Time Counts: Future Time Perspective, Goals, and Social Relationships
Frieder R. Lang
Humboldt University of Berlin
Laura L. Carstensen
On the basis of postulates derived from socioemotional selectivity theory, the authors explored the extent
to which future time perspective (FTP) is related to social motivation, and to the composition and
perceived quality of personal networks. Four hundred eighty German participants with ages ranging
from 20 to 90 years took part in the study. In 2 card-sort tasks, participants indicated their partner
preference and goal priority. Participants also completed questionnaires on personal networks and social
satisfaction. Older people, as a group, perceived their future time as more limited than younger people.
Individuals who perceived future time as being limited prioritized emotionally meaningful goals (e.g.,
generativity, emotion regulation), whereas individuals who perceived their futures as open-ended
prioritized instrumental or knowledge-related goals. Priority of goal domains was found to be differently
associated with the size, composition, and perceived quality of personal networks depending on FTP.
Prioritizing emotion-regulatory goals was associated with greater social satisfaction and less perceived
strain with others when participants perceived their future as limited. Findings underscore the importance
of FTP in the self-regulation of social relationships and the subjective experience associated with them.
Social relationships are a powerful resource for successful ag-
ing. There is considerable evidence that the quantity and, even
more so, the quality of social relationships are reliably associated
with better physical and psychological functioning (e.g., Fratigli-
oni, Wang, Ericsson, Maytan, & Winblad, 2000; Uchino, Ca-
cioppo, & Kiecolt-Glaser, 1996). In view of this, life span scholars
have emphasized the proactive role of the individual in managing
the social and psychological resources that contribute to successful
development across adulthood (e.g., M. M. Baltes & Carstensen,
1996; Carstensen, Graff, & Lang, 2000). It has been argued that
through the selection of goals and environments, people influence
the course and quality of their individual lives (see M. M. Baltes
& Carstensen, 1996; Bandura, 1997; Caspi, Bem, & Elder, 1989).
However, not much is known about the ways in which an individ-
ual’s goals or motivations are actually reflected and regulated in
relationships with others (e.g., Ferring & Filipp, 1999; Hill, 1991),
and whether a congruence between goals and social relationships
contributes to improved social functioning (e.g., Jarvinen & Nich-
olls, 1996). This issue is particularly relevant in later adulthood
because personal networks do tend to reduce in size in old age,
with emotionally close social partners being maintained while
more peripheral social partners are increasingly excluded (Lang,
2000; Lang & Carstensen, 1994; Lang, Staudinger, & Carstensen,
1998). The goal of this study is to explore the possible matches
between social motivation and the composition and quality of
personal networks. More specifically, we examine whether the
associations of social motivations on the composition and on the
perceived quality of personal networks differ depending on age-
related differences in future time perspective (FTP).
The present research advances and extends assumptions of
socioemotional selectivity theory (Carstensen, 1993; Carstensen,
Isaacowitz, & Charles, 1999), according to which an individual’s
perceptions of his or her remaining time to live determines the
priority of specific goal contents. Because chronological age is
systematically related to time left to live, age-related patterns of
prioritized goal contents emerge.
Socioemotional Selectivity Theory:
FTP, Chronological Age, and Goals
Essentially, socioemotional selectivity theory predicts that indi-
viduals select goals in accordance with their perceptions of the
future as being limited or open-ended, and that such time-
perspective congruent selection is adaptive. When time is per-
ceived as expansive, goals aimed at optimizing the future are
prioritized. Such goals often pertain to the acquisition of knowl-
edge or to seeking contacts that could be useful in the more distant
future. This also includes goals related to the task of finding out
about one’s role in the society (e.g., receiving social acceptance),
and to vocational or career interests (e.g., become financially
Frieder R. Lang, Department of Education, Humboldt University of
Berlin, Berlin, Germany; Laura L. Carstensen, Department of Psychology,
The present research was supported by German Research Foundation
Grant Ba 902/11-2 awarded to Margret M. Baltes and Frieder R. Lang.
Data collection for this study took place while Frieder R. Lang was at the
Free University, Berlin, Germany. We are most grateful to Margret M.
Baltes, who secured most of the funding for this research and who pro-
foundly enriched this work with her ideas. Frieder R. Lang also expresses
his deep gratitude for her influence as a mentor. We are thankful to the
many colleagues and students who have made this work possible. We are
also indebted to Jens B. Asendorpf and Jutta Heckhausen for their most
valuable comments on an earlier version of this article and to Iain Glen for
Correspondence concerning this article should be addressed to Frieder
R. Lang, Humboldt University of Berlin, Department of Education (Faculty
of Arts IV), Geschwister-Scholl-Strasse 7, 10099 Berlin, Germany. E-mail:
Psychology and Aging
2002, Vol. 17, No. 1, 125–139
Copyright 2002 by the American Psychological Association, Inc.
In contrast, when time is perceived as limited, emotionally
meaningful goals become relatively more important because they
are typically associated with achieving short-term benefits. In our
research, we distinguish two subtypes of emotionally meaningful
goals—one relating to the regulation of emotions and one relating
to generativity goals. Emotion regulation pertains to self-
regulatory goals such as seeking to be in control over one’s
emotions or seeking meaningful emotional experiences. Generat-
ivity includes goals such as being or becoming a “keeper of the
meaning” (Vaillant & Milofsky, 1980, p. 1349) as well as taking
responsibility for future generations (e.g., Ryff & Heincke, 1983).
Generativity goals have been found to be most prominent in later
adulthood (McAdams, Hart, & Maruna, 1998). For example, in an
experimental study using a game-simulation method, Van Lange,
De Bruin, Otten, and Joireman (1997) found stronger prosocial
orientations among older adults as compared with young adults.
McAdams et al. (1998) suggested that an increased commitment
towards generativity in later life may reflect an individual’s desire
for “symbolic immortality.” Consequently, striving for generativ-
ity when individuals perceive their time as running out may also
indicate an (emotion-regulatory) effort to symbolically expand
According to socioemotional selectivity theory, age differ-
ences in goal selection are linked to the individual’s perception
of time. When younger people imagine conditions in which
their time is limited (e.g., anticipating a geographic relocation),
or when older people imagine conditions in which their time is
unlimited (e.g., a life-extending medical advance), age differ-
ences in goal selection disappear. For example, when the imag-
ined time frame is limited, younger people have been found to
prefer familiar social partners (Fredrickson & Carstensen,
1990), and when the imagined time frame is extended, the bias
displayed by older people had disappeared (Fung, Carstensen,
& Lutz, 1999). In a more naturalistic study aimed at decoupling
chronological age from position in the life cycle, a sample of
relatively young men who found themselves at different points
in the life cycle (based on HIV status) was studied. Consistent
with experimental studies, those research participants who were
closest to the end of their lives displayed similar goals to very
old adults (Carstensen & Fredrickson, 1998). Emotionally
meaningful goals therefore appear to be prioritized as people,
old or young, approach the end of their lives.
So far, however, no empirical study has investigated the gener-
ality of socioemotional selectivity theory across a broad and di-
verse set of goals, on the basis of a continuous measure for FTP,
and on the basis of a heterogeneous sample of mostly healthy
adults covering a broad age range from early to late adulthood. In
addition, to date, most of the empirical research stemming from
socioemotional selectivity theory has focused on goal selection
rather than on the adaptive outcomes associated with the selection
of particular social goals. It is thus still not well understood in
which ways the selection of specific goal contents also benefits
social functioning and the perceived quality of social relationships.
For example, one expectation is that when individuals set goal
priorities that correspond with their perceptions of the future, time
as either limited or expansive, they benefit in terms of perceived
quality of relationships.
Matching Goals and FTP to Personal Networks
In a series of studies, Lang and his colleagues (Lang, 2000,
2001; Lang & Carstensen, 1994; Lang et al., 1998) have reported
findings about the associations between FTP and social relation-
ships in later life. For example, in a study with old and very old
adults, Lang (2000) found that close emotional relationships were
more stable than peripheral social relationships over a 4-year time
interval. Most social relationships that had been discontinued were
terminated deliberately and did not end because of morbidity or
mortality. Most important, reductions in the number of social
relationships were more often observed when individuals felt that
their life was coming to an end. Across the 4-year period, people
who felt they were near the end of their lives increased emotional
closeness with family members and with social companions more
strongly than with those who gave tangible support. In this re-
search, individual goals were not measured and it cannot therefore
be concluded from these findings that the effects of subjective
nearness to death on personal networks and emotional closeness
with social partners also reflect differences in goal priority. More-
over, the findings did not differentiate among types of received
support (e.g., advice or help with errands) and between positive
and negative aspects of social relationships, which are found to
constitute separate and distinct dimensions of social experience
(e.g., Okun & Keith, 1998; Rook, 1984).
Although it makes logical sense that personal networks are
actively regulated in accordance with goals, alternative arguments
for the benefits of having large, diversified, and heterogeneous
personal networks in later life are conceivable. Restrictions in
social contact may reduce the likelihood for incidental or certain
casual social experiences. Indeed, gains and losses are inevitably
associated with any selected pursuit (P. B. Baltes, 1987). Yet there
is also empirical evidence that goal-motive congruence may be
especially adaptive. Brunstein, Schultheiss, and Grassman (1998)
found that only the achievement of goals that are congruent with
motivational dispositions contributes to enhanced well-being. For
example, when an individual is committed to intimacy-related
goals while having a strong power or achievement motive dispo-
sition, progress towards intimacy goals will not necessarily en-
hance his or her well-being. It seems that the pursuit of some goals
can even be dysfunctional when incongruent with the individual’s
motive or context.
Similarly, when goals are incongruent with one’s FTP, pursuing
them could result in detrimental outcomes. According to socio-
emotional selectivity theory, such incongruency may relate to
either (a) seeking instrumental or knowledge-related goals when
the future is perceived as limited or (b) pursuing emotionally
meaningful goals when the future is perceived as open-ended. One
possible consequence may be that individuals are more likely to
experience social strain and dissatisfaction. For example, when
pursuing instrumental goals in one’s social worlds (e.g., seeking to
learn new skills from others) while having a sense that one’s future
time is limited, individuals may more often feel impatient, disap-
pointed, or irritated with others who are unable to meet their
expectations. In contrast, when individuals prioritize emotionally
meaningful goals although experiencing their future time as ex-
pansive, individuals may not invest much into knowledge-
enhancing social contacts. However, these individuals are also
likely to be confronted with questions about their long-term future
LANG AND CARSTENSEN
plans by their social partners who want them to do well. Avoiding
knowledge-related and informational exchanges with these social
partners may thus result in a lowered quality of relationships.
But how about when FTP is in accordance with one’s preferred
goals? For example, when the future is perceived as relatively
open-ended and filled with opportunities, people are expected to
engage in activities and plans that extend into the future, even
when this bears risks (e.g., de Volder & Lens, 1982). In fact,
perceiving one’s future as expansive may be an important require-
ment for young and middle-aged adults who are confronted with
family and career goals. Prioritizing goals in a timely manner may
also be associated with more positive social experiences. When the
future appears restricted in time, people may seek to use their
remaining time for plans and social experiences that appear most
emotionally meaningful to them, and thus selectively invest in
those relationships that appear to match with these expectations.
Again, doing so is likely to result in greater social satisfaction. To
sum up, we expect that the perceived quality as well as the
quantitative aspects of personal networks (size, composition) will
be differently associated with goals, depending on whether goals
match with FTP.
The Present Research and Major Hypotheses
In the present research, we addressed two questions. First, we
asked whether individuals preferably select emotion-regulatory
goals if they perceive their future time as limited. This research
question aimed to replicate previous findings across a broad set of
goals in an heterogeneous sample of young, middle-aged, and
older German adults. Second, we expected that when individuals
perceive their future as limited, prioritizing emotion-regulatory
goals will determine the composition and quality of personal
networks more strongly than when they perceive their future as
open-ended. We hypothesized that emotionally meaningful goals
are prioritized when individuals perceive their future time as
limited. In contrast, when future time is perceived as open-ended,
we expected a preference for knowledge-related and instrumental
social goals. Finally, we hypothesized that goals are differently
associated with the size and quality of personal networks depend-
ing on FTP. In specific, we expected that when one’s future is
perceived as limited, emotionally meaningful goals are associated
with smaller networks, and with greater social satisfaction. We
also expected that knowledge-related goals will be most strongly
reflected in the size and composition of personal networks when
individuals perceive their future as open-ended.
In total, 480 adults took part in a one-session (or if necessary, two-
session) interview that included physical, psychological, and sociological
assessment instruments. Participants were recruited through probability
sampling from the local registration office in Berlin in 1997 (in Germany,
each citizen must be registered), stratified by year of age, cohort, and
gender. In each birth year, 4 men and 4 women agreed to participate. The
480 participants represented 31.3% of a pool of 1,531 persons who were
contacted prior to the study. Of these 1,531 persons, 1,022 (66.8%) took
part in a short telephone interview at the occasion of the first contact. An
analysis of sample selectivity revealed small effects of sample selectivity
(d ? .25). Participants, as compared with nonparticipants, were better
educated and were more satisfied with their health.
Three age-cohort groups were distinguished. Young adults ranged
from 20 to 40 years (M ? 30.7, SD ? 5.7, N ? 160), middle-aged adults
ranged from 45 to 65 years (M ? 55.7, SD ? 5.8, N ? 160), and old adults
ranged from 70 to 90 years (M ? 80.7, SD ? 5.9, N ? 160). Average years
of education (including professional training) was 13.0 (SD ? 2.9). Of the
480 participants, 48.8% were married, 17.5% were widowed, 9.2% were
divorced, and 24.6% had never been married. A total of 66.5% of partic-
ipants had at least one living child. Sociodemographic characteristics
differed depending on age-cohort group. Older adults were most likely to
be widowed, ?2(2, N ? 480) ? 117.7, p ? .01. In contrast, young adults
reported more years of education, F(2, 477) ? 13.0, p ? .01; were more
likely to be unmarried, ?2(2, N ? 480) ? 125.2, p ? .01; and were less
likely to have children, ?2(2, N ? 480) ? 58.7, p ? .01.
Instruments, Measures, and Constructs
Participants took part in individual interview sessions that were mostly
done in a university building (with a few exceptions, e.g., when partici-
pants could not easily leave their home). Participants received DM 50 ($20)
for participating in the study.
FTP was assessed with a German version of the Future Time Perspective
Scale developed by Carstensen and Lang (1996). Participants rated on a
scale from 1 (very good) to 5 (not at all) the degree to which they agreed
with each of 10 items. Sample items are “Many opportunities await me in
the future,” “Most of my life (still) lies ahead of me,” “I have the sense that
time is running out,” “As I get older, I begin to experience that time is
limited,” and “My future seems infinite to me.” The internal consistency
reached an alpha of .92. The scale was transformed into T scores, with
values above 50 indicating a more open-ended time perspective and values
below 50 indicating a more limited time perspective. To interpret results,
a tercile split on the FTP scores was used, resulting in three groups of
participants with limited (lower third), indefinite (middle third), and open-
ended (upper third) time perspectives.
Personal networks characteristics.
were assessed with the circle-diagram method (Kahn & Antonucci, 1980).
The circle-diagram consists of three concentric circles grouped around a
small circle in which the German word Ich (I) is written. The inner circle
represents network members to whom the participant feels very close, so
close that it would be “hard to imagine life without them.” The middle
circle refers to those network members to whom the participant felt close
but not “quite so close.” The outer circle lists those network members to
whom the participant “feels less close, but who are still important.” After
participants had reported the names of their network partners in each of the
three circles, they reported additional information on each of their network
partners (i.e., gender, type of relationship—spouse, child, other kin, non-
kin—and length of relationship). The personal-network size was computed
as the sum of all social partners named in the circle diagram ranging from 0
to 57 network members (M ? 11.6, SD ? 7.4). Four participants did not
report any network partners. In addition, three scores were used to indicate
the composition of personal networks, that is, the number of (a) friends, (b)
nonkin network partners who have been known for less than 10 years
(novel nonkin partners), and (c) relatives. All three indicators were mod-
erately to strongly correlated with the size of personal networks (number of
friends: r ? .66, p ? .01; number of nonkin known less than 10 years: r ?
.68, p ? .01; number of relatives: r ? .55, p ? .01). In order to focus only
on information unrelated to personal network size, all three measures were
residualized on the size of networks. The resulting residualized (corrected)
scores reflected the relative proportion of the three relationship character-
istics in the personal network. It should be noted that the number of friends
and the number of novel network partners correlated substantially (r ? .55,
p ? .01). The two variables indicate distinct but interdependent character-
istics of the personal network (i.e., friendship, relationship length).
Personal networks of participants
FUTURE TIME PERSPECTIVE, GOALS, AND SOCIAL RELATIONSHIPS
Received advice (number of advice-giving supporters).
reported the names of partners who had provided advice during the past
month. Participants were asked to name those partners to whom they had
turned to when seeking advice during the past 4 weeks. In order to avoid
confusion with network size, the sum of received advice was residualized
on personal network size (r ? .32, p ? .01). Higher values on the resulting
score indicated a greater number of advice-giving social partners in the
Social satisfaction was assessed with two items. Participants rated how
satisfied they were with their social partners in general, and how satisfied
they were with their family and relatives on a rating scale ranging from 1
(very dissatisfied) to 5 (very satisfied). The two measures correlated .41
(p ? .01, ? ? .59). A T-transformed mean composite of the two items was
used, with higher values indicating greater social satisfaction.
Perceived social strain was assessed with a scale that consisted of four
items: annoyance with others, feelings of being overly demanded, feeling
obstructed by others, and unfulfilled need for support. In order to assess
unfulfilled need for support, participants reported in yes–no format whether
they had desired any of three types of support (advice, esteem support,
tenderness) but had not received sufficient amounts. Participants rated how
often they had felt annoyed and overly demanded by others during the
past 4 weeks (from 5 ? very often to 1 ? not at all). Participants indicated
their agreement with the item “Others have hindered me in achieving my
goals” on a scale ranging from 5 (very true) to 1 (not true at all), ? ? .68.
A T-transformed mean composite of the items was used, with higher values
indicating greater perceived strain with social partners. Perceived social
strain and social satisfaction correlated ?.37 (p ? .01).
Socioeconomic status was assessed by two indicators of years in edu-
cation and training and by occupation prestige. Years of education was
assessed with two items asking participants for their highest educational
degree and for their professional qualification. Answers were recoded into
the number of years needed to earn the respective degree and/or qualifi-
cation. This resulted in an average of 13.8 years in education or profes-
sional training (SD ? 2.7). Occupational prestige was assessed with
ratings of the relative social status of the participant’s current or last job on
the basis of Wegener’s Magnitude Prestige Scale (cf. Mayer, Maas, &
Wagner, 1999). Professional occupations were coded according to the
International Standard Classification of Occupations (ISCO). A statistical
program was used to assign ISCO scores of occupational prestige, resulting
in a scale of Occupational Prestige ranging from 20 to 186 (M ? 74.1,
SD ? 30.2). Education and occupational prestige correlated .51 (p ? .01).
Physical and mental health.
Physical health was assessed with a Ger-
man version of the somatization scale of the Symptom Checklist–90–R
(SCL-90-R; Derogatis & Cleary, 1977). Participants rated on a scale from
4 (very much) to 0 (not at all) the extent to which they experienced any of
12 somatic or functional complaints during the last 7 days. The sum score
of the somatization scale ranged from 0 to 40 (M ? 7.2, SD ? 6.4, ? ?
.82), with higher values indicating more somatic complaints. Depression
was assessed with the Center for Epidemologic Studies–Depression Scale
(CES-D; Radloff, 1977). Participants rated on a scale from 3 (very often)
to 0 (not at all) to 3 (very often) how often they had experienced each of
20 symptoms during the last week (internal consistency; M ? 12.0,
SD ? 7.7, ? ? .84). Somatic complaints and depression scores correlated
.51 (p ? .01).
Socioeconomic status, physical and mental health, and gender (with
values of 1 ? men and 2 ? women) were used as covariates in all analyses.
Card-Sort Tasks of Goal Priority and Partner Preference
In two card-sort tasks—one on subjective priority of goals and one on
preference for social partners—participants sorted cards (size: 3.7 in. ? 2.5
in.) with descriptions, written in large letters, of various social goals (see
Table 1 for a list of items), or of social partners (see Table 2 for a list of
items). Prior to the social goal card sort and the partner-preference card
sort, participants completed an 18-item food-preference card sort in order
to have a practice trial. Participants first completed the card sort of partners
before sorting the social goal cards. This was necessary because the card
sort of social goals consisted of two subsequent tasks (see below). In order
to preclude ipsative effects on the results of the card sort, participants were
allowed to create as many piles with as many cards as they needed. Cards
also could be completely excluded.
Verbal instructions for the partner-preference card sort were as follows:
On these cards you will find descriptions of different persons with
whom one might have contact in everyday life. Please read each of
these cards carefully and imagine someone described on the card.
Please sort the cards into several piles with respect to how much you
would like to spend one day or evening together with that person.
Again, please begin on the left side with the pile of those persons with
whom you would prefer most to spend one day or evening. Please
order those persons with whom it is less or not at all preferable for you
to spend one day or evening on the right side of that first pile. You
may build as many piles as you think necessary.
Verbal instructions for the social-goal card sort were as follows:
On these cards you will find descriptions of different goals and plans
that one can have and find important in life. Please read each of these
cards carefully and sort the cards into several piles with respect to how
important these goals and plans are for you personally. Again, please
begin on the left side with the pile of those goals and plans which are
most important to you. Please order those goals and plans that are less
important or not important at all to you into piles on the right side.
You may build as many piles as you think necessary.
Scoring of the card-sort tasks.
from 2 to 10 for the goal-sort task (M ? 3.5, SD ? 1.1) and from 2 to 9
for the partner-preference-sort task (M ? 3.8, SD ? 1.2). The aim of the
card-sort tasks was to identify prioritized goals or social partners. For
scoring purpose, all stacks were reordered into three piles: (a) the pile with
the most prioritized goals or partners (Pile 3), (b) the pile with least
prioritized goals or partners (Pile 1), and (c) all remaining middle piles with
goals or partners of indefinite priority (Pile 2). In order to obtain priority
scores, each card was then assigned a score in the rank order of the piles
resulting in a 3-point rating for each card (ranging from 3 ? most prior-
itized goals or preferred partners to 1 ? not prioritized goals or not
Treatment of inconclusive card sorts.
pile of goal cards and were excluded from analyses that included the
card-sort data. For 88 participants (18.3%) who differentiated only two
piles of social goals, priority rankings of a second trial of the card-sort task
on goals were used. In this second trial, all participants sorted those cards
that were ranked as most important in the first card sort once again (an
additional instruction was “Please sort the goals with respect to how much
you think of these goals”). Together with this second goal sort, all partic-
ipants differentiated at least three piles in the goal card-sort task. It should
be noted that all reported results remained unchanged when including only
results of the first card-sort trial. In the partner-preference card sort, all
participants created at least two piles. No additional card sort was done in
the partner card-sort task.
Identification of constructs (factors) of goal priority and partner pref-
erence with exploratory and confirmatory factor analyses.
prioritized goals and partner preferences were identified with two varimax-
rotated exploratory factor analyses. Resulting factors represented sets of
cards that were likely to be ordered into the same stack. Orthogonal factor
rotation was chosen to obtain statistically independent constructs. Before
further examining the statistical associations of the constructs, we tested
the factorial invariance across age-cohort groups. Invariance of factors can
be seen as an essential validity condition. Demonstrating invariance se-
The initial number of stacks ranged
Five participants created just one
LANG AND CARSTENSEN
cures that meanings of factors do not differ between groups (MacCallum &
Austin, 2000). Analyses of factorial invariance relied on structural equation
modeling using the LISREL 8.2 package (Jo ¨reskog & So ¨rbom, 1993).
Following the recommendation of MacCallum and Austin (2000), the
root-mean-square error of approximation (RMSEA) with a cutoff value
close to .06 is used as a criterion for model fit. In addition, it is the
chi-square value divided by its associated degrees of freedom to be close
to 2. For the purpose of identification, all latent factors in the measurement
models were fixed to 1. No correlations among error residuals of observed
variables were allowed. Maximum likelihood procedure was used for
estimation of parameters.
Constructs of goal priority.
As shown in Table 1, all items had highest
loadings (above .35) on their theoretically specified factor. The four
obtained factors were labeled (a) Social Acceptance, (b) Autonomy, (c)
Generativity, and (d) Emotion Regulation. Two items had nonspecified
second factor loadings above .35 on other factors. The goal “Be available
to others who need to be comforted” was associated with the Social
Acceptance factor. The goal “have large experience of life” loaded on the
Autonomy factor. However, there was no indication that these unexpected
loadings changed the substantial meaning of the four factors.1
Goodness-of-fit statistics of the free estimation model were close to
satisfactory, ?2(471, N ? 475) ? 682.9, p ? .01 (?2/df ? 1.45; RMSEA ?
.053). In a second model, all factor loadings were forced to be equal across
age-cohort groups to test whether factor loadings were invariant of age
cohort. No significant change of the chi-square was observed, ??2(66, N ?
475) ? 78.7, p ? .10, indicating that factors were defined by the same item
loadings across age groups. In a next step, error residuals of items were
fixed to be equal across age groups. No significant increase of the chi-
square was found, ??2(40, N ? 475) ? 12.2, p ? .10. In a last step, the
exact loading pattern, as reported in Table 1 (including invariant error
residuals), was fitted to replicate the observed correlation matrices within
each of the three age groups, resulting in a significant improvement of the
model fit, ??2(47, N ? 475) ? 126.0, p ? .01; ?2(530, N ? 475) ? 674.8,
p ? .01; (?2/df ? 1.27; RMSEA ? .042).
Constructs of network partner preferences.
varimax-rotated factor solution for the card sort of social partners. All
items had highest loadings (above .40) on the theoretically specified factor.
The five factors were labeled as (a) Preference for Friend or Acquaintance
(e.g., “a close friend,” “a sympathetic acquaintance”), (b) Preference for
Knowledgeable Partner (e.g., “an author of a book that I have read”), (c)
Preference for Controversial Partner (e.g., “a recent acquaintance with
whom I have nothing in common”), (d) Preference for Formal Partner
(e.g., “a medical doctor”), and (e) Preference for Family or Relative (e.g.,
“a close member of my family”). It should be noted that the observed factor
structure did not justify a differentiation between the preference for “close
friend” and for “casual friend or acquaintance.” This may reflect a partic-
ularity of the German context in which “true” friendship may be cogni-
Table 2 shows the
1Exclusion of items that showed unexpected loadings did not change the
observed factor structures (of both card-sort tasks). Scores of factors that
were extracted after excluding items with unexpected loadings correlated
between .96 and 1.0 with scores of corresponding factors that were ex-
tracted after all items were included.
Priority of Goal Domains: Factor Loadings of Goal Ratings (Varimax; N ? 475)
Goal Rotated factor solution (social goal domain)
G1: Social acceptance (I)
6 Have good friends who accept me the way I am
8 Have close friends who trust me
10 Be able to confide in a close friend at any time
11 Receive good advice on important decision
12 Not feel lonely
G2: Autonomy (IV)
3 Determine my future by myself
5 Receive approval for my work
1 Have strong power of discernment
22 Be financially independent
19 Be well educated and knowledgeable
G3: Generativity (II)
14 Be available to others who need to be comforted
13 Leave my mark on this world
15 Give my knowledge/experience on to others
16 Help others to find their purpose in life
17 Have large experience of life
7 Be with people who set high value on my opinion
G4: Emotion regulation (III)
4 Be autonomous in my feelings
18 Know myself and my feelings very well
20 Have control over my feelings
2 Not depend on someone else’s feelings
Explained variance of factor (total: 39.6%) 15.4 6.79.9 7.5
printed in bold. G ? goal.
Factor numbers in parentheses indicate the order of factor extraction. Highest loadings of items are
FUTURE TIME PERSPECTIVE, GOALS, AND SOCIAL RELATIONSHIPS
tively represented as “elective affinity” (or “quasi-kinship”). Consequently,
“close friends” were not differentiated from sympathetic acquaintances and
may have not been identified well in this task. Two items showed second
loadings (greater than .35) on unexpected factors: preferring “a recent
acquaintance with whom I have much in common” loaded on Preference
for Knowledgeable Partner; preferring “a stranger of my age” loaded on
Preference for Controversial Partner. Again, we did not find any indication
that these unexpected loadings changed the substantial meanings of the
factors (see Footnote 1).
Goodness-of-fit indices for the freely estimated measurement model
across age groups were not fully satisfactory, ?2(372, N ? 475) ? 631.6,
p ? .01 (?2/df ? 1.70; RMSEA ? .066). No significant differences were
observed when constraining factor loadings, ??2(58, N ? 475) ? 63.1,
p ? .10, and error residuals, ??2(36, N ? 475) ? 27.8, p ? .10, to be equal
across age groups. When fitting the exact rotated factor solution (see Table
2) to the observed correlation matrix within each age group (including
invariant error residuals), no significant improvement of the model fit was
observed, ?2(29, N ? 475) ? 11.9, p ? .05. Fit statistics of this final model
were fairly satisfactory, ?2(495, N ? 475) ? 734.4, p ? .01 (?2/df ? 1.48;
RMSEA ? .055).
Statistical analyses were done with the statistical software package
SPSS for Macintosh computers (1995). Hierarchical regression analyses
were computed and interpreted according to the recommendations of
Aiken and West (1991). Statistical tests of interaction effects were
based on the product of the respective continuous predictor variables.
Such a procedure is associated with a reduced likelihood to detect
“true” interaction effects in field studies that are based on random
samples (McClelland & Judd, 1993). One reason is the constrained
statistical power of the product term. McClelland and Judd (1993),
therefore, suggested that one should only expect small effect sizes (e.g.,
accounting for 1% of the variation of the dependent variable) for
significant interaction effects (p ? .05), which may nevertheless be
considered as substantive.
Results are reported in two sections corresponding to our two
main research questions. In the first section, we inspect associa-
tions between FTP, chronological age, and patterns of goal and
partner preferences. In specific, we examine whether FTP differ-
entially predicts patterns of goal and partner preference. In the
second section, we examine the effects of goal priority, partner
preference, and FTP on the size, composition, and perceived
quality of personal network.
Domains of Partner Preference: Factor Loadings of Preference Ratings (Varimax; N ? 475)
Rotated factor solution (domain of partner
P1: Friend/acquaintance (II)
2 A close friend
20 A sympathetic acquaintance
6 A recent acquaintance with whom I have much in common
8 A friend who I haven’t seen for a long time
17 A casual acquaintance
P2: Knowledgeable partner (I)
11 An author of a book that I have read
13 An artist whose work I admire
1 An interesting stranger
9 A clergy person (priest, rabbi, etc.)
P3: Controversial partner (V)
3 A recent acquaintance with whom I have nothing in
4 Someone who I know well but do not like
P4: Formal partners/service (IV)
5 A medical doctor
7 An attorney
15 The mayor of my residential district
14 A stranger of my age
P5: Family/relative (III)
10 A close member of my family
12 A younger relative
16 A relative of same age
Explained variance of factor (total: 50.8%)12.1 16.26.0 7.3 9.2
printed in bold. P ? partner preference.
Factor numbers in parentheses indicate the order of factor extraction. Highest loadings of items are
LANG AND CARSTENSEN
Predicting Patterns of Goal and Partner Preferences
From FTP and Age
FTP as a proxy of chronological age.
search is that FTP is connected with chronological age. In fact,
FTP was found to be strongly negatively associated with age
cohort (r ? ?.70). The factor communality of chronological age
and FTP was .852. Note that this association remained fairly
unchanged when controlling for covariates (i.e., marital status,
gender, health, socioeconomic status; r ? ?.69). In order to
differentiate age-associated variance of FTP from nonshared vari-
ance of age cohort (i.e., not associated with time perspective) in the
following analyses, age cohort was residualized on FTP. Residual
age contains therefore only age-associated information uncorre-
lated with FTP. Note also that when replacing FTP with chrono-
logical age in all of our statistical analyses, the same pattern of
findings emerged, with old age representing a limited FTP and
young age representing an open-ended time perspective.
FTP, goal priorities, and partner preferences.
plays the correlations of FTP and characteristics of personal net-
works with goal and partner preferences. It should be noted that all
goal and partner preferences pertained to factor scores of the
varimax-rotated factor solutions. Therefore, correlation coeffi-
cients among goals and among partner preferences were zero
(because of the orthogonal factor rotation). As shown in Table 3,
FTP was differentially associated with patterns of goal and partner
preference. These associations are illustrated in Figures 1 and 2.
For illustrative purpose, a tercile split on FTP scores was per-
formed, resulting in three groups of participants with limited,
indefinite, and open-ended time perspectives.
Figure 1 shows the associations between priority of goals and
the three groups of FTP. The effect illustrated in Figure 1 was also
found to be significant (p ? .01) in a 3 (FTP) ? 4 analysis of
variance (ANOVA), with domain of goal priority as a within-
subject factor, F(6, 1416) ? 7.4, p ? .01. As shown in Figure 1,
generativity and emotion-regulatory goals were prioritized among
individuals who perceived their future time as limited. In contrast,
social acceptance and autonomy were prioritized among partici-
pants who felt their future to be open-ended.
Figure 2 shows that when perceiving the future as limited,
individuals preferred relatives and formal partners (e.g., helpers)
over other social partners. In contrast, when perceiving the future
as open-ended, knowledgeable partners, friends or acquaintances,
and controversial partners were preferred over relatives and formal
partners. The effect illustrated in Figure 2 was found to be signif-
icant (p ? .01) in a 3 (FTP) ? 5 ANOVA, with domains of
preference as a within-subject factor, F(8, 1900) ? 6.2, p ? .01.
A premise of our re-
Table 3 dis-
Predicting Personal Network Characteristics From Social
Goals, Partner Preferences, and FTP
In the second section of our analyses, we examined whether
goals and preferences were differently associated with the size,
composition, and perceived quality of personal networks depend-
ing on levels of FTP. We computed seven hierarchical regression
analyses with each of seven dependent variables of personal net-
work characteristics (i.e., size, residualized number of relatives,
friends, nonkin known less than 10 years, and of advice givers,
perceived strain with others, and social satisfaction).2All possible
two-way interaction effects of goal priority or partner preference
with FTP were tested. In a first step, we inspected the bivariate
correlations among the variables to explore whether quantitative
and qualitative characteristics were reflected in observed patterns
of goal and partner preferences.
Bivariate associations of personal networks, goal priority, and
Table 3 summarizes the bivariate correla-
tions between priority of goal domains and partner preferences
with the quantity and perceived quality of personal networks.
When prioritizing social acceptance goals, individuals reported
larger personal networks than others, more friends, and more
advice givers, but also perceived greater strain with network part-
ners. Prioritizing autonomy goals was associated with perceiving
greater social strain. When prioritizing generativity goals, individ-
uals reported more relatives, fewer friends, fewer new nonkin
partners, and fewer advice givers and expressed greater social
satisfaction. When individuals prioritized emotion regulatory
goals, they reported smaller personal networks, more relatives,
fewer friends, fewer novel nonkin partners (known less than 10
years), and fewer advice givers. When preferring friends or ac-
quaintances, participants reported larger networks, fewer relatives,
more friends, and more novel nonkin partners, but also perceived
greater strain and felt less satisfied with social partners. Preference
for formal partners was associated with smaller networks and with
fewer novel nonkin in the network. When expressing a preference
for family or relatives, this was reflected in a larger number of
relatives, fewer friends, and fewer new nonkin partners. Moreover,
a strong preference for family or relatives was associated with
perceiving less social strain and greater social satisfaction.
Interaction effects of goal priority and partner preferences with
FTP on personal networks.
In a next step, we computed seven
hierarchical regression analyses with size, composition, and per-
ceived quality of personal networks as criterion variables, and
domains of goal priority, domains of partner preferences, residual
age, and FTP as predictors (entered in that order into the equation,
followed by tests of interaction effects). The focus of these anal-
yses was to test all possible interaction effects of goal priority or
partner preference with FTP on characteristics of personal net-
works (including also those for which we did not have predic-
tions). Tables 4 and 5 show the unstandardized coefficients and
2The observed associations between the moderator variables raised the
issue of homogeneity of variances (i.e., homoscedasticity). The following
measures were taken in order to test for violations of the homoscedasticity
assumption: (a) Outliers were detected with the criterion of standardized
residuals in excess of ?3.3 (Tabachnik & Fidell, 1996). Few outliers were
identified and all reported effects remained significant after exclusion of
these outliers. (b) A visual inspection of regression errors (residuals) was
plotted against the predicted values in order to detect homoscedasticity.
Only with respect to Preference for Controversial Partner could heterosce-
dasticity be detected. After a square-root transformation of this variable,
heteroscedasticity was clearly reduced. This procedure did not affect any of
the reported results. (c) Homogeneity of variances was tested across the
three groups of limited, indefinite, and open-ended time perspective. Fmax
was found to be acceptable and below 1.2 for all variables. Bartlett-Box F
was significant at p ? .05 for salience of acceptance goals, Preference for
Friend/Acquaintance and Preference for Controversial Partner. After ex-
clusion of outliers (see above), Bartlett-Box F lost significance for these
FUTURE TIME PERSPECTIVE, GOALS, AND SOCIAL RELATIONSHIPS
their unique contributions to the explained variance for the final
model (as presented in the tables). Reported effects remain un-
changed when entering covariates into the equation (i.e., gender,
socioeconomic status, depression, health). It should be noted that
when replacing FTP by chronological age in the regression anal-
yses, effects of chronological age replicated the observed effects of
FTP throughout (with effects of limited time perspective reflecting
effects of older age). Significance tests of interaction effects refer
to effects of the continuous variables. However, for illustrations of
interaction effects we used the tercile split of FTP scores as
Predicting the size and composition of personal networks.
ble 4 shows the results of four hierarchical regression analyses of
size and composition of personal networks on FTP, residual age,
priority of goals (social acceptance, autonomy, generativity, emo-
tion regulation), and partner preferences (friend/acquaintence,
knowledgeable partner, controversial partner, formal partner, fam-
ily/relative). As shown in Table 4, four interaction effects of FTP
and emotion-regulation goals on size and composition of personal
networks were found to be significant (p ? .05). Main effects of
FTP were significant (p ? .05) in all four regression models
before entering the interaction terms into the regression equation.
Figure 3 illustrates the observed significant interaction effect of
emotion-regulatory goals and FTP on the size of personal network,
t(461) ? 2.9, p ? .01. When individuals perceived their futures as
limited, the priority of emotion-regulatory goals was negatively
associated with the size of their personal networks (r ? ?.20, N ?
160, p ? .05), whereas there was no such effect when they
perceived their future as open-ended (r ? ?.02, N ? 160, ns).
Three additional interaction effects between priority of emotion-
regulatory goals and FTP on composition of personal networks
were observed (see Table 4). These effects indicated that associ-
ations between network composition and priority of emotion reg-
ulation were strongest among participants with an open-ended
FTP. Among participants who had an open-ended FTP, low pri-
ority of emotion regulation was associated with reporting a larger
(residual) number of friends, t(461) ? ?3.0, p ? .01; a larger
(residual) number of new nonkin partners, t(461) ? ?3.2, p ? .01;
and a smaller (residual) number of relatives, t(461) ? 3.6, p ? .01.
When priority of emotion-regulatory goals was high, number of
friends, relatives, and new nonkin in the network were comparable
across the three groups of FTPs. This finding indicates that a low
priority of emotion regulation was related to the network compo-
sition among participants who perceived their future as expansive.
These results do not indicate whether the priority of emotion-
regulatory goals among participants who perceived future time as
limited was also associated with positive emotional outcomes, and
with receiving advice support.
Predicting receipt of advice and perceived quality of personal
Our next analysis explored the emotion-related out-
comes of social goals and FTP in relation to the number of advice
givers, social satisfaction, and perceived strain with partners. Ta-
ble 5 shows the results and final models of three hierarchical
regression analyses (that were computed analogously to those
reported in Table 4). Four significant interaction effects on the
three indicators of perceived quality of personal networks were
observed: (a) an interaction effect of FTP ? Preference for Knowl-
edgeable Partner on the Number of Advice Givers, t(461) ? 2.4,
p ? .05; (b) two interaction effects of FTP ? Priority of Emotion-
Correlations of Social Goals and Partner Preferences With Network Characteristics, Covariates, and Future Time Perspective (FTP; N ? 475)
Priority of goal
Size of personal network
No. of relatives (residual)
No. of friends (residual)
No. of new nonkin (residual)
No. of advice givers (residual)
Perceived strain with others
Years of education
Somatic complaints (SCL-90-R)
(after controlling for covariates 8–13)
Priority of social goals (Columns 2–5) and partner preferences (Columns 6–10) refer to factor scores of two orthogonal factor rotations: Correlations among goals and correlations among
preferences are therefore zero. CES-D ? Center for Epidemiologic Studies Depression Scale; SCL-90-R ? Symptom Checklist–90–R.
* p ? .05.
** p ? .01.
LANG AND CARSTENSEN
Regulation Goals, t(461) ? 2.2, p ? .05, and of FTP ? Preference
for Knowledgeable Partner, t(461) ? ?2.1, p ? .05, on Perceived
Strain with Partners; and (c) on interaction effects of FTP ?
Priority of Emotion-Regulation Goals on Social Satisfaction,
t(461) ? ?2.0, p ? .05. Figures 4, 5, and 6 illustrate the observed
interaction effects. Note that a tercile split of FTP was used in
these figures only for the purpose of illustration. FTP was entered
as a continuous variable in all analyses.
Figure 4 shows that Preference for Knowledgeable Partner was
positively associated with a greater number of advice-giving support-
ers among the group of participants who perceived their future time as
open-ended (r ? .20, N ? 160, p ? .05). No such association was
ns) or indefinite (r ? .06, N ? 160, ns). This finding indicates that
even when individuals who perceive their future as limited express a
preference for knowledgeable partners, they appear to rely on one
such partner rather than on two or more such partners. It may be that
individuals who perceive their future as limited and who seek advice,
do this in a more emotionally close relationship than individuals who
perceive their future as open-ended. However, with the present data
set, we are unable to further investigate this speculation.
Figure 5 shows that perceiving strain in the personal network
was differentially associated with emotion-regulatory goals and
with preference of knowledgeable partners depending on FTPs.
When individuals perceived their future time as limited, they
experienced less strain when prioritizing goals relating to emotion
regulation (r ? ?.16, N ? 160, p ? .05). In addition, Figure 5B
illustrates that not preferring knowledgeable partners is associated
with less social strain among individuals who perceive their future
as restricted (r ? .18, N ? 160, p ? .05).
Finally, a congruence of goals and time perspective was found
to also be associated with social satisfaction. Figure 6 shows that
when individuals perceived their future as limited, prioritizing
emotion-regulation goals was also associated with greater social
Priority of goals is differentially associated with future time perspectives. N ? 475.
Preference for partners is differentially associated with future time perspectives. N ? 475.
FUTURE TIME PERSPECTIVE, GOALS, AND SOCIAL RELATIONSHIPS
satisfaction (r ? .19, N ? 160, p ? .05). No such association was
found in the group of individuals who perceived the future as
open-ended (r ? ?.04, N ? 160, ns).
To sum up the results, FTP was found to be differentially
associated with goal and partner preferences. When future time
was perceived as limited, emotion regulation, generativity, family
or relatives, and formal helpers were preferred. In contrast, when
future time was perceived as open-ended, goals relating to social
acceptance, to autonomy, and to seeking contact with friends or
acquaintances were prioritized. The second part of our analyses
revealed that a preeminence of emotion-regulatory goals was dif-
ferently associated with the size, composition, and perceived qual-
ity of personal networks depending on FTP. For example, when
time perspective was limited, preferring emotion-regulatory goals
was associated with smaller personal networks. When time per-
spective was expansive, a preference for knowledgeable partners
was associated with receiving more advice from social partners.
Finally, a preference for emotion-regulatory goals was associated
with greater satisfaction and fewer perceived strains, but only
among those participants who perceived their future as limited.
We examined the associations among FTP, social goals, and
personal networks across the adult life span. Findings indicate that
goals are selected in congruence with one’s construal of future
time as being restricted or expansive, and more important, that
such congruence contributes to a more positive quality of one’s
relationships across adulthood. In specific, prioritizing emotionally
meaningful goals when the future is perceived as limited was
associated with smaller personal networks, greater social satisfac-
tion, and less social strain. Such associations were not found when
individuals perceived their future as expansive. In this latter case,
however, a preference for knowledgeable social partners was as-
sociated with perceiving less strain with partners. Findings lend
support to the claim of socioemotional selectivity theory that when
individuals select goals and partners in accordance with their view
of their remaining time to live, they benefit more from their social
contexts. We discuss the implications of our findings with respect
to FTP, chronological age, goal preference, and the quantity and
perceived quality of personal networks.
FTP and Chronological Age
FTP was strongly associated with chronological age. In fact, our
measure of FTP shared more than 50% of the variance with age,
suggesting that FTP captured much of the psychological informa-
tion associated with age differences in our cross-sectional study
(cf. Wohlwill, 1970). No other variable was found to be as strongly
associated with age. Moreover, the association of FTP with age
remained unchanged even after controlling for effects of health.
This may also serve to underscore the validity of our measure of
FTP as a proxy variable of age-associated information. Conse-
quently, it seemed inadequate to us to include both age and FTP
simultaneously into our analyses without losing information be-
cause of statistical collinearity (Tabachnik & Fidell, 1996). How-
ever, we found that the observed effects of FTP could be perfectly
replicated when replacing FTP with chronological age throughout
our analyses (with older age mirroring limited FTP). This points to
a caveat of our study that needs to be considered when interpreting
the results. It is not possible to draw conclusions about the inter-
play of chronological age with FTP in regulating personal networks
Four Regression Analyses of Time Perspective and Goals on Personal Network Composition
Size of network Relativesa
Nonkin known less
than 10 yearsa
Goals (G) and partner preferences (P)
G1: Social acceptance
G4: Emotion regulation
P2: Knowledgeable partner
P3: Controversial partner
P4: Formal Partner/service
Two-way interaction effect
FTP ? Emotion Regulation
health) to the equation. No further interaction effects with future time perspective (FTP) were significant. Standard errors are shown in parentheses. Adj.
aNumber of relatives, friends, and recent nonkin partners were residualized on network size.
* p ? .05.
Unique R2of effects in the final regression model. Effects remain unchanged after entering covariates (gender, socioeconomic status, depression,
bUnstandardized beta coefficients (B) of the final model.
LANG AND CARSTENSEN
with cross-sectional data. Clearly, longitudinal research is needed to
more accurately disentangle the effects of FTP from those of chro-
nological age. Intraindividual change in FTP across the life span
appears plausible. Findings from experimental studies suggest, for
example, that manipulation of FTPs could easily be reversed in young
as well as in later adulthood (e.g., Fung et al., 1999).
Prioritizing Emotionally Meaningful Goals When the
Future Is Limited
Our first research question aimed at replicating earlier findings
in a broad and age-heterogeneous sample and using a continuous
measure of FTP. We found that individuals who perceived their
future to be limited placed higher priority on goals or social
Regression Analyses of Future Time Perspective (FTP) and Goals on Social Satisfaction and Advice
Perceived strain Social satisfaction
Goals (G) and partner preferences (P):
G1: Social acceptance
G4: Emotion regulation
P2: Knowledgeable partner
P3: Controversial partner
P4: Formal partner/service
Two-way interaction effects
FTP ? Emotion Regulation
FTP ? Knowledgeable Other
n.s.— 0.12* (.06)
— 0.03* (.01)
depression, health) to the equation. No further interaction effects with FTP were significant. Standard errors are shown in parentheses. Adj. ? adjusted.
aNumber of supporters (advice, esteem) were residualized on network size.
* p ? .05.
Unique R2square of effects in the final regression model. Effects remain unchanged after entering covariates (gender, socioeconomic status,
bUnstandardized beta coefficients (B) of the final model.
personal networks when the future is perceived as limited. Future time
perspective was used as continuous variable in the regression analyses. A
tercile split was done in order to illustrate the significant interaction effect
(see Table 3).
Prioritizing emotion-regulatory goals is associated with smaller
more often advice from others when the future is perceived as open-ended.
Future time perspective was used as continuous variable in the regression
analyses. A tercile split was done in order to illustrate the significant
interaction effect (see Table 3).
Preferring knowledgeable others is associated with receiving
FUTURE TIME PERSPECTIVE, GOALS, AND SOCIAL RELATIONSHIPS
partners associated with strong emotional meaning. In contrast,
when individuals perceived their future as open-ended, they ex-
pressed preferences for knowledgeable partners, and for goals
aimed at achieving or maintaining autonomy and social accep-
tance. These findings replicate results from earlier studies (e.g.,
Carstensen et al., 1999; Fredrickson & Carstensen, 1990) but also
serve to demonstrate the generality of the phenomenon.
For example, in our research, we relied on a relatively broad and
heterogeneous set of social goals. This allowed us to identify
distinct subtypes of emotionally meaningful goals, one aiming at
emotion regulation and the other being a set of generativity goals.
Both subtypes of emotionally meaningful goals were preferred
when individuals perceived their future as limited. Note that per-
ceptions of future time were not residualized on age. Our findings
on generativity are therefore consistent with studies that report
stronger prosocial orientations in late adulthood as compared with
early adulthood (e.g., Ryff & Heincke, 1983; Van Lange et al.,
1997). One should recall that our assessment of generativity goals
also included aspects of symbolic immortality (e.g., “leave my
mark on this world”). These findings suggest that priority of
generativity in the second half of life appears to reflect an in-
creased need of emotional meaning when one’s remaining lifetime
Furthermore, the finding that individuals who perceived their
future time as expansive were more strongly committed to auton-
omy goals and to social-acceptance goals is also in accordance
with predictions of socioemotional selectivity theory. Seeking to
have success in one’s career, wanting to become knowledgeable,
or seeking social partners who express their acceptance and trust
seem to be important goals when individuals perceive their future
as expansive. We do not argue that autonomy goals or social-
acceptance goals are not also emotionally relevant. Rather we
argue that prioritizing autonomy or social-acceptance goals is
associated with a more instrumental and information-seeking in-
terest in one’s social partners (e.g., “ . . . to accept me the way I
am,” or “ . . . to approve my independence”). Such partners may or
may not also provide emotionally meaningful experiences to the
individual, but they are primarily contacted for instrumental
What about findings concerning preferences for formal partners,
and for friends or acquaintances? In fact, at first sight our finding
that formal partners (e.g., medical doctor) were preferred over
friends when individuals perceived their future time as limited
seems to be at odds with the theory. Note, though, that our measure
of Preference for Friends or Acquaintances mostly concerned
casual friends and acquaintances. Thus, the reduced preference for
friends or acquaintances when time is perceived as limited is
consistent with other findings that suggest a reduced contact with
less close social partners when approaching the end of life (e.g.,
Lang, 2000). The greater preference for formal partners when
future time is limited is more difficult to integrate. Seeking or
preferring formal partners (e.g., a lawyer to set up one’s testimony,
or a medical doctor) may also imply a regulation of emotion when
time appears limited. In addition, information or benefits from
formal partners (or “a stranger of my age”) can typically be
of emotion-regulatory goals and (b) preference for knowledgeable partner
depending on future time perspectives. Future time perspective was used as
continuous variable in the regression analyses. A tercile split was done in
order to illustrate the significant interaction effect (see Table 3).
Perceived social strain is differently associated with (a) priority
emotion-regulatory goals depending on future time perspectives. Future
time perspective was used as continuous variable in the regression analy-
ses. A tercile split was done in order to illustrate the significant interaction
effect (see Table 3).
Social satisfaction is differently associated with priority of
LANG AND CARSTENSEN
received immediately, in contrast to rewards associated with
knowledgeable partners. Preference for formal partners, thus, may
reflect (short-term) exchange orientations as, for example, Clark
(1984) has proposed. One consequence of such orientations may
be that older individuals transfer their instrumental needs to ex-
changes with formal partners in order to maximize their potential
for meaningful, intrinsic benefits from their informal partners.
Matching Goals and Personal Networks to FTP
A premise of our research is that social goals are central to
self-regulation, to social relationships, and to emotional experience
(e.g., Brunstein et al., 1998; Lansford, Sherman, & Antonucci,
1998; Omodei & Wearing, 1990). Not surprisingly, characteristics
of personal networks were associated with consistent goal and
partner preferences. For example, when prioritizing social accep-
tance, individuals reported larger personal networks, a greater
proportion of friends, and receiving advice more often. In contrast,
when prioritizing emotion-regulatory goals, individuals reported
smaller networks and received advice less often. Theoretically
more important, however, was the finding that the associations
between an individual’s goal or partner preferences and the char-
acteristics of his or her personal network were found to depend on
time perspective. That is, when the future appeared restricted in
time, emotion-related social goals and partner preferences influ-
enced the composition of personal networks as well as the quality
of social contact. However, when the future was perceived as
expansive, instrumental and knowledge-related goals had an influ-
ence on the composition of personal networks.
In those instances where emotion-regulatory goals were associ-
ated with the composition of personal networks, even though the
future was perceived as open-ended, participants did not fare as
well as their counterparts, reporting fewer friends, fewer novel
nonkin partners, and more relatives in the personal network. This
finding can be interpreted in light of our other finding that the
priority of emotion-regulatory goals when perceiving the future as
open-ended was associated with greater social strain. While
emotion-regulatory goals were generally related to more emotion-
ally meaningful social contact, the outcomes of such contacts were
more likely to be negative (when not congruent with the time
perspective). Seeking to maximize emotionally meaningful expe-
rience in one’s social environment may not be adaptive when the
future is perceived as being filled with possibilities. To be clear,
we do not make “all or none” distinctions about goal pursuits.
Socioemotional selectivity theory claims that the same essential set
of goals motivates social behavior across the life span. There are
many times when opportunities for expanding one’s horizons do
not conflict with pursuing emotionally meaningful goals. Indeed,
often times, different types of goals are bundled together. Rather,
the theory addresses those times when goals compete. In early
adulthood, spending time with others who do not seem to have
much in common with one’s self (i.e., controversial others) may
contribute to enhancing one’s knowledge about one’s self (and
thus be at times preferred over being with one’s parents). More-
over, for a 20-year-old, emphasizing emotional satisfaction over
information acquisition in college, for example, may not be adap-
tive. Forgoing parties and attending singularly to learning even
difficult material may be highly adaptive. For an 80-year-old
returning student, emphasizing emotional satisfaction in one’s
college studies may be highly adaptive.
Another finding was that the quality of personal networks also
reflected a match between FTP and goals. Although there is
considerable agreement that having or pursuing goals is generally
relevant to positive or negative emotional experience (e.g., Diener,
Suh, Lucas, & Smith, 1999), the focus of our research lay on the
contents of preeminent goals. We found that associations between
specific goals and the quality of social exchanges differed when
individuals perceived their future time as restricted or as expan-
sive. When time was perceived as limited, high importance of
emotion regulation and low preference for knowledgeable partners
was associated with less social strain. In contrast, when time was
perceived as open-ended, high importance of emotion regulation
and low preference for knowledgeable partners was associated
with greater social strain. Individuals may feel more often strained
when they pursue goals that are not in accordance with their
perceived future time. This also points to the distinction between
antecedent and response-focused emotion regulation (e.g., Gross,
1998). For example, when time is scarce, individuals who priori-
tize emotionally meaningful goals may be better at selecting part-
ners who reflect such needs.
One limitation must be considered when interpreting these find-
ings. As previously stated, findings are based on cross-sectional
data. Consequently, results should not be interpreted in terms of
change. It remains an open question whether individuals who
experience changes in FTP and social goals also engage in efforts
to modify their personal networks in accordance with such
changes. In addition, the cross-sectional character of our study
prevents any conclusion about the direction of observed effects.
Perceived quality of personal networks may also trigger individ-
uals to prioritize goals that are not in accordance with their FTP.3
For example, individuals may focus on goal domains that are not
congruent with a limited time perspective when emotion-
regulatory goals are blocked by their partners. Also, our findings
are based exclusively on individual self-reports. However, social
relationships, per definition, involve at least two partners who
influence the course and outcome of interactions. Consequently,
regulation of personal networks is a product of complex social
transactions between at least two individuals who may or may not
pursue congruent or compatible social goals. Delving further into
the regulatory effectiveness of social transactions that show a
congruence between goal and time perspective appears a promis-
ing venue for future research. For example, we would predict that
when emotion-regulatory goals that are congruent with time per-
spective are blocked by the social behaviors of a specific partner,
individuals would rather seek other emotionally meaningful expe-
riences (e.g., generativity) than give up the priority of emotion
To conclude, our findings lend support to the notion that effects
of FTP on the regulation of personal networks depend on the
content of prioritized goals. Selecting goals in congruence with
perceptions of remaining time in life appears to contribute to better
perceived quality of one’s personal network. Individuals who
recognize that their time is limited seem to benefit in their social
worlds when they focus on goals relating to the maximization of
their emotionally meaningful experience.
3We are grateful to an anonymous reviewer for pointing us toward this
alternative interpretation for our finding.
FUTURE TIME PERSPECTIVE, GOALS, AND SOCIAL RELATIONSHIPS
Aiken, L. S., & West, S. G. (1991). Multiple regression: Testing and
interpreting interactions. Newbury Park, CA: Sage.
Baltes, M. M., & Carstensen, L. L. (1996). The process of successful aging.
Aging and Society, 16, 397–422.
Baltes, P. B. (1987). Theoretical propositions of life-span developmental
psychology: On the dynamics between growth and decline. Develop-
mental Psychology, 25, 611–626.
Bandura, A. (1997). Self-efficacy. The exercise of control. New York:
Brunstein, J. C., Schultheiss, O. C., & Grassman, R. (1998). Personal goals
and emotional well-being: The moderating role of motive dispositions.
Journal of Personality and Social Psychology, 75, 494–508.
Carstensen, L. L. (1993). Motivation for social contact across the life span:
A theory of socioemotional selectivity. Nebraska Symposium on Moti-
vation, 40, 209–254.
Carstensen, L. L., & Fredrickson, B. (1998). Influence of HIV status and
age on cognitive representations of others. Health Psychology, 17,
Carstensen, L. L., Graff, J., & Lang, F. R. (2000). Psychology’s contribu-
tions to gerontology. In J. M. Clair (Ed.), Multidisciplinary perspectives
on aging: Progress and priorities (pp. 29–248). Amityville, NY:
Carstensen, L. L., Isaacowitz, D. M., & Charles, S. T. (1999). Taking time
seriously: A theory of socioemotional selectivity. American Psycholo-
gist, 54, 165–181.
Carstensen, L. L., & Lang, F. R. (1996). Future Orientation Scale. Un-
published manuscript, Stanford University.
Caspi, A., Bem, D. J., & Elder, G. H., Jr. (1989). Continuities and
consequences of interactional styles across the life course. Journal of
Personality, 57, 375–406.
Clark, M. S. (1984). A distinction between two types of relationships and
its implications for development. In J. C. Masters & K. Yarkin-Levin
(Eds.), Boundary areas in social and developmental psychology (pp.
241–270). New York: Academic Press.
Derogatis, L. R., & Cleary, P. A. (1977). Confirmation of the dimensional
structure of the SCL-90: A study in construct validation. Journal of
Clinical Psychology, 33, 981–989.
de Volder, M. L., & Lens, W. (1982). Academic achievement and future
time perspective as a cognitive-motivational concept. Journal of Per-
sonality and Social Psychology, 42, 566–571.
Diener, E., Suh, E. M., Lucas, R. E., & Smith, H. L. (1999). Subjective
well-being: Three decades of progress. Psychological Bulletin, 125,
Ferring, D., & Filipp, S.-H. (1999). Soziale Netze im Alter: Selektivi-
taet in der Netzwerkgestaltung, wahrgenommene Qualitaet der Sozi-
albeziehungen und Affekt [Social networks in old age: Socioemo-
tional selectivity, perceived quality of social interaction, and affect].
Zeitschrift fuer Entwicklungspsychologie und Paedagogische Psy-
chologie, 31, 127–137.
Fratiglioni, L., Wang, H.-X., Ericsson, K., Maytan, M., & Winblad, B.
(2000). Influence of social network on occurrence of dementia: A
community-based longitudinal study. Lancet, 355, 1315–1319.
Fredrickson, B. L., & Carstensen, L. L. (1990). Choosing social partners:
How old age and anticipated endings make people more selective.
Psychology and Aging, 5, 335–347.
Fung, H. H., Carstensen, L. L., & Lutz, A. M. (1999). Influence of time on
social preferences: Implications for life-span development. Psychology
and Aging, 14, 595–604.
Gross, J. J. (1998). Antecedent- and response-focused emotion regula-
tion: Divergent consequences for experience, expression, and
physiology. Journal of Personality and Social Psychology, 74, 224–
Hill, C. A. (1991). Seeking emotional support. The influence of affiliative
need and partner warmth. Journal of Personality and Social Psychol-
ogy, 60, 112–121.
Jarvinen, D. W., & Nicholls, J. G. (1996). Adolescents’ social goals, beliefs
about the causes of social success, and satisfaction in peer relations.
Developmental Psychology, 32, 435–441.
Jo ¨reskog, K. G., & So ¨rbom, D. (1993). LISREL 8: User’s reference guide.
Chicago: Scientific Software International.
Kahn, R. L., & Antonucci, T. C. (1980). Convoys over the life course.
Attachment, roles and social support. In P. B. Baltes & O. G. Brim
(Eds.), Life-span development and behavior (pp. 254–283). New York:
Lang, F. R. (2000). Endings and continuity of social relationships:
Maximizing intrinsic benefits within personal networks when feeling
near to death? Journal of Social and Personal Relationships, 17,
Lang, F. R. (2001). Regulation of social relationships in later adult-
hood. Journal of Gerontology: Psychological Sciences, 56B, P321–
Lang, F. R., & Carstensen, L. L. (1994). Close emotional relationships in
late life: Further support for proactive aging in the social domain.
Psychology and Aging, 9, 315–324.
Lang, F. R., Staudinger, U. M., & Carstensen, L. L. (1998). Socioemotional
selectivity in late life: How personality and social context do (and do
not) make a difference. Journal of Gerontology: Psychological Sciences,
Lansford, J. E., Sherman, A. M., & Antonucci, T. C. (1998). Satisfaction
with social networks: An examination of socioemotional selectivity
theory across cohorts. Psychology and Aging, 13, 544–552.
MacCallum, R. C., & Austin, J. T. (2000). Applications of structural
equation modeling in psychological research. Annual Review of Psychol-
ogy, 51, 201–226.
Mayer, K. U., Maas, I., & Wagner, M. (1999). Socioeconomic conditions
and social inequalities in old age. In P. B. Baltes & K. U. Mayer (Eds.),
The Berlin aging study (pp. 227–255). New York: Cambridge University
McAdams, D. P., Hart, H. M., & Maruna, S. (1998). The anatomy of
generativity. In D. P. McAdams & E. de St. Aubin (Eds.), Generativity
and adult development (pp. 7–43). Washington, DC: American Psycho-
McClelland, G. H., & Judd, C. M. (1993). Statistical difficulties of detect-
ing interactions and moderator effects. Psychological Bulletin, 114,
Okun, M. A., & Keith, V. M. (1998). Effects of positive and negative social
exchanges with various sources on depressive symptoms in younger and
older adults. Journal of Gerontology: Psychological Sciences, 53B,
Omodei, M. M., & Wearing, A. J. (1990). Need satisfaction and
involvement in personal projects: Toward an integrative model of
subjective well-being. Journal of Personality and Social Psychol-
ogy, 59, 762–769.
Radloff, L. S. (1977). The CES-D Scale: A self-report depression scale for
research in the general population. Applied Psychological Measure-
ment, 1, 385–401.
Rook, K. (1984). The negative side of social interaction: Impact on
psychological well-being. Journal of Personality and Social Psychol-
ogy, 46, 1097–1108.
Ryff, C. D., & Heincke, S. G. (1983). Subjective organization of person-
ality in adulthood and aging. Journal of Personality and Social Psychol-
ogy, 44, 807–816.
SPSS Macintosh (Version 6.1.1) [Computer software]. (1995). Chicago:
Tabachnik, B. G., & Fidell, L. S. (1996). Using multivariate statistics (3rd
ed.). New York: Harper Collins.
LANG AND CARSTENSEN
Uchino, B. N., Cacioppo, J. T., & Kiecolt-Glaser, J. K. (1996). The Download full-text
relationship between social support and physiological processes: A
review with emphasis on underlying mechanisms and implications for
health. Psychological Bulletin, 119, 488–531.
Vaillant, G. E., & Milofsky, E. (1980). Natural history of male psycho-
logical health: IX. Empirical evidence for Erikson’s model of the life
cycle. American Journal of Psychiatry, 137, 1348–1359.
Van Lange, P. A., De Bruin, E. M., Otten, W., & Joireman, J. A. (1997).
Development of prosocial, individualistic, and competitive orientations:
Theory and preliminary. Journal of Personality and Social Psychol-
ogy, 73, 733–746
Wohlwill, J. F. (1970). The age variable in psychological research. Psy-
chological Review, 77, 49–64.
Revised July 25, 2000
Revision received April 2, 2001
Accepted April 13, 2001 ?
FUTURE TIME PERSPECTIVE, GOALS, AND SOCIAL RELATIONSHIPS