Download as pdf or txt
Download as pdf or txt
You are on page 1of 14

Two Views of One Relationship: Comparing Parents' and Young Adult Children's Reports

of the Quality of Intergenerational Relations


Author(s): William S. Aquilino
Source: Journal of Marriage and Family , Nov., 1999, Vol. 61, No. 4 (Nov., 1999), pp.
858-870
Published by: National Council on Family Relations

Stable URL: https://www.jstor.org/stable/354008

REFERENCES
Linked references are available on JSTOR for this article:
https://www.jstor.org/stable/354008?seq=1&cid=pdf-
reference#references_tab_contents
You may need to log in to JSTOR to access the linked references.

JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide
range of content in a trusted digital archive. We use information technology and tools to increase productivity and
facilitate new forms of scholarship. For more information about JSTOR, please contact support@jstor.org.

Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at
https://about.jstor.org/terms

National Council on Family Relations is collaborating with JSTOR to digitize, preserve and
extend access to Journal of Marriage and Family

This content downloaded from


103.59.198.83 on Mon, 28 Aug 2023 19:50:10 +00:00
All use subject to https://about.jstor.org/terms
WILLIAM S. AQUILINO University of Wisconsin

Two Views of One Relationship:


Comparing Parents' and Young Adult Children's
Reports of the Quality of Intergenerational Relations

Patterns of agreement and disagreement on the perspectives of family members and the factors
quality of intergenerational relationships were ex- that affect the level of agreement in self-report stud-
plored in a sample of parents and young adult ies (Tein, Roosa, & Michaels, 1994). The degree
children. Data on parent-child closeness, contact, to which the choice of informant alters the results
control, and conflict were taken from parent and of research is poorly understood. In this article, I
child interviews in the longitudinal National Survey compare the perspectives of parents and their chil-
of Families and Households. Parents gave more dren who are making the transition to adulthood.
positive reports than their adult child on six of the Paired data from the longitudinal National Survey
eight relationship indicators where parent and child of Families and Households (NSFH) are used to
answered identical questions. Parents were espe- estimate differences between the reports of parents
cially likely to report higher levels of closeness. and young adult children on emotional closeness,
Three patterns of dyadic agreement were identified: contact, control, and conflict in their relationship.
high agreement (54%), parent more positive than The intent is not only to describe differences, but
child (25%), and child more positive than parent to test multivariate models predicting the circum-
(21%). Despite these differences in perspective, stances under which divergent perspectives are
regression models predicting intergenerational most likely to occur and to assess the influence of
closeness and conflict were nearly invariant across
divergent perspectives on research outcomes.
the parent and child data.

RATIONALE
The choice of informant for data collection has
become an increasingly critical issue in research onA recent study of preadolescent children and their
intergenerational relationships. There have been rel- parents suggests that the point of view reflected in
atively few studies of systematic differences in the self-report data is an important aspect of research
design. Tein et al. (1994) found little evidence of
cross-generational, convergent validity on five
subscales of the Child's Report of Parental Behav-
Department of Child & Family Studies, University of Wisconsin,
ior Inventory (CRPBI). Correlations between par-
1430 Linden Drive, Madison, WI 53706 (aquilino@facstaff.
wisc.edu). ent and child reports were quite low, ranging from
.13 to .36 (with most in the .20-.25 range). Level of
Key Words: intergenerational relations, measurement validity, agreement was linked to child's age (lower parent-
multiple informants. child agreement among older children), family size

858 Journal of Marriage and the Family 61 (November 1999): 858-870

This content downloaded from


103.59.198.83 on Mon, 28 Aug 2023 19:50:10 +00:00
All use subject to https://about.jstor.org/terms
Two Views of One Relationship 859

(less agreement in families used. Amato et al. with


and Aquilino three
relied on parents'
or m
children), and child's well-being reports of support exchange, but Cooney &
(lower Uhlen-
agreeme
among more depressed berg, White, and UmbersonSimilarly,
children). used adult children's cor
lations between college reports. This conclusion is based
freshmen and on speculation,
their not pare
on subscales of the CRPBI were also found to be on direct comparison of multiple informants.
quite low (Schwarz, Barton-Henry, & Pruzinsky, Rossi and Rossi (1990) conducted one of the few
studies that has examined how the results of predic-
1985) with correlations between mothers and chil-
tive models vary, depending on whose perspective is
dren ranging from .30 to .40 and between fathers
and children from .19 to .29. Tein et al. concluded used. Their research included interviews with multi-

that researchers should not aggregate across the ple generations of the family, including the adult
reports of multiple family members but should con- children and the parents of the primary respondents.
sider conducting a separate analysis for each fam- They found that young adults' retrospective ac-
ily informant. counts of family cohesion, parental affection, and
The generational stake theory (Acock & Bengt- emotional closeness when they were adolescents
son, 1980; Bengtson & Kuypers, 1971) may help in were positively related to current affective closeness
understanding why the reports of parents and young with parents. However, Rossi and Rossi found evi-
adult children differ. This theory emphasizes the dence of continuity only in the reports of the adult
different psychological needs of the two genera- child. With retrospective data from the parents' per-
tions who are at contrasting points of the life cycle. spective, they found little association between early
Youth tend to emphasize conflict with parents and family life and current relations with adult children.
exaggerate differences in order to achieve a clearer They concluded that parents are more apt than
sense of emancipation and to facilitate separation grown children to view past family relationships
from the family of origin. "Each generation has an through a rosy lens and to see current relations with
investment in the generational bond. But, for youth, children as less contingent on the past. In Aquilino's
the 'stake' is more toward maximizing a sense of (1997) study of parent-adult child relationships
separate identity; for parents, the investment pays based on NSFH data, however, considerable evi-
off in maximizing continuity" (Acock & Bengtson, dence of continuity in relationships was generated
p. 512). Young adult children may have little moti- with data from the parents' perspective, using
vation to put a positive spin on relationships with prospective rather than retrospective data.
parents. In contrast, if parents strive to maintain in- Results of a cross-cultural comparison (German
tergenerational continuity, they may be more moti- and American) of parent-adolescent relationships
vated than adult children to present a picture of (Barber, Chadwick, & Oerter, 1992) are at odds
strong intergenerational ties and to avoid revealing with the Rossi and Rossi (1990) conclusion. Barber
problematic aspects of the relationship. Evidence and his colleagues examined linkages between
that supports this theory has been found with parental behaviors and adolescent psychological
matched pairs of parents and adolescents (Noller, well-being (self-esteem, self-derogation, self-worth).
Seth-Smith, Bouma, & Schweitzer, 1992), parents The results they obtained when using adolescent
and college students (Thompson, Clark, & Gunn, perceptions of parental behavior could be replicated
1985), and middle-aged children and their parents (in large part) using the reports of parental behavior
(Bond & Harvey, 1991). given by the parents. Barber et al. consistently found
In research on intergenerational relations, will significant correlations between parent-reported
substantive results differ, depending on whether the parental behaviors and the well-being outcomes re-
parent's or the adult child's reports are used? There ported by the adolescents.
is little empirical evidence relevant to this question. In sum, the few empirical studies of multiple
In a study of the impact of parental divorce on the family informants do not yield a clear picture of
exchange of help between adult children and par- differences in response tendencies between the
ents, Amato, Rezac, and Booth (1995) reported that generations in a family and provide conflicting evi-
their main finding (that support exchange was not dence of the impact of the informant on research
adversely affected by divorce) was at odds with outcomes. My purpose in this article is to provide a
several prior studies (e.g., studies of Cooney & more complete assessment of the magnitude and
Uhlenberg, 1990; Umberson, 1992; White, 1992), implications of differences between parents and
but was consistent with the results reported by young adult children in reports of the quality of the
Aquilino (1994). Amato and his colleagues attrib- intergenerational relationship. The analysis focuses
uted this pattern to differences in the informants on the following questions:

This content downloaded from


103.59.198.83 on Mon, 28 Aug 2023 19:50:10 +00:00
All use subject to https://about.jstor.org/terms
860 Journal of Marriage and the Family

Question 1: What are the mean-level


the null differ-
hypothesis that the effects of predic-
ences and the magnitudetorsof
on relationship outcomes would be in-
the correlations
variantand
between parents' ratings across theyoung
sources of data. adult
children's ratings of emotional closeness,
shared activities, conflict, and control?
METHODS Gen-
erational stake theory suggests that parents
Data are taken from the quality
will report higher relationship first and second waves
and of
the NSFH conducted
fewer problems than will adult children. in 1987-1988 and 1992-1993.
The NSFH collected detailed family life interviews
from 13,007of
Question 2: What patterns respondents drawn from a multistage
parent-child
area probability
agreement or disagreement can sample of the United States. See
be identified
Sweet, Bumpass, and Call
across the set of intergenerational (1988) for a full descrip-
measures?
tion of the
This analysis goes beyond survey. The screening
looking at aggre-response rate at the
gate, mean differences first wave was 91%, and
between 74% of eligible
parents respondents
and
adult children and moves to a dyadic
were successfully interviewed.level
At Time 2,ofthe study
analysis based on matched pairs
design called (parent
for a reinterview of all and
primary respon-
adult child). I expected dents interviewed at Time
to identify 1, as well
three dis-as additional
tinct patterns: dyads ininterviews
which with the primary
the parentrespondent's
re- spouse or
partner, a child of the respondent
ported more positive relationships than the aged 10 to 24
child, dyads in which the
years, and achild reported
parent of the respondent. The follow-
up interviews
more positive relationships took place
than the in 1992 and 1993.
parent,
The NSFH subsample
and dyads with high agreement betweenused in this analysis in-
par-
cludesset
ent and child across the the 18-
ofto 24-year-old
relationship children who were
quality variables. interviewed at Time 2 and their parents who were
interviewed at both times. Parents were the pri-
mary influence
Question 3: What factors NSFH respondents theat Timepat-
1. The children
werebetween
tern of dyadic agreement 12-18 years oldparent
at Time 1, and
and were the
child? This involvedfocal
developing anThat
children in the interview. ex-
is, at Time 1,
ploratory, multivariate
parentsmodel predicting
reported extensively on their parenting of
and relations with
which pattern of agreement this particular child. All focal
a parent-child
children lived with
dyad would fall into. Predictors the parent at par-
included Time 1. At Time
ent and child demographic 2 the parentscharacteristics,
were reinterviewed and asked an ex-
family background factors, current
tensive series of questions aboutliving
their current rela-
arrangements and role tions with the focal child. The focal
constellations ofchildbothinter-
parent and child, and measures of parent- view at Time 2 also contained a subset of identical
child relations taken at Time
questions about the 1quality
(when chil- with
of the relationship
dren were adolescents).each parent. About 82% of the eligible parents and
62% of the young adult children were successfully
Question 4: To what degree interviewed at and in what
Time 2, resulting in a sample size
manner does the choice of informant affect of 1,090 parent-child dyads. Missing data reduced
the results of multivariate models predicting the sample size for testing mean differences be-
parent-child relationship quality? The first tween groups on individual items to 1,062. In mul-
step in answering this question involved de- tivariate models, missing data further reduced the
veloping and testing an invariant factor struc- analytic sample to 1,027 cases. The likelihood of a
ture model for parent and child reports of re- parent-child dyad being in the sample at Time 2
lationship quality. I hypothesized that two was lower when the original respondent (the par-
factors would emerge from the data whether ent) was male, Black, or Hispanic, and had fewer
measured from the child's perspective or years of education; when the family structure at
from the parent's perspective: a warmth- Time I was single parent or stepparent; and when
closeness factor and a conflict-control factor. the focal child was male and was older. There was
In the second step, structural equation model- little difference between respondents and nonre-
ing was used to compare multivariate models spondents on the Time 1 parenting measures
based on children's reports of the relation- (warmth-support, democratic discipline, coercive
ship and based on parents' reports. I tested control and conflict).

This content downloaded from


103.59.198.83 on Mon, 28 Aug 2023 19:50:10 +00:00
All use subject to https://about.jstor.org/terms
Two Views of One Relationship 861

Sample Characteristics
habiting; most of the rem
or dating. Thirteen percen
The average age of parents at Time
lationship. 2 was
Most of 47 y
the
84% were between the ages of 40 and 59. Ave
(70%). About 1 in 7 was
age of the children was 21 years.
for work. One There are
fifth of th
twice as many mothers (70%) as fathers (30%
become parents. Nearly h
the sample. This occurs
sidebecause
in theirhouseholds
parent's houw
children who were not living with both biolog
parents were double-sampled at Time 1, resulting
Measures
many more female-headed households than wo
be expected in a typical
Wordingprobability
of items and response scales sample.
for the mea- C
dren who were not living with both biological
sures of parent-child relationship quality used in
ents were much morethis likely
analysis is toshownbe in theliving with
Appendix. Eight mea- t
mother than with their father. Average
sures were constructed from items that were iden- educat
attainment of parents was
tical across 13 and
the parent years: 14%
child interviews: global had
finished high school, relationship
43% were high school
quality, humor-closeness, tension in gr
ates, 24% had completed 1-3
the relationship, years
extent of activities,
of shared leisure college,
19% were college graduates. Dueparental
parental need for control, to disapproval
the overs of
pling of families withchild's decisions, extent of open disagreements, par
an absent biological
there are a large number of single-parent
and the frequency of fights and arguments in the and s
parent families. About half of the focal child
relationship.
were living in families with two biological pare
at Time 1. Thirty-one percent were living
ANALYSIS AND RESULTS
in sin
parent families and about 18% in stepparent f
lies. In two-thirds of the interviews in steppa
Mean Level Comparisons
families, the biological parent was interviewed,
in one third the stepparent was ininterviewed.
The first question addressed the analysis was the
children were evenly generational
split stake by sex.
effect, Nearly
the notion 40% w
that parents will
enrolled in some type of
report school,
a more positive view ofmost in re-
intergenerational colleg
small percentage in vocational or adult
lationships than their young technical
children. Paired t scho
Among those not enrolled in on
tests were computed school,
each of the eight25%
relation- had
ceived some form of shippostsecondary
measures. The results support thiseducation,
hypothesis,
21% were high school graduates
especially for warmth-closeness only. Seven
(Table 1). Parents
cent were high school gave
drop-outs.
significantly higher ratings
Twenty-eight
than adult children p
cent of the children were currently
on all four of married
the closeness indicators and reported or

TABLE 1. MEANS, CORRELATIONS, AND PAIRED t TEST RESULTS FOR RELATIONSHIP VARIABLES

Parents' Child's
Variable M (SD) M (SD) Correlation

Global relationship quality 8.27 8.09** .39


(1.69) (1.86)
Easy to laugh and have a good time together 4.38 4.23*** .26
(.77) (.82)
Feel on edge or tense when together 1.61 1.97*** .18
(.91) (1.00)
Frequency of leisure activities together 3.65 3.35*** .39
(1.36) (1.39)
Parent would like more influence over child's life and decisions 2.36 2.66*** .20
(1.15) (1.23)
Extent of parental disapproval of child's decisions and performance 2.38 1.98*** .31
(.93) (.79)
Average frequency of open disagreements between parent and child 2.00 1.67*** .43
(2.26) (1.95)
Frequency of arguing, fighting, having difficult time with each other 1.69 1.82** .28
(1.05) (.97)

Note: Data from parents and adult children, National Survey of Families & Households
**p < .01. ***p < .001.

This content downloaded from


103.59.198.83 on Mon, 28 Aug 2023 19:50:10 +00:00
All use subject to https://about.jstor.org/terms
862 Journal of Marriage and the Family

higher relationship quality,though positive and


more significant
ease and(p humor
< .01), were quite
in
the relationship, less tension,
low, ranging from and more
.18 to .43. shared
The magnitude of the
leisure activities than did correlations
their is consistent with other
offspring. studiesmag-
The of
nitude of the mean differences is small to moder- agreement between parent and child ratings (Tein
ate, ranging from .20 to .40 standard deviations. et al., 1994; Schwarz et al., 1985) and suggests that
The paired comparisons are in line with genera- the perspectives of parents and adult children may
tional stake theory on only two of the four control- be quite different in many dyads.
conflict indicators (Table 1). As hypothesized,
young adult children were significantly more likely
Patterns of Parent-Child Agreement
than parents to report that parents wished more in-
fluence over the child's life. Children also reportedThe second step in the analysis was to identify
more frequent arguments and fights. On two of the patterns of parent-child agreement across the set
control-conflict indicators, however, results were of relationship measures. First, difference scores
in the opposite direction. In ratings of parental dis-were computed for each of the eight relationship
approval of the child's decisions, parents gave sig- measures by subtracting the child's score from the
nificantly higher ratings than did the young adult parent's score in each dyad. Thus, difference
children. This is the largest mean difference (aboutscores greater than zero mean that the parent gave
.50 standard deviations). This suggests that adult a higher rating than the child, difference scores
children may not be fully aware of how their par- lower than zero mean that the child gave a higher
ents feel about many aspects of the child's life andrating than the parent, and difference scores close
that, in many cases, parents keep feelings of disap-to zero indicate high agreement. Cluster analysis
proval to themselves. Parents also reported signifi-then was used with the set of eight distributions of
cantly more open disagreements than did the difference scores. Three clusters were specified for
young adult children. Although this variable wasthe solution. Cases with missing values on more
supposed to measure overt, observable behavior, itthan two of the eight indicators were dropped from
is possible that parents took the opportunity in thethe analysis, reducing the sample size to 1,027
survey to mention areas of disagreement that theydyads. The tension variable was reverse-coded to
do not openly discuss with their offspring. Children be consistent in direction with the other three
may assume that parental silence on an issue, such closeness variables.
as how the child dresses or spends money, indi- The difference score means for the dyads as-
cates approval or acceptance. signed to the three clusters are shown in Table 2.
In sum, the results are consistent with genera- Cluster 1 reflects the high agreement pattern. The
tional stake theory. At the aggregate level, parentsmean difference scores for dyads in this cluster
described the intergenerational relationship in sig- are close to zero or at least closer to zero than the
nificantly more positive ways than did children. Inmeans for the other two clusters, indicating that
only two instances did adult children give moreparent and child gave similar ratings across the
positive descriptions than parents. Correlations be- set of eight relationship indicators. This was the
tween the parent and child ratings (Table 1), al- largest cluster, with 556 of the 1,027 dyads (54%).

TABLE 2. RESULTS OF CLUSTER ANALYSIS BASED ON PARENT-CHILD DIFFERENCE SCORES

Cluster 2 Cluster 3
Cluster 1 Parent More Positive Child More Positive
Difference Score High Agreement Than Child Than Parent
Global relationship quality -.00 2.40 -2.09
Ability to laugh, have a good time together .09 .76 -.41
Tension in relationship (reverse coded) .32 1.14 -.41
Shared leisure activities .19 .85 -.08
Parent wants more influence over child -.36 -.91 .62
Parental disapproval of child's decisions .10 -.51 .31
Extent of open disagreements -.03 -.44 .67
Extent of fights, arguing, difficulties in relationship -.19 -.67 .67
n 556 256 215

Note: Means for


parents had hig
close to zero ind

This content downloaded from


103.59.198.83 on Mon, 28 Aug 2023 19:50:10 +00:00
All use subject to https://about.jstor.org/terms
Two Views of One Relationship 863

This cluster does not contain need for control. Oneonly dyads
of the surprising outcomesreport
of
high quality relationships; this research wasdyads
the size of this in
cluster. which
It included par
and child agree that215 the relationship
of the 1,027 dyads, or 21% of the sample.is poor
mediocre also fall into this cluster. In summary, slightly more than half (54%) of
Cluster 2 represents dyads in which parents the parent-child dyads fall into the high agreement
clearly provided more favorable ratings of the cluster. The remaining 46% of the dyads are char-
parent-child relationship than did the children. This acterized by higher levels of disagreement between
was the second largest cluster, with 256 of the the parent and child reports. Although the majority
1,027 dyads (25%). Among the dyads in this clus- of these dyads involve parents giving more posi-
ter, the parent gave higher ratings than the adult tive ratings (25%), there are many dyads (21%) in
child on all four warmth-closeness indicators and which it is the adult child who provides the rosier
view of the intergenerational relationship.
lower ratings than the adult child on all four control-
conflict indicators.
Cluster 3 represents a pattern that has received
Predicting Cluster Membership
little attention: dyads in which the adult child pro-
vides more favorable ratings of the parent-child re-What factors influence the propensity of dyads to
lationship than the parent does. This group is notexhibit a particular pattern of agreement or dis-
apparent when looking only at aggregate differences agreement on intergenerational relations? The next
across groups. Among these dyads, the adult childstep in the analysis was to develop a predictive
gave higher ratings of global relationship quality model of cluster membership and to identify the
and ease and humor in the relationship and reportedindividual and family characteristics that might
lower levels of tension. Members of these dyads
differentiate the high agreement, parent-more-
favorable, and child-more-favorable patterns of
did not differ in rating the frequency of social inter-
action. The difference score mean was close to zero. dyadic agreement. The dependent variable in this
The adult child also reported less control-conflict analysis, cluster membership, is a triadic, categori-
than did the parent on all four measures, indicatingcal variable. Thus, multinomial logistic regression
fewer arguments and fights, fewer open disagree- was the most appropriate modeling technique
ments, less parental disapproval, and less parental (Greene, 1992). Multinomial logistic regression is

TABLE 3. MULTINOMIAL LOGIT MODEL PREDICTING CLUSTER MEMBERSHIP

Cluster 2 Cluster 3
Parent More Positive Child More Positive
Predictor Than Child Than Parent

Parent education (vs. high school gradu


Less than high school -.28 -.82**
Some college .15 .17
College graduate .09 .48t
Parent's family status, Time 1 (vs. b
Single parent .10 .18
Remarried parent .14 .39
Stepparent 1.00** .58
Child's education or enrollmen
College graduate .18 -1.31**
Enrolled in college .19 -.78**
Enrolled in technical or high school .28 -.19
Some college or technical school .27 -.24
Less than high school .44 -.13
Child's current union status (vs. going stea
Married -.21 -.82**
Cohabiting .20 .28
Dating .42* .01
No relationship .25 .09
Child's parental status (1 = has child) .35 .44t
Child's religiosity -.09t -.07
Parent's democratic discipline, Time 1 -.06 -.29**
Coercive control-conflict, Time 1 -.20't.43**

tp<.10. *p<.05. **p < .01. ***p < .001.

This content downloaded from


103.59.198.83 on Mon, 28 Aug 2023 19:50:10 +00:00
All use subject to https://about.jstor.org/terms
864 Journal of Marriage and the Family

in thelogit
an extension of the binary parent-more-favorable
model cluster and the more
to situations
likely it was
where the dependent variable has to be in the high
three or agreement
more group.
cat-
egories. The multinomial Therelogit
may be less model
divergence insimultane-
perspective be-
tween generations
ously contrasts the likelihood of when children and
a dyad parents in
being
share strong religious
Cluster 2 (parent more favorable) values. Families
versus Clusterin which 1
(high agreement) and the likelihood
parents reported of
higher levels of a dyad
coercive control
being in Cluster 3 (child more
(arguing, yelling,favorable)
disagreements) at Time versus
1 when
Cluster 1 (high agreement). The
children were estimates
adolescents were also lessin
likelyeach
to
equation control for the bepropensity of the
in the parent-more-favorable dyad
category. Prior re-to
fall into the cluster not included in the contrast. search with these data (Aquilino, 1997) has shown
There is no equation directly contrasting Cluster that
2 there is continuity in parental assessments of
to Cluster 3 because those parameters are a mathe- the parent-child relationship from adolescence to
matical function of the first two equations. The re-young adulthood. Parents who experienced diffi-
sults of the multinomial analysis are shown inculty with their adolescent children are less likely
Table 3. Predictor variables that had no significantto report a rosy picture of the relationship when the
effects in either equation were dropped from thechildren have become young adults.
final model. These included parent's age, sex, There were more significant predictors in the
race-ethnicity, income at Time 1, and union transi-
second equation that contrasted the child-more-
tions between Time 1 and Time 2; the child's age,favorable cluster to the high agreement group
sex, employment status, and coresidence in the(Table 3). Parent's education has a linear relation-
parental household; and parent-child closeness ship to the probability of a dyad displaying the
and parental restrictiveness at Time 1 when thechild-more-favorable pattern. The more educated
child was an adolescent (taken from parent reportsthe parent, the more likely the dyad is to show this
in the original interview). pattern. Dyads in which the parent had less than a
There were relatively few significant predictors
high school education were least likely to fall into
of the tendency of dyads to fall into the parent-this cluster. There is evidence that more highly ed-
more-favorable pattern than into the high-agreementucated parents at midlife are more likely than less
pattern. The strongest finding was that stepparent-educated parents to focus on opportunities for self-
stepchild dyads were more likely to have this pat- development and autonomy (Aquilino & Supple,
tern than dyads involving biological parents and1991; Schnaiberg & Goldenberg, 1989) and are less
children, regardless of whether the biological parentsanguine about continued coresidence and other
was still married to the other biological parent, re-forms of support for young adult children. Thus,
married, or a single parent. Adult stepchildren de- more educated parents may present a less idealis-
scribed a less favorable view of the relationship than
tic view of intergenerational relationships, which
did their stepparents. Dyads in which the adultmay increase the likelihood that an adult child could
child was dating (as opposed to going steady, the provide a more favorable report of the relationship
omitted category) were also significantly more than the parent.
likely to be in the parent-more-favorable cluster. The adult child's education and enrollment sta-
The origins of this effect are not clear. The period tus was also significantly related to the probability
of romantic involvement before the child makes a
of falling in Cluster 3, but in a direction opposite to
commitment may be a time of increased tension for the effect of parental education. Dyads in which the
the young adult child. Parents may apply pressureadult children were college graduates or college
to move toward commitment or may communicatestudents were less likely to fall into the child-more-
their reservations about the child's potential partner. favorable cluster than dyads in which the child had
Children likely would find these behaviors aver-lower levels of education. The more educated chil-
sive, and parents may see them as a positive con-dren may take a more critical, evaluative stance vis-
tribution to the child's welfare. Thus the dating a-vis their families of origin. Children's perception
situation might detract from the child's experienceof similarity to parents may decrease as their at-
of relationship quality but not from the parent's. tainment increases, especially if children acquire
Two other predictors reached trend levels of sig-more education than their parents.
nificance (p < .10), child's religiosity and parent's Dyads in which the adult child was married
coercive control at Time 1. The greater the child's were significantly less likely to fall into the child-
involvement in religion (frequency of attending re- more-favorable category. This effects likely stems
ligious services), the less likely the dyad was to befrom the fact that parents' views of intergenerational

This content downloaded from


103.59.198.83 on Mon, 28 Aug 2023 19:50:10 +00:00
All use subject to https://about.jstor.org/terms
Two Views of One Relationship 865

closeness become more positive


offspring. after an
The parent-more-favorable adult
pattern was
has married (Aquilino,most 1997), making
likely in stepparent-stepchild itand
dyads less
when li
that the child would have
adult more
children were positive
still dating view
rather than going
the relationship than steady.
the parent. The marginally
nificant effect for the adult child becoming a
ent is in the opposite direction. Impact of the Informant onThese dyad
more likely to be in the Predicting
child-more-favorable
Relationship Quality
ter than in the high agreement group. This is
consistent with priorTheresearch
final phase of the analysisthattakes on the most
shows that
ents experience more critical question: Are the differences
difficulty with in perspective
adult child
after their offspringbetween
have parent and made
child large enough
ato relatively
produce e
divergent results
transition to parenthood in models predicting relationship
(Aquilino, 1997).
Finally, there were quality? Confirmatory factor analysis was effects
significant used to fo
parental behavior at reduce
Time the number1. of dependent
Democratic variables and to discip
was negatively construct
related to identical outcome variables
Cluster 3from the
members
and coercive control parent and
was child data. Empiricalrelated
positively literature on the
tofac- fa
in Cluster tor structure of scales
democratic3. Less measuring relationships
discipline (not be- al
tween parents and younger
ing the child to participate children was usedmak
in decision as a
and higher levels ofguide in developing a hypothesized
coercive control factor sugges
structure
authoritarian style of for the parent childdiscipline.
parental relationship measuresThe
used in re
this research. Factorwere
suggest that when children analyses of raised
parent-child rela-
wit
tions parent-child
authoritarian style, the inventories have yielded adyad
consistent was
set of m
likely to fall into thedimensions describing the parent-child relation-pat
child-more-favorable
of ship (Barnes & Farrell,
agreement. One possibility is that 1992; Barber
adultet al. child
1992;
Robin, Koepke,
orientation toward parents is& more
Moye, 1990;positive
Simons, Beaman,in
ilies that have a Conger,
more & Chao, 1993; Tein et al.,
conservative, 1994). Most
family v
bent. Children's desire to distance themselves from measures can be subsumed within two dimensions
parents and family may be less urgent in this type (Amato, 1990): warmth-support and control. Fre-
of setting. quently used indicators of support include expres-
In summary, only a few clear findings emerge sion of affection, emotional closeness, parental
from this attempt to predict the agreement pat- acceptance, open communication, help with prob-
lems, and parental involvement. The dimension of
terns of the parent-child dyads in this sample. The
child-more-favorable pattern is most likely to parental control encompasses a broad array of par-
occur when parents have high education and chil- enting behaviors including punishing, yelling and
dren have low educational attainment and when scolding, and asserting more control versus allow-
ing autonomy (Demo, Small, & Savin-Williams,
the family is characterized by conservative, reli-
gious, and more authoritarian values. The child's1987).
marriage decreases the likelihood of the child-In the confirmatory factor analysis, I hypothe-
more-favorable pattern by increasing parents' sized that the eight relationship measures in this
research would define two latent dimensions char-
propensity to have a positive view of relations with

TABLE 4. AN INVARIANT FACTOR MODEL FOR PARENT AND CHILD RATINGS OF THEIR RELATIONSHIP

Factor 1 Factor 2
Variable Closeness Control-Conflict

Global relationship quality .76 .00


Ability to laugh, have a good time together .72 .00
Tension in relationship (reverse coded) .63 .00
Shared leisure activities .55 .40
Parent wants more influence over child .00 .50
Parental disapproval of child's decisions .00 .70
Extent of open disagreements .00 .62
Extent of fights, arguing, difficulties in relationship .00 .38

Note: This is the completely standardized solution fitted in LISREL8. Factor loadings were co
across parent and child ratings. The factor intercorrelation for the parent data was -.57, and for th
chi-square with 43 degrees of freedom was 304.7 (p < .001); goodness-of-fit index was .97.

This content downloaded from


103.59.198.83 on Mon, 28 Aug 2023 19:50:10 +00:00
All use subject to https://about.jstor.org/terms
866 Journal of Marriage and the Family

closeness
acterizing the parent-child dimension only, its highest
relationship, loading.
closeness
and control. The closeness factor would consist of This was done to make the factor scores as distinct
as possible from each other. The four regression
global relationship quality, ease-humor, low tension,
and shared leisure activities. The control factor
models (a parent and child model for each of the
would consist of parental desire for influence,
two dimensions) were fitted simultaneously in
parental disapproval of child, frequency of open
LISREL-8. There were four dependent variables:
disagreements, and frequency of fights and argu-
the parent's rating and the child's rating of close-
ments. The hypothesized model specified two fac-
ness, and the parent's rating and the child's rating
tors, with each indicator loading on one factor only.
of control-conflict. Predictors included a set of par-
Loadings of the closeness indicators on the control
ent characteristics (age, sex, race-ethnicity, educa-
factor were constrained to be zero, as were the con-
tion, income, family type, and union transitions be-
trol indicators' loadings on the closeness factor. The
tween waves from the Time 1 and Time 2 parent
model specified an invariant factor solution across
interviews), a set of child characteristics (age, sex,
the parent and child data (factor loadings con- education-enrollment, union status, employment,
strained to be equal across informants). Results parental
of status, coresidence status, and religiosity
fitting this model in LISREL-8 showed that one from the child interview at Time 2), and measures
change in model specification was needed to obtain
of parental behavior when the child was an adoles-
a satisfactory solution. Shared leisure activities
cent (support, democratic discipline, restrictiveness,
loaded significantly on the control-conflict factor,
and coercive control from the parent interview at
as well as on the closeness factor. The loadingTime
is 1).
positive, suggesting that families with higher lev-The initial model tested the null hypothesis that
els of conflict and control issues also tend to have the regression parameters for all predictors would
higher levels of social interaction (and that parentsbe identical for the parent and child ratings on each
and children with low levels of contact are also
dependent variable. The two sets of beta coeffi-
likely to report low levels of conflict). The final
cients in the parent and child models predicting
model shown in Table 4 was identical for the par-
closeness were constrained to be equal. The same
was done for the coefficients in the two control-
ent and child data, and all loadings were constrained
to be equal across the groups. This invariant fac-
conflict equations. The fit of the totally invariant
tor model provided an acceptably good fit to model
the was reasonably good (chi-square with 70 de-
data. The goodness-of-fit index was .97. Levels grees of freedom of 132.3, p = 0.00001; adjusted
below .95 indicate that the model needs improve-goodness-of-fit index of 0.93). However, models
ment. The chi-square is significant (p < .001), with
in- a goodness-of-fit index below .95 can usually
dicating that the invariant model does not fit the
be improved by relaxing constraints on the model
data perfectly. The chi-square test for goodness-of-
(Bollen, 1989). Inspection of the LISREL modifica-
fit is quite sensitive to sample size (Bollen, 1989).
tion indices showed that freeing only a few of the
With large samples, it is often significant, even
equality parameters would greatly improve the
when the model fits the data reasonably well. The
overall fit of the model. In the final model, freeing
two factors are significantly correlated (-.57 the
for regression parameters on just 4 of the 40 predic-
parents, -.53 for children). The success of confir-
tors (20 independent variables in each model) im-
matory factor analysis in finding an invariant proved
so- the fit substantially, yielding a model chi-
lution suggests that the measures do have similar
square of 76.1 with 66 degrees of freedom (p = .19)
psychometric properties across the two groups andof an adjusted goodness-of-fit index of .96. Thus,
informants and provide equally valid measuresdespite
of the differences in parent and child perspec-
the intergenerational relationship for both groups.
tives on the intergenerational relationship, as shown
The final step in the analysis explored the by the cluster analysis, a nearly invariant predictive
question of whether the parent data and child data
model fit the data for parents and children well.
would yield identical results in regression modelsThe results of the final regression models are
predicting parent-child relationship outcomes. The
shown in Table 5. In the models predicting close-
closeness and control-conflict scores were com- ness and control-conflict, effects of the parent
puted by summing the standardized scores for each characteristics are completely invariant across the
of the variables that loaded on each factor. Factor parent and child ratings. Both the parent's and
weights were not used in this computation. Al- child's ratings of closeness were higher for African
though shared leisure activities had a significant Americans and lower when parents were college
loading on both factors, it was included on the graduates, single parents, stepparents, and when

This content downloaded from


103.59.198.83 on Mon, 28 Aug 2023 19:50:10 +00:00
All use subject to https://about.jstor.org/terms
Two Views of One Relationship 867

TABLE 5. REGRESSION MODELS PREDICTING PAR


OF CLOSENESS AND CONTROL-CONFLICT IN RELATIONSHIP

Closeness Ratings Control-Conflict Ratings


Predictor Parent Child Parent Child

Parent characteristics
Age .01 .01 .01 .01
Sex (1 = male) .01 .01 -.12*** -.12**
Race-ethnicity (vs. White)
Black .08** .08** .02 .02
Hispanic -.04 -.04 .06* .06*
Education (vs. high school graduate)
Less than high school .04 .04 -.07* -.07*
Some college -.01 -.01 .02 .02
College graduate -.09** -.09** .03 .03
Family income, Time 1 -.04 -.04 .02 .02
Family income missing -.01 -.01 .00 .00
Parent's family status, Time 1 (vs. biological parent in
Single parent -.06t -.06t .02 .02
Remarried parent .03 .03 -.01 -.01
Stepparent -.15*** -.15"** -.01 -.01
Parent's union transitions between Time 1 and Time 2 (v
New marriage or cohabitation -.01 -.01 -.02 -.02
Separated-divorced -.09*** -.09*** .05* .05*
Child characteristics
Age .00 .00 -.12*** -.12***
Sex (1 = male) -.01 -.01 .02 .02
Employment status (vs. employed)
Unemployed -.04 -.04 .05* .05*
Not in labor force -.00 -.00 -.01 -.01
Education or enrollment status (vs. high sc
College graduate .01 .01 -.05t -.05t
Enrolled in college .05 .05 -.11"** -.11***
Enrolled in technical or high school .03 .03 -.05t -.05t
Some college or technical school .00 .00 .03 .03
Less than high school .04 .04 .07** .07**
Current union status (vs. going steady)
Married .09** .09** -.16*** -.16***
Cohabiting -.01 -.01 .03 .03
Dating -.04 -.04 .05t .05t
No relationship .01 .01 -.05* -.05*
Parental status (1 = has a child) -.04 -.04 .11*** .11***
Coresides with parent .11*** .02 .17*** .17***
Religiosity .07** .07** -.06* -.06*
Parental behavior, Time 1 (parent report)
Support-closeness .20*** .10** -.01 -.01
Democratic discipline .09** .09** -.05* -.05*
Restrictiveness -.07* -.07* .04 .04
Coercive control-conflict -.23*** -.04 .26*** .15***
R2 .24 .12 .28 .22

Note: n = 1,027 parent-


and child groups. The
dom (p = .185); adjusted
tp<.10. *p<.05. **

parents had separa


(Table 5) w
first and second
parent in
and
conflict were
only highe
one p
parent tus on warmth-closeness-needed to be freed toor
separated im- d
surement, and
prove the fit were
of the model. When parents and chil-
ents with less
dren lived together, than
parents reported significantly
regression coefficien
higher levels of closeness to children than when

This content downloaded from


103.59.198.83 on Mon, 28 Aug 2023 19:50:10 +00:00
All use subject to https://about.jstor.org/terms
868 Journal of Marriage and the Family

they lived apart. In contrast, coresidence


tween the two closeness status
models is for the parent's
had no impact on adultreportchildren's
of coercive controlperceptions
at Time 1. This had a of
highly
closeness to their parents. Thissignificant,
pattern negative iseffect in the parent
consistent
with previous studies (Aquilino, 1997;
model but no significant effect inAquilino
the child model. &
Supple, 1991) and showsIt appears,
thatthen, parents
that in studyingreport
the long-term posi-
ef-
fects coresident
tive experiences with their of parent-adolescent conflict,
adult the children
results may
and feel less close to children who have left home. vary depending on whose perspective is used.
Adult children do not share this perspective. In the models predicting current control-conflict,
The effects of all other child characteristics on invariant betas were found for three of the four
the two relationship dimensions are invariant parental behavior predictors. In both models, Time
across the parent and child data. Ratings of close- 1 support and Time 1 restrictiveness had no impact
ness were higher when the adult child was marriedon Time 2 control-conflict, and democratic disci-
and when the child reported high religiosity. The pline had a negative effect in both models. Time 1
marriage effect is consistent with the hypothesis coercive control had a positive effect on Time 2
(Aquilino, 1997; Bengtson & Black, 1973) that in- control-conflict in both models. However, the size
tergenerational relations become more positive asof the effect was significantly larger for parents.
children move into adult roles and their lives be- In summary, the regression models predicting
come more similar to the lives of their parents. Allcurrent relationship quality between parents and
eight of the child characteristic variables were adult children were close to invariant across the
significantly related to control-conflict, and the two informants. Except for the effects of coresi-
betas were invariant across the two groups. Control-dence on warmth-closeness, all of the effects of
conflict ratings were negatively related to the child'sparent characteristics and child characteristics were
age, education, and religiosity. Control-conflict rat-identical in the parent and child models. The most
ings were also lower when the child was married or discrepant findings across informants were for the
had no current relationship than when the child waslong-term influence of parental coercive control at
going steady and were higher when the son or Time 1 on closeness and control-conflict when the
daughter was unemployed and looking for work, children were young adults. This suggests that
had made the transition to parenthood, or still livedthere is some disparity in the perspectives of par-
with the parent. ents and adolescents on control issues and that both
The final set of predictors in these equations
reports may be needed to understand the control di-
were the reports of parenting behavior when themension more fully.
child was an adolescent (taken from the Time 1 in-
terview with parents; see Aquilino, 1997, for a fuller CONCLUSIONS
description of these indicators). Although three of
the eight parameters needed to be freed from the The central question explored in this research was:
equality constraint to improve the fit of the modelIn research on family relationships, does it matter
(Table 5), the most noteworthy finding from this which family member provides the self-report data
part of the analysis is the degree of consistency in
on relationship quality, contact, and interaction
findings across the models for parents and chil- patterns? Results suggest that the answer is yes.
dren. In the three instances in which the beta coef- There are systematic differences in the perspec-
ficients differed across models, the effect was tives of parents and adult children on the nature
stronger for parents than for children. This pattern and quality of their relationships. At the aggregate
likely reflects the contribution of method variance. level, a sample of parents is likely to provide an
The parent models have the same informant for overall rosier view of intergenerational relation-
predictor and outcome, but the child models have ships than is a sample of adult children. This is
different informants for predictor and outcome. consistent with the generational stake theory
In the models predicting warmth-closeness, the (Acock & Bengtson, 1980) and with the findings
positive effect of democratic discipline and the of Rossi and Rossi (1990). High agreement be-
negative effect of parental restrictiveness are in- tween parent and adult child from the same family
variant across the models for parents and children. characterized only about half of the dyads in the
Parental support at Time 1 is positively related to NSFH sample. One of the most important findings
closeness at Time 2 in both models, although the of this research, however, is that when disagree-
effect is significantly stronger in the parent model ment occurs, it is not always in the direction of
than in the child model. The largest difference be- parents giving the more positive report. There are

This content downloaded from


103.59.198.83 on Mon, 28 Aug 2023 19:50:10 +00:00
All use subject to https://about.jstor.org/terms
Two Views of One Relationship 869

many cases in which to estimate


the the contribution
adult of method
child variance expres
to
more positive view than the results ofthe parent.
models based This pa
on single informants.
appears to be especially Althoughlikely
the research described
when in this article
parents
high educational attainment provides some insightsand children
into the subjective experi- hav
educational attainment and when children were ence of closeness and conflict in relationships be-
raised in more conservative, religious, authoritar-
tween parents and young adult children, there are a
ian families. number of other indicators that need a more thor-
To what extent does the choice of informant ough review. These include parent and child reports
have the potential to alter the results of research
of help exchange, as well as subjective perceptions
aimed at predicting the quality of parent-adult child
of equity in exchange; perceptions of the role perfor-
relations? The number of parent-child dyads char-mance of self and other (e.g., the older generation's
acterized by substantial disagreement suggests performance of the grandparent role, the younger
that results might vary greatly depending on who generation's performance in the parent role, and
provides the data. Nonetheless, I was able to fitbotha generations' performances in spousal roles).
Research contrasting parents' and children's retro-
factor model for the parent-child relationship indi-
cators that was invariant over the parent and child
spective accounts of earlier family life would help
ratings and that fit the data reasonably well. Fur-
greatly in understanding the construction and re-
construction of family memories. There is much to
ther, in predicting relationship outcomes, regression
models for the parent and child reports of closeness
be learned from viewing the divergent perspectives
of individuals in relationship as a substantive
and control were nearly invariant. This suggests that
it is not a foregone conclusion that different fam-
question in its own right, rather than simply error
variance.
ily informants will always yield different results in
models predicting relationship outcomes. The poten-
tial for informant variance may depend on the na- REFERENCES
ture of the variables used in the analysis. In this
Acock, A., & Bengtson, V. (1980). Socialization and attri-
study, discrepant results in predicting current close-
bution processes: Actual versus perceived similarity
ness and control were found for one aspect of ear-
among parents and youth. Journal of Marriage and the
lier parent-child relations-coercive control and
Family, 42, 501-515.
conflict when the child was an adolescent. This Amato,
sug- P. (1990). Dimensions of the family environ-
gests that children's and parents' perspectives onment as perceived by children: A multidimensional
scaling analysis. Journal of Marriage and the Family,
control issues and conflict may be different and that
52, 613-620.
the divergent views may influence the outcome of
Amato, P., Rezac, S., & Booth, A. (1995). Helping be-
research on these dimensions. Regression models tween parents and young adult offspring: The role of
also showed that estimating the effect of home- parental marital quality, divorce, and remarriage.
Journal of Marriage and the Family, 57, 363-374.
leaving versus continued coresidence on the qual-
Aquilino, W. (1994). Impact of childhood family disrup-
ity of the parent-child relationship depends on thetion on young adults' relationships with parents.
informant. Coresidence exerted a strong influence Journal of Marriage and the Family, 56, 295-313.
Aquilino, W. (1997). From adolescent to young adult: A
on parents' experience of closeness to adult children
prospective study of parent-child relations during the
but had no impact on adult children's feelings of
transition to adulthood. Journal of Marriage and the
closeness to parents. In assessing the effects of de-
Family, 59, 670-686.
mographic characteristics, family type, role transi-
Aquilino, W., & Supple, K. (1991). Parent-child rela-
tions and parental satisfaction with living arrange-
tions, and religiosity on relationship outcomes, data
ments when adult children live at home. Journal of
from parents and data from children yielded iden-
Marriage and the Family, 53, 13-27.
tical results.
Barber, B., Chadwick, B., & Oerter, R. (1992). Parental
In conclusion, theoretical and empirical scholar-
behaviors and adolescent self-esteem in the United
ship on intergenerational relations would benefit States and Germany. Journal of Marriage and the
Family, 54, 128-141.
from more attention to the issue of divergent per-
Barnes, G., & Farrell, M. (1992). Parental support and
spectives in families. I believe there is a strong case
control as predictors of adolescent drinking, delin-
for making the collection of equivalent data from quency, and related problem behaviors. Journal of
both generations a standard practice in designing re-Marriage and the Family, 54, 763-776.
Bengtson, V. L., & Black, K. D. (1973). Intergenera-
search on parents and adult children. The robustness
tional relations and continuities in socialization. In
of predictive models across family informants at
P. Baltes & W. Schaie (Eds.), Life-span developmen-
different life stages needs to be demonstrated em-tal psychology: Personality and socialization (pp.
pirically. Such designs would also allow researchers
207-294). New York: Academic Press.

This content downloaded from


103.59.198.83 on Mon, 28 Aug 2023 19:50:10 +00:00
All use subject to https://about.jstor.org/terms
870 Journal of Marriage and the Family

Bengtson, V. L., & Kuypers,Schwarz,


J. A. J. C., Barton-Henry,
(1971). M. L., & Pruzins
Generational
(1985). Assessing
differences and the developmental child-rearing
stake. Agingbehaviors:
and A com
Human Development, 2, 249-260.
ison of ratings made by mother, father, child an
Bollen, K. (1989). Structuralling on the CRPBI.
equations Child Development,
with latent vari- 56, 462
ables. New York: Wiley & Sons.
Simons, R., Beaman, J., Conger, R., & Chao, W.
Bond, J. B., & Harvey, C. (1991). Ethnicity
Childhood experience, and inter-
conceptions of parentin
generational perceptions ofattitudes
family of spouse as determinants
solidarity. of parental b
Interna-
tional Journal of Aging andior. Human
Journal of Development,
Marriage and the Family,
33, 55, 91
33-44. Sweet, J., Bumpass, L., & Call, V. (1988). The
Cooney, T., & Uhlenberg, P. (1990). The role of divorce inand content of the National Survey of Familie
men's relations with their adult children after midlife.Households (Working Paper NSFH-1). Madison:
Journal of Marriage and the Family, 52, 677-688. versity of Wisconsin, Center for Demography
Demo, D., Small, S., & Savin-Williams, R. (1987). Fam- Ecology.
ily relations and the self-esteem of adolescents and Tein, J., Roosa, M., & Michaels, M. (1994). Agreement
their parents. Journal of Marriage and the Family, 49, between parent and child reports of parental behaviors.
705-715. Journal of Marriage and the Family, 56, 341-355.
Greene, W. (1992). LIMDEP version 6: Users manualThompson, L., Clark, K., & Gunn, W. (1985). Develop-
and reference guide. Bellport, NY: Econometric Soft- mental stage and perceptions of intergenerational
ware. continuity. Journal of Marriage and the Family, 47,
913-920.
Noller, P., Seth-Smith, M., Bouma, R., & Schweitzer, R.
(1992). Parent and adolescent perceptionsUmberson,
of family D. (1992). Relationships between adult chil-
functioning: A comparison of clinic and dren nonclinic
and their parents: Psychological consequences
for both generations. Journal of Marriage and the
families. Journal ofAdolescence, 15, 101-114.
Family, 54, 664-674.
Robin, A., Koepke, T., & Moye, A. (1990). Multidimen-
White, L. K.
sional assessment of parent-adolescent relations. (1992). The effect of parental divorce and
Psy-
chological Assessment, 2, 451-459. remarriage on parental support for adult children.
Rossi, A., & Rossi, P. (1990). Of human bonding:
Journal
Parent-
of Family Issues, 13, 234-250.
child relations across the life course. New York: Aldine
de Gruyter.
Schnaiberg, A., & Goldenberg, S. (1989). From empty
nest to crowded nest: The dynamics of incompletely
launched young adults. Social Problems, 36, 251-269.

APPENDIX

ITEM WORDING, RESPONSE SCALES, AND COMPUTATION OF SCORES FOR PARENT-CHILD RELATIONS VARIABLES

Warmth-closeness items

1. Global relationship quality: Taking things all together, on a scale from 0 to 10, where 0 is really bad and 10 is absolutely
perfect, how would you describe your relationship with your (parent/child)?
2. Humor/closeness: It's easy for me to laugh and have a good time with my (parent/child). (1 = strongly disagree, to 5 =
strongly agree.)
3. Tension: I feel on edge or tense when I'm with my (parent/child). (1 = strongly disagree, to 5 = strongly agree.)
4. Shared activities: Over the last 3 months, about how often have you spent time with your (parent/child) in leisure activi-
ties, working on something together, or just having private talks? (1 = not at all, 2 = less than once a month, 3 = one to
three times a month, 4 = about once a week, 5 = more than once a week.)

Control-conflict items

5. Parental control/influence: (I/my mother/my father) would like more influence over (my/my child's) decisions. (1 =
strongly disagree, to 5 = strongly agree.)
6. Parental disapproval: Average rating over six items. Inapplicable items were excluded from the computation. Each item
was coded 1 = extremely happy, 2 = somewhat happy, 3 = neither happy nor unhappy, 4 = somewhat unhappy, 5 = ex-
tremely unhappy. How (do you/does your parent) feel about: how well you've done in school? the (boyfriend-girlfriend)
you're going steady with? your decision to get married at the age you did? your choice of a (husband/wife)? the occupa-
tion or career that you want?
7. Extent of open disagreements: Computed as the prorated number of the items answered "yes." In the last 3 months, have
you and your (parent/child) had open disagreements about: dress; dating; friends; getting a job or a better job; sexual be-
havior; drinking, smoking, or drug use; money; helping around the house; how late you stay out at night; your (hus-
band/wife/partner); raising your children?
8. Fights, arguments: During the last 3 months, how often did you argue or fight or have a lot of difficulty with your (par-
ent/child)? (1 = not at all, to 5 = more than once a week).

Note: From the National Survey of Families & Households, Time 2 parent and youth interviews.

This content downloaded from


103.59.198.83 on Mon, 28 Aug 2023 19:50:10 +00:00
All use subject to https://about.jstor.org/terms

You might also like