times of measurement are
i
ntrig
uing.
Among data collected in
1978-1979
,
t
h
e
globa
l
and
ro
l
e-specific ap-
proaches accounted for
virtua
ll
y
equa
l
proportions of
var
i
ance in
we
ll
-
being, while in the
1991
data,
role
-
specific approaches accounted for
significantly greater variability than
glo
bal
approac
h
es.
Again, although
these results
cou
l
d stem from important
preex
i
sting differences be-
tween the samples, they are also consistent with
i
ncreasing
discontinu
-
ity in women's ro
l
e
patterns and support our ear
li
er suggestion that
role-specific approaches may become
increasing
l
y u
sef
ul.
There are important limitations that must be taken into account
when considering our
findings
.
First, the research was
cross-sectiona
l
,
precluding any
con
j
ecture about the direction of the links between per-
ceived generativity in specific roles and
well
-
being.
It
is
quite possible
that individuals with positive
we
ll
-
being
simp
l
y report
perceiv
i
ng
themselves as more generative
beca
use they have a
genera
ll
y
pos
i
tive
outlook (Kotre, 1984). Presumably, however, such
individua
ls
a
l
so
would be more likely to report higher competence and satisfaction.
The discrepant patterns for satisfaction and competence in this study
seem to argue against a contagion perspective, although
on
l
y
l
ongit
u
di-
nal research will sort out this
re
l
ationship
.
Another concern
i
s
that the items we used to indicate generativity
were
origina
ll
y developed as
i
ndicators of
ro
le
quality (Baruch, Barnett,
&
Rivers, 1983). In our view, the ability to be generative in a given role may
be
a
very
l
egitimate, though distinct, component of
ro
le
q
u
ality
.
On
t
h
e
basis of the work of our judges, we are confident that the items we
se
l
ec
t
ed
represent role-specific generativity. Stronger comparability of generativ-
ity measures across roles would be gained, however, if the items
co
ul
d
be more similar in both content and meaning.
Our sample
is
small and
certain
l
y
not
representative, given that
a
ll
the 1991 respondents worked in one bank. We believe
t
h
at we have
provided sufficient evidence to justify
l
arger and more rigorous
stu
dies
of
both men and women.
Fina
ll
y, there are many other domains of generative activity we
235

Ma
c
D
e
r
mi
d
,
H
e
i
lbrun, and
D
e
Ha
an
should have asked about. The most serious omission probably was that
of generativity
e
x
pressed in
th
e
role of friend. In the future, more
free
-
dom needs to be allowed for
r
espondents to tell
u
s
how they construct
ge
nerativity.
Mor
e
sophisticated ways of thinking about and measuring
the breadth of
ge
nerative
ex
p
re
ss
ions also
ne
e
d to be
developed
.
Fi-
n
a
lly,
because of our
desir
e
for comparability between the
1978
-
1979
and
1991
data, only two
(i.
e
.,
productiv
i
ty and procreativity) of the
five
type
s
of generative expression that were of
int
er
e
s
t could
b
e e
xplored.
Important
r
e
search task
s
lie
a
head.
Fir
s
t, we need to understand
the
r
e
lationship
s
betw
e
en
rol
e
-
sp
e
cific generativity and the factors that
shape participation in
a
given
role
.
For
exampl
e
,
what impedes
genera
-
tive perceptions and
behavior
s
?
What facilitates generativity? What
is
th
e
process
throu
g
h which individuals
choo
s
e
to be (or not
b
e
)
genera
-
tive, and how
i
s
th
e
mode
o
f
ex
pr
es
sion chosen? In
short
,
which
indi
-
viduals will
f
eel
generative
a
nd which individuals will
be
gener
a
tive,
and in which domains under which conditions? The generativity of men
a
nd women
mu
s
t be studied longitudinally, in
a
wider
a
rr
a
y of domains.
It
also would be very helpful to understand difference
s
between birth
c
ohorts,
sin
ce
our
finding
s
hinted
a
t
pos
s
ible
s
hift
s
in
patt
e
rn
s
of
gen
-
era
tivity
acros
s
tim
e
.
Finally,
w
e
need to
under
s
tand the
con
se
quences
of generativity (and
s
tagnation) for other domains of life, making clear
distinctions among generativity,
well
-
being, and personality.
We have contributed to a way of thinking about generativity that
acknowledges
ex
pression in multiple
role
s
.
Our aim was
t
o
explore
an approach that allows us to
b
e
tt
e
r
und
e
rstand individual
diff
e
renc
e
s
across time and context. In
s
o doing, we hope to come
close
r to doing
justice to the full complexity of the lives of both women and
men
.
ACKNOWL
E
DGMENT
S
In addition to funding from the MacArthur Network for Research
on Successful Midlife Development, support for this research was re-
ceived from the Purdue Agricultural Experiment Station and a
Purdu
e
Univer
s
ity
Biom
e
dical
Res
ea
rch Support Gr
a
nt
(5111187
-
0700). We
gratefully acknowledge support and assistance from Rosalind Barnett,
Ph arne
Cam
a
rena, Ann
C.
C
router, Cynthia
Darling
-
Fish
e
r, Tim
Dwyer, Joanne Gold, Joan Jurich, Eric McCollum,
E
lizabeth Paul,
Carol
Ryff,
D
e
na Targ, the
sta
ff of the Murray Center, and the
respon
-
d
e
nts whose
effo
r
t
s
made this
r
e
search possibl
e
.
We also appreciate
h
e
lpful suggestions from
thre
e
anonymous reviewers.
2
36

E
m
p
loy
ed
Mo
t
h
e
r
s
in
Multi
p
l
e
R
o
les
NOT
ES
1.
B
eca
u
se
ind
iv
idu
a
l
it
e
m
sc
or
es
w
ere not
avai
l
a
bl
e
in the
B
a
ru
c
h
an
d
B
arn
e
tt
d
a
t
a a
r
c
hi
ve,
Cro
nb
ac
h
's
a
lph
as
are
ca
lcul
ated
u
s
in
g
o
nl
y
th
e
1
991
d
ata
se
t.
2.
S
ee
not
e
l
a
bo
ve
.
3.
See
n
o
t
e
l
a
bo
ve
.
4.
All
pat
h
a
n
a
ly
ses
w
ere run
w
i
t
h
a
nd
w
ithout
th
e
wome
n
w
h
o
h
ad
c
hil
dren
u
n
d
er seven
o
r
w
h
o
h
a
d
l
eft
h
ome;
fin
d
i
n
gs
we
r
e
virt
ua
ll
y
id
en
t
ica
l.
5.
T
h
ere
w
as
one
d
i
ffere
n
ce
i
n
th
e
a
n
s
w
er
o
pt
i
o
n
s
offere
d
sp
ou
ses
in
1
978-
1
979
a
nd
1
99
1:
i
n
1
978
-
1
97
9
spou
ses
were
g
i
ven
th
e
op
tion of
res
p
on
d
i
n
g
"
un
-
certain"; in
1
991 spouses were not
give
n this op
tion
.
T
h
e
1
6
respo
nd
ent
s
w
ho
respon
d
e
d
"
un
certai
n
"
i
n
1
978-
19
79 were
ex
clud
e
d
fr
o
m
th
e
co
mp
ar
i
s
on
o
f
r
e-
sponses on
t
hi
s
item
.
Pat
h ana
l
yses
w
e
re
d
one
se
p
arat
e
l
y
f
or
full
-
t
i
me
a
nd
par
t
-
tim
e
wo
rkers in
eac
h
s
u
bsa
mp
I
e
.
In
th
e
1
978
-
1979
s
ubs
a
m
p
I
e
,
fin
d
i
n
gs
w
ere
vi
r
tu
-
a
ll
y
ident
ica
l
f
or
fu
ll
-
time
an
d
part
-
time workers.
Al
thou
g
h
th
ere
w
ere not eno
u
g
h
p
art
-
time
w
o
rk
ers in
th
e
1
99
1
s
ub
sa
mp
I
e
for
re
li
a
bl
e
pa
th ana
l
yses
,
the
p
atterns
for
th
e
full
-
ti
m
e
wo
rk
ers were
s
imi
l
ar
t
o
th
ose in
th
e
o
vera
ll
sa
mpl
e
.
R
EFE
R
ENCES
Arc
h
er
,
Sa
ll
y
L.
(1992)
.
A
f
em
i
nis
t'
s
appr
o
ac
h to
id
e
nt
i
ty
resea
r
c
h
.
In
G.
R.
A
d
a
m
s
,
T.
P
.
G
ull
otta
,
&
R
.
Mon
t
emayor
(E
d
s
.
),
Ado
l
esc
e
nt
i
d
e
ntity
f
ormati
o
n
.
Ne
wbu
ry
Park,
CA:
Sage
.
B
ake
m
a
n,
R
o
ger
,
&
G
o
tt
m
a
n
,
J
o
hn
M
.
(1
986).
O
bs
e
rving
interaction:
An
intr
o
d
u
c-
tion
to
s
e
q
u
entia
l
ana
l
ys
i
s
.
Ca
mb
rid
ge
:
Ca
mbridge Uni
ve
r
s
i
ty Press.
Ba
rn
e
tt
,
Rosa
lin
d C.
(1
991)
.
Mu
l
tip
l
e
r
o
l
e
s,
gende
r
,
and
p
syc
ho
l
o
gica
l
distr
e
ss.
W
e
ll
es
-
l
ey
Co
ll
ege
W
orking
Pa
p
er
n
o
.
233.
W
e
ll
es
l
ey
,
M
A:
W
e
ll
es
l
ey
C
oll
ege
Ce
nt
er
for
R
ese
a
rc
h
o
n
Wome
n
.
Barnett
,
Rosa
l
i
nd
c.,
&
Ba
ruc
h
,
Grace
K
(1
978
)
.
W
ome
n
i
n
t
h
e
m
iddl
e
years:
A
cri
tiq
ue of
resea
r
c
h
an
d
th
eory
.
P
sych
o
l
ogy
o
f
W
omen
Quart
e
rly,
3(2),
1
87-197.
B
arne
tt,
Ro
sa
li
nd
C.
,
&
Ba
ru
c
h, Grace
K.
(198
5
).
W
ome
n
's
invol
ve
m
e
nt in
mult
i
pl
e
ro
l
es
and
p
syc
h
o
logica
l
di
stress
.
J
ourna
l
of
Persona
l
ity and
Socia
l
Psycho
l
ogy,
4
9
,
1
35-
1
45.
Barnett,
Ros
a
lin
d
C
.,
&
Ba
ru
c
h,
G
r
ace
K
(
1
987)
.
S
o
cia
l
ro
l
es
,
gen
d
er,
a
n
d
p
syc
ho
-
l
ogica
l
distress
.
In
R.
C.
B
arne
tt
,
L.
Biener
,
&
G.
K
B
a
ru
c
h
(
E
d
s.
),
Gen
d
e
r
and
Stress
.
New York: Free Press.
Barne
tt,
R
osa
li
nd
C.
,
Mars
h
a
ll
,
Nan
cy
L.
,
&
Sayer
,
A
l
ine
.
(1
99
1). P
o
sit
iv
e
spill
over
e
ff
e
cts
from job
to
hom
e
:
A
clos
e
r
l
ook
.
We
ll
es
l
ey
Co
ll
ege Working
Pa
p
er no
.
222
.
W
e
ll
es
l
ey,
M
A:
W
e
ll
es
l
ey
Co
ll
ege
Ce
nt
e
r for
R
esearc
h on
W
omen.
B
aruch, Grace
K
,
&
B
a
rn
e
tt
,
R
osa
l
in
d. (1986).
R
o
l
e
qu
a
li
ty
,
mult
i
pl
e
r
ol
e
i
n
vo
l
ve-
me
nt
,
an
d
p
sycho
lo
gica
l
we
ll
-
b
eing
i
n mi
dli
fe
women
.
J
ourna
l
of
Pers
o
na
l
ity
and
Socia
l
Psycho
lo
gy,
5
1
(3),
578
-
585
.
B
a
ru
c
h, Grace
K
,
Barnett
,
R
o
sa
lind
c.,
&
R
i
ve
r
s
,
Cary
l.
(1
983).
Life
p
rints
.
Ne
w
York:
P
lu
me
Books
.
23
7

M
ac
D
ermi
d,
H
e
ilbrun, and
D
e
Ha
an
Bur
ea
u of the Censu
s
.
(1980)
.
Stati
s
tical
Ab
s
tra
c
t
of
t
h
e
Unit
e
d
Stat
es
,
101
s
t
Ed
it
i
on
.
U.S.
Dep
a
rtment of Commerce.
Bur
e
au of the
Ce
nsu
s
.
(1987).
Stati
s
ti
c
al
Ab
s
tr
a
ct
of
th
e
Unit
e
d
St
a
t
es
,
107th E
d
i
t
i
o
n
.
U.S
.
D
e
p
a
rtm
e
nt of
Comm
e
rce
.
Bur
e
au of
th
e
Ce
nsus
.
(1991)
.
Stati
s
ti
c
al
Ab
s
tract
of
th
e
Uni
te
d
Stat
es
,
111
th
Ed
i
t
ion
.
E
conomics and
St
a
t
i
stics Administration.
Bure
a
u of
L
a
bo
r
Statistics
.
(19
7
9,
J
a
nu
a
ry)
.
C
PI
D
e
tail
e
d
R
e
port.
U
.
S.
D
e
p
a
rtment
of
L
a
bor.
Bureau of Labor
St
a
tisti
c
s
.
(1990
,
Decemb
e
r)
.
C
PI
D
e
tai
l
e
d R
e
po
r
t.
U
.
S.
Departm
e
nt
of Labor.
Bur
e
au of Labor
Stati
s
tic
s
.
(1991,
Augu
s
t).
Workin
g
wo
me
n
:
A
c
ha
rtbook. U.S.
De
-
p
ar
tm
e
nt of
L
a
bor.
Campb
e
ll,
A
.,
Conv
e
rs
e
,
P.,
&
Ro
g
e
rs
,
W
.
L.
(1976)
.
T
h
e
quality
of
Am
e
ri
c
an
l
i
f
e.
Ne
w
Y
ork:
Ru
ss
e
ll
S
age
Found
a
tion.
C
oh
e
n,
J
ac
ob
.
(1960).
A c
o
e
ffici
e
nt of
a
g
re
em
e
nt for
nom
i
n
a
l
sc
a
l
es.
Educat
ional
a
nd Psychological
M
eas
ur
eme
nt
,
2
0,
3
7
-
46.
D
a
rlin
g-
Fisher,
c.
,
&
L
e
idy,
D.
Klin
e
.
(1988)
.
M
e
a
s
urin
g
Er
iksoni
a
n development
in
th
e
a
dult:
Th
e
modified
E
r
i
kson
p
s
ychosoci
a
l
st
age inven
t
o
r
y,
Ps
y
c
holo
gi
cal
R
e
ports,
6
2
,
7
47
-
754
.
Dero
ga
tis,
L.
R.,
Lipm
a
n
,
R
.
S
.
,
Rick
e
ls,
K.,
Uhlenhuth,
R.
H.,
&
Covi,
L.
(1974)
.
The Hopkins Symptom
C
heckl
is
t:
A
self
-
r
e
port symptom
inv
e
nto
ry
.
B
e
h
a
vioral
S
c
i
e
n
ce,
19,
1
-
14.
Erikson,
E
rik H. (1950). Childhood and
Soci
e
ty
.
New York: Norton.
E
rikson,
E
rik H., Erikson, Ioan
.
M,
&
Kivnick,
H
e
len
Q.
(1986).
V
ita
l
involv
e
m
e
nt
i
n old
ag
e.
New York:
W
.
W.
Nor
ton.
Fisk
e
,
M
a
r
j
ori
e.
(1980)
.
C
hangin
g
hi
e
r
a
rchies of
commitm
e
nt in
a
du
l
thood
.
In
N
e
il
J
.
Sm
e
ls
e
r
&
E
r
ik H.
Erik
s
on
(Ed
s
.
),
Th
e
m
e
s
of
work and
lov
e
in
a
du
lthood (pp.
238
-
264).
C
ambridge, MA:
H
a
rva
r
d University
Press
.
Fran
z
,
Car
ol
E
.,
&
White
,
K
a
thl
ee
n
M
.
(1985).
Individu
a
tion and
att
ac
hment in
per
sona
l
ity
d
e
velopment
:
Extending
Erik
s
on's
theory
.
Journal
of
P
er
sonality,
53
,
2
24
-
2
5
5.
Gergen
,
M
a
ry M. (1990)
.
Finished at 40: Women's
d
e
ve
l
opment
w
i
thi
n the
p
a
tri
a
r
-
chy
.
Psy
c
ho
l
o
g
y of
Wom
e
n
Quart
e
rly,
1
4
,4
7
1
-
49
3
.
Gie
l
e,
Janet Zollinger.
(1982)
.
Wom
e
n in
a
dulthood
:
Un
a
nswered
questions
.
In
Janet
Z
olling
e
r Giele
(
E
d
.
),
Wom
e
n in
th
e
midd
l
e
y
e
ars (pp.
1
-35
).
New York:
John
Wil
e
y and
Sons
.
Gillig
a
n,
C.
(1982)
.
Adult development and
wom
e
n's
d
e
velopment
:
Arr
a
ngem
e
nts
fo
r a
m
a
r
r
iag
e.
In Janet
Z
ollinge
r
G
iel
e
(
E
d.),
Wom
en
i
n
th
e
m
id
d
l
e
y
e
ars (pp.
89
-
114). New York: John Wiley and
Sons
.
Gorsuch
,
Rich
a
rd
L.
(1983). Factor
Analys
i
s.
Hillsdal
e
,
NJ:
Law
re
nce
E
rlb
a
um
.
Grot
e
va
nt
,
H. D. (1987). Toward a process
mod
e
l
of
id
e
ntity
form
a
tion. Journal
of
Adol
e
sc
e
n
t
R
e
s
e
arch,
2
(
3
),
203
-
222.
Hornstein
,
Ga
il
A.
(1986)
.
The
structu
rin
g
of
id
en
tity
a
mon
g
m
i
dlif
e
women
a
s
a
238

Employ
e
d
Moth
e
rs
in
Multipl
e
Rol
e
s
function of
th
e
ir
degr
ee
of
involv
e
ment in employment.
Journal of
P
e
rsonality,
5
4,
551
-
5
75.
Hul
s
izer, D., Murphy,
M
.
,
Noam,
G
.
,
Taylor,
c.,
Erikson,
E.,
&
Erikson,
J.
(1981)
.
On
g
ene
r
at
i
vity and identity: From
a
conv
e
rs
a
tion with Erik and
Jo
a
n
Erikson
.
Ha
r
vard
E
ducational
R
e
vi
e
w,
51,
249
-
269.
Io
re
sko
g
,
K
a
rl
G
.
(1979).
A
gen
e
ral
a
pproach to confirmatory
m
ax
imum likelihood
factor
analysi
s
with
a
dd
e
ndum. In Karl
G
.
Io
resko
g
&
Dag Sorbom,
Adva
n
ce
s
i
n
facto
r
analysi
s
an
d
structural
e
quation
m
od
e
ls
(pp.
21
-
44)
.
Cambridge, MA:
Abt
Book
s.
Ioresko
g
,
Karl
G.,
&
Sorbom,
Da
g.
(1988).
LISREL
7-
A
Guid
e
to
th
e
Pro
g
ra
m and
Application
s
.
Chicago:
SPSS.
Juhas
z,
Anne
McCre
ar
y.
(1989)
.
A
role
-
bas
e
d approach to adult
de
v
elopm
e
nt:
Th
e
tripl
e-
h
e
lix
mod
e
l.
I
n
t
e
rnational Journ
a
l
of
A
g
ing and Human
D
e
v
e
lopm
e
nt
,
2
9(4),
3
01
-3
15.
Kotr
e
,
John.
(1984)
.
Outlivi
n
g
th
e
se
lf
Gene
rativ
i
ty and
th
e
i
nt
er
pr
e
tation of
liv
es
.
Baltimore,
MD
:
Johns Hopkins University
Press
.
Kroge
r
,
Jane,
&
H
as
lett,
Steph
e
n
J.
(1991). A comparison of ego identity status
tr
a
nsition
pathw
a
ys
and
chan
ge
rat
e
s
a
cross
five
identity domains.
Int
e
rnational
Journal of
A
g
ing
and Human D
e
v
e
lopm
e
nt,
3
2
(4),
303
-
330.
Lon
g
,
Judy
,
&
Port
e
r,
Kar
e
n
L.
(1984). Multiple roles of midlife women: A case
fo
r new directions in theory,
r
e
se
a
rch, and
poli
c
y.
In
G
.
B
a
ruch
&
J
.
Brooks-
G
unn
(
E
ds.),
Wom
en
i
n
m
i
dlif
e
.
New York:
Plenum
.
M
a
cDermid,
Shell
e
y M.,
&
Gillespie, Laura
K.
(1992,
May)
.
G
e
n
e
rativity
in
multipl
e
rol
e
s
.
P
a
per presented
durin
g
symposium
Eme
r
g
i
n
g
p
e
rsp
e
ctiv
es
in adult
p
e
r
s
on
-
a
lity
d
e
v
e
lop
me
nt
:
R
e
s
e
arch
co
n
ce
rning
g
e
n
e
rativity,
ch
a
ired by
E.
de
St.
Aubin
at the annual
m
ee
tin
g
of the Midwestern
Psychologic
a
l
Association,
Chicago
.
McAd
a
ms,
D
a
n
P.
(1988).
Pow
e
r,
intimacy and
th
e
li
f
e
story.
New York: Guilford
.
McAdam
s
,
D
a
n P.,
Ru
e
t
z
el,
Karin,
&
Foley,
J
ea
nn
e
M. (1986).
Comple
x
ity and
gene
ra
tivity at midlife:
Rel
a
tions
a
mong social motives, ego
d
e
velopment,
a
nd
adults'
plan
s
fo
r the future.
Journal of
P
e
rsonality and
Social
Psychology,
50(4),
800
-
807.
McAdams, Dan
P
.
,
&
St.
Aub
i
n,
E
d
d
e
.
(199
2
)
.
A theory of
gen
e
rativity and its
a
ssessment
throu
g
h
self
-
report,
beh
a
vior
a
l
a
cts, and
n
a
r
r
ative
th
e
mes in autobi-
ography
.
Journal of
P
e
r
s
onality and
Social
Psycholo
g
y,
62
(6), 1003-1015.
McAdams,
D
a
n P.,
St.
Aubin,
E
d
d
e
,
&
Lo
ga
n,
R
.
L.
(1993)
.
Ge
n
e
rati
v
ity
amon
g
young,
midlif
e
,
a
nd older adults
.
Psycholo
g
y and
A
g
in
g
,
8(2),
221
-
230.
McKinlay
,
Sonj
a
M.,
Tri
a
nt,
R
a
ndi
S
.
,
McKinl
a
y,
John
B
.
,
Brambilla, Donald
J
.
,
&
Ferdock,
Matth
e
w
.
(1990).
Multipl
e
roles
fo
r
middl
e-
a
ge
d
wom
e
n and
th
e
ir
imp
a
ct on
he
a
lth. In M. Ory
&
H
.
Warne
r (Eds.),
G
e
nd
e
r,
h
e
alth and
long
e
vity
.
New York: Sprin
ge
r.
Neug
a
rt
e
n
,
B
e
rnice
L.
(1968). The
awar
e
ness
o
f middle
ag
e
.
In
B.
L.
Neugarten
(Ed
.
),
Middle
a
ge
a
nd
a
g
ing (pp.
93
-
98).
Chic
ago
:
University of Chicago
Press
.
Ochs
e
,
R
.,
&
Plu
g
,
C.
(1986)
.
Cross
-
cultur
a
l
investigation of the
v
a
lidity of Erikson's
2
39

Mac
D
e
rmid,
H
e
il
brun, and
D
e
Haan
t
h
e
o
ry of
p
ersona
li
ty
d
e
v
e
l
opment.
J
ourna
l
of
P
e
rsona
l
ity and
S
o
cia
l
Psyc
h
o
l
ogy,
5
0
,
1
24
0
-
1
252.
Pearlin,
L.
K.,
&
Sc
hoo
l
er,
C.
(1978)
.
T
h
e
structure of
cop
i
ng.
J
our
n
al of
H
e
alt
h
and
Soci
al
B
e
h
avior
, 1
9,
2
-
21.
P
e
t
erson
,
B
ill
E.,
&
S
t
ewart
,
A
bigai
l
J.
(199
0
)
.
Us
i
n
g
p
e
rs
o
n
al
an
d
fictio
n
a
l
docu
-
ments to assess
psychos
o
cia
l
deve
l
opment: A
cas
e
s
t
ud
y of Vera
Brittain
'
s
Ge
n
-
era
tiv
i
ty
.
Psyc
h
o
lo
gy and
Aging,
5,
400
-
4
11.
Pe
t
erso
n
,
B
ill
E
.
,
&
Stewart,
A
b
igai
l
J
.
(1992,
May)
.
T
h
e a
ssessment of gener
a
tiv
i
ty
u
s
i
ng
th
e
Ca
li
fornia
Q
-S
or
t.
Pap
e
r
p
resen
t
e
d during the symposi
u
m
Em
e
rg
i
n
g
p
e
rsp
e
ctives
in
adult
p
e
rsona
l
ity
d
e
v
e
lopm
e
nt:
R
ese
arc
h
concernin
g
g
e
n
e
rativity
,
c
h
aire
d
b
y
E.
d
e
St.
A
ub
in
a
t
th
e
a
n
nu
a
l
m
eeting of
t
h
e
Midw
estern
Psyc
h
o
lo
g
i
-
ca
l
Ass
o
cia
t
ion,
C
h
icago.
R
osen
b
erg,
M
.
(1979).
Conceiving
t
h
e
self.
New York: Basic
B
o
oks
.
R
u
pp,
M
i
c
h
ae
l
T.,
&
Sega
l
,
Richard
.
(1989)
.
Confirm
a
tory factor
a
n
a
l
ysis of
a
pro
-
fess
ion
a
li
sm
sc
a
l
e
in
p
har
m
acy.
J
ournal of
Socia
l
and
Administra
ti
v
e
Pharm
a
cy,
6(1),
3
1
-
38.
R
yff
,
Ca
ro
l
D
.,
&
Hein
cke,
S
.
G.
(1
983)
.
S
ub
jective
organiza
t
i
o
n of
persona
l
ity in
a
dulthoo
d and aging.
J
ournal of
P
e
rsona
l
ity and
Soci
a
l
P
s
ycho
lo
g
y,
44
,
80
7-
8
1
6
.
R
yff
,
Caro
l
D.
,
&
Mig
d
a
l
,
S
u
san
.
(1984).
Int
imacy and generativity:
Self-perce
i
ve
d
trans
i
t
i
ons.
S
i
g
n
s
,
9(3),
47
0
-
481.
S
h
erma
n
,
E
d
m
und
.
(1
987)
.
Meanin
g
in
mid
-
l
if
e
transitions
.
Al
b
any:
Stat
e
Univ
e
rs
i
ty
of
N
e
w
Y
o
rk
Press
.
Stewar
t
,
A
bi
ga
il
J
.,
Franz, Caro
l
E.
,
&
Layto
n
,
L.
(1988).
Th
e
chang
i
ng
self:
Using
persona
l
d
oc
u
men
t
s
to
s
tud
y
li
ves
.
J
ourna
l
of
P
e
r
s
ona
l
ity,
56, 41
-
74
.
S
tewar
t,
A
bi
gai
l
J
.,
Fra
n
z
,
Ca
rol
E
.
,
Paul
,
E
li
za
b
et
h
,
L.
,
&
Pe
t
e
r
son
,
B
.
E.
(199
1
).
R
e
vise
d
co
d
i
n
g
manua
l
for
t
h
r
ee
asp
ects
of
dev
e
l
opment of
t
h
e
adu
l
t
se
lf:
I
d
e
ntity,
intimacy
,
a
nd
g
e
nerativity.
Unpubli
s
h
e
d
c
odin
g
man
u
a
l.
Ann
A
rbo
r
:
Unive
r
sity
of Michi
gan
.
S
t
e
w
art
,
A
bi
ga
il
J
.,
&
Go
ld
-
S
t
e
i
n
b
erg
,
S
h
ar
on
.
(199
0
).
M
idlif
e
women's
poli
tica
l
co
n
scio
u
s
n
es
s
:
Case
st
udi
es
of
psyc
h
osoc
i
a
l
d
eve
l
opmen
t
a
nd
po
li
tica
l
co
m
mit-
me
nt.
P
s
y
c
h
o
l
ogy
of
W
omen
Quart
e
rl
y
,
14,
543
-
566
.
Ta
n
ak
a
,
J.
S
.
&
H
ub
a, G.
J
.
(198
4)
.
Confirm
a
t
ory
hi
erarc
hi
ca
l
fac
t
or
ana
l
yses
o
f
p
syc
holo
gica
l
di
s
t
ress
measures
.
J
ourna
l
o
f
P
ersona
l
ity and
Socia
l
Psyc
h
o
l
ogy,
46,
62
1
-635
.
Va
ill
ant
,
Geo
r
ge
E
.
,
&
M
il
ofsky,
E
v
a
.
(198
0
).
Na
tu
ra
l
hi
story
o
f
ma
l
e
psyc
h
o
l
og
i
ca
l
hea
lth
I
X
:
Empirica
l
evi
d
ence for
Erikson'
s
mo
d
e
l
of
th
e
l
ife
c
ycle
.
Am
e
rican
J
ourna
l
o
f
Psyc
h
iatry
,
137(11)
,
1348
-
1359
.
Va
n
d
e
W
a
t
er
,
Donna
A.,
&
M
cA
d
ams, Dan
P
.
(1989)
.
Ge
nerati
v
i
ty
an
d
Er
i
ks
on
's
"Belief in
t
h
e
Species
.
"
J
ourna
l
of
R
e
s
ea
rch
in
P
e
rsona
l
ity,
2
3,
4
35
-
449
.
W
a
t
erma
n
,
A
.
S.
(1985).
Id
enti
ty in the
co
nt
ex
t of
ado
l
esce
nt
psyc
h
o
l
ogy. In Alan
S
.
W
a
t
erma
n
(E
d.
)
,
I
d
e
ntity in
ado
l
e
sc
e
nce:
Process
e
s
and
c
o
n
t
ex
t.
S
a
n
Fra
n
c
i
sco
:
J
osse
y
-
Bass
.
24
0