Вы находитесь на странице: 1из 11

PERSONALITY AND SOCIAL PSYCHOLOGY BULLETIN

Granqvist / ATTACHMENT AND RELIGIOSITY IN ADOLESCENCE

Attachment and Religiosity in Adolescence:


Cross-Sectional and Longitudinal Evaluations

Pehr Granqvist
Uppsala University

The study offers the first investigation of cross-sectional and lon- ration of the environment (Kirkpatrick, 1992, 1994,
gitudinal associations between attachment and religiosity with 1995, 1999). Hence, the attachment aspects of religios-
adolescent participants. Adolescence is a life period linked to ity, just as the child’s attachment to parents, serve the
attachment transitions and religious changes. The research was ontogenetic function of obtaining/maintaining felt
conducted to help resolve inconsistent results from previous cross- security (see, e.g., Sroufe & Waters, 1977).
sectional versus longitudinal studies and studies on attachment To examine the empirical applicability of attachment
to parents versus peer attachment. Time 1 questionnaire data theory to adults’ religiosity, two opposing hypotheses
was collected from 196 Swedish adolescents (M age = 16 years); were derived from attachment theory (Kirkpatrick,
143 participants completed the 15-month follow-up. Results 1992; Kirkpatrick & Shaver, 1990). The original corre-
from the cross-sectional analyses generally supported the social- spondence hypothesis was based on Bowlby’s (1969,
ized correspondence and emotional compensation hypotheses, 1973, 1980) discussion of continuity in inner working
particularly in relation to attachment to mother. Results from the model (IWM) functioning as underlying temporal conti-
longitudinal analyses were more mixed. The discussion inte- nuity in the individual’s relationship orientation. As
grates the hypotheses with Kirkpatrick’s previous proposal, applied to religion, security of attachment was hence
argues for a main focus on attachment to parents rather than proposed to constitute the foundation on which subse-
peer attachment, and offers suggestions for future attachment quent religiosity was built. In contrast, based on
and religion studies. Ainsworth’s (1985) discussion of insecurely attached
persons’ need for surrogate attachment figures, the
compensation hypothesis stated that insecure attach-
Attachment theory, as formulated by Bowlby (1969, ment was the foundation for subsequent religiosity (i.e.,
promoting the use of God as a compensatory attachment
1973, 1980) and extended by Ainsworth (Ainsworth,
Blehar, Waters, & Wall, 1978) and others, is an empiri- figure).
cally well-corroborated framework that describes norma- Retrospective studies of attachment to parents have
tive aspects and individual differences of child-parent found insecure attachment to be linked to significant
relations and their socioemotional correlates in subse- religious changes and sudden religious conversions, rep-
quent development (see, e.g., Bretherton, 1985, 1987, resenting attachment main effects in support of the com-
1991; Cassidy & Shaver, 1999; Colin, 1996). The theory is pensation hypothesis (Granqvist, 1998; Granqvist &
one of the leading relationship-oriented paradigms in Hagekull, 1999, 2001; Kirkpatrick & Shaver, 1990). Com-
developmental, personality, and social psychology, as
Author’s Note: Thanks to Birgitta Brandhofer, Niklas Dahlquist, and
well as in the psychology of religion. Paula Grinde for assistance with the data collections. Special thanks to
anonymous reviewers and to my advisor, Berit Hagekull, for many good
Attachment and Religion: Empirical Complexities
suggestions on previous drafts of this article. The research presented
Some believers’ relationships with God bear striking herein was supported by a Grant (Dnr 1999-0507:01,02) to Professor
Berit Hagekull from the Bank of Sweden Tercentenary Foundation.
resemblances to the child’s attachment to caregivers and
Communication regarding this article should be addressed to Pehr
appear to meet the most important criteria for defining Granqvist, Department of Psychology, Uppsala University, P.O. Box
attachment relationships, including seeking closeness to 1225, SE-751 42, Uppsala, Sweden; e-mail: pehr.granqvist@psyk.uu.se.
God in prayer and through rituals and using God as a PSPB, Vol. 28 No. 2, February 2002 260-270
safe haven during distress and as a secure base for explo- © 2002 by the Society for Personality and Social Psychology, Inc.

260
Granqvist / ATTACHMENT AND RELIGIOSITY IN ADOLESCENCE 261

pensation effects from attachment to parents have been Granqvist and


obtained on other variables as well (e.g., belief in a per- Hagekull (1999) Kirkpatrick (1998b, 1999)
sonal God, personal relationship with God), but these
effects have been moderated by parental religiousness. Secure Socialized correspondence: Concurrent correspondence:
attachment Religiosity reflecting social Correspondence between
Insecure participants have had higher religiousness at
learning of parent’s mental models of self/
low levels of parental religiousness (Granqvist, 1998; religious standards others and God at a
Kirkpatrick & Shaver, 1990). given time
However, secure participants have had higher reli-
giousness at high levels of parental religiousness, thus Insecure Emotional compensation: Longitudinal compensation:
attachment Religiosity reflecting Religiosity tied to
supporting the original correspondence hypothesis
affect regulation strategy attachment system
(Granqvist, 1998). Another way to describe these inter- to obtain/maintain felt dynamics; prone to
actions is to say that parental and offspring religiousness security increases over time
is positively linked in secure dyads and independent in
Figure 1 Description of two attempts to clarify previous inconsistent
insecure dyads (Granqvist, 1998; Kirkpatrick & Shaver, results on attachment and religiosity.
1990).
Cross-sectional studies of peer attachment have
shown secure attachment to be linked to higher reli- corresponds to parental religiosity rather than to the
giousness than insecure attachment, hence supporting security of the relationship per se. The compensation
the correspondence hypothesis (Granqvist & Hagekull, hypothesis was slightly revised (Granqvist & Hagekull,
2000; Hyde, 2000; Kirkpatrick, 1998b; Kirkpatrick & 1999) in that a mechanism of affect regulation, in search
Shaver, 1992). The effect sizes of these findings have of felt security, was proposed to underlie the insecure
been small (typically, d < .20), and level of parental (or individual’s relationship with God (“emotional compen-
some other attachment figure’s) religiousness has not sation”). These revisions primarily attempted to resolve
been considered in the statistical analyses. In contrast, the findings from the studies of attachment to parents
longitudinal studies of peer attachment have found rather than from the peer attachment studies. (See Fig-
ambivalent/preoccupied and fearful-avoidant attach- ure 1 for a schematic description of Kirkpatrick’s and
ment to predict increased religiosity over time our attempts to clarify previous inconsistencies in
(Kirkpatrick, 1997, 1998b), thus supporting the compen- results.)
sation hypothesis. In the latter study, however, dismissing/ Our revised hypotheses have recently received empir-
avoidance was linked to decreased religiosity. ical support (Granqvist & Hagekull, 1999, 2001). The
socialized correspondence hypothesis was supported in
Two Attempts to Clarify the Complexities
that perceived security of parental attachment was asso-
The above complexities in results obtained in studies ciated with a newly constructed Socialization-Based Reli-
on attachment to parents versus peer attachment and giosity Scale (SBRS) that was designed to tap degree of
studies using cross-sectional versus longitudinal designs transmission of parental religious standards to the off-
have yielded two different attempts at clarification. spring. In addition, security was linked to non-
Kirkpatrick (1998b, 1999) suggested that the discrepant conversions and gradual increases in religiosity, as
findings from cross-sectional versus longitudinal studies opposed to sudden conversions, the latter of which are
highlight how the correspondence effects are more thought by many researchers to reflect an extreme
immediate and are due to corresponding mental models attempt to resolve emotional turmoil (see Hood, Spilka,
of self/others and God at a specific time (henceforward Hunsberger, & Gorsuch, 1996). The increases in religi-
labeled “concurrent correspondence”). In contrast, as osity of relatively secure offspring were characterized by
the compensation effects are delayed, it was hypothe- early onset and occurred during life situations pointing
sized that the religiosity of insecure people is more tied to the importance of socialization of significant others’
to attachment system dynamics (henceforward referred religious standards (“themes of correspondence,” e.g.,
to as “longitudinal compensation”). close friendships with believers, membership in reli-
We have instead argued for the need to reformulate gious organizations). Life themes of emotional turmoil,
the original hypotheses (see Granqvist, 1998; Granqvist & such as relationship problems with parents and peers,
Hagekull, 1999). The revised correspondence hypothe- personal crises, mental or physical illness (“themes of
sis states that the religiosity of individuals in secure compensation”), were not characteristic of these reli-
dyads, via social learning, stems from the socialization of gious changes. The emotional compensation hypothesis
religious standards held by the primary attachment fig- was supported in that insecurity was associated with a
ures in childhood (“socialized correspondence”). The newly constructed Emotionally Based Religiosity Scale
individual’s religiosity in secure dyads thus presumably (EBRS) that was designed to tap affect regulation strate-
262 PERSONALITY AND SOCIAL PSYCHOLOGY BULLETIN

gies with direct reference to the individual’s relationship home, also is tied to adolescence (Roof & Hadaway,
with God. In addition, insecurity was linked to sudden 1979; Roof & McKinney, 1987), making this period of
religious conversions and to increases in religiosity, char- attachment transition ideal for studying increases as well
acterized by suddennness/intenseness, late onset, and as decreases in religiosity.
themes of compensation, whereas themes of correspon-
The Present Study
dence were uncharacteristic.
It is an open question if the socialized correspon- The present study is the first investigation of cross-
dence and emotional compensation hypotheses are sectional and longitudinal associations between attach-
applicable to the links between peer attachment and reli- ment and religiosity using participants in adolescence.
giosity in adolescents. Also, no study has examined the The study sought to clarify previous inconsistencies in
relative ability of attachment to parents and peer attach- the literature and theory as well as to examine the rela-
ment to predict unique variance in religiosity. tive ability of parental and peer attachment to explain
unique variance in religiosity.
Attachment and Religion in Adolescence Predictions were derived from our socialized corre-
The work on attachment and religion has relied on spondence and emotional compensation hypotheses.
adult samples or, in one case (Kirkpatrick, 1998b), a Hence, for cross-sectional relations, the emotional com-
young adult sample. It is important also to examine ado- pensation hypothesis predicted that insecurity of attach-
lescents, for several reasons. First, adolescence repre- ment to parents and peer attachment at Time 1 (T1)
sents a life period during which transition of attachment would be related to religiosity as based on affect regula-
takes place, where attachment components are gradu- tion (emotionally based religiosity) and religious fluctu-
ally transferred from parents to peers (e.g., Allen & ations (both increases and decreases). For cross-sectional
Land, 1999; Fraley & Davis, 1997; Hazan & Zeifman, and longitudinal relations, and for participants who had
1999). Although parents remain important in fostering experienced increased religiosity, either prior to T1 or
continuity of adolescent adaptation, adolescents show between T1 and Time 2 (T2), insecurity was expected to
an increasing autonomy vis-à-vis parents (e.g., Armsden be linked to religious changes marked by suddenness
& Greenberg, 1987; Lapsley, Rice, & Fitzgerald, 1990; and intenseness and occurring during life situations of
Moore, 1987; Palladino Schultheiss & Blustein, 1994; emotional turmoil (themes of compensation).
Papini & Roggman, 1992; Rice, 1990). This scenario is The socialized correspondence hypothesis predicted
normative but some (particularly insecure) individuals that security at T1 would be linked to religiosity as based
are less likely to build close, trusting, and satisfactory on social learning of parental religiosity (socialization-
peer relations (e.g., Lapsley et al., 1990; Palladino based religiosity) and religious stability (i.e., little fluctu-
Schultheiss & Blustein, 1994; Rice, 1990), implying that ation in religiosity). Insofar as increased religiosity had
felt security cannot be derived either by turning to par- been experienced, either prior to T1 or between T1 and
ents or peers for support. The attachment transitional T2, security was expected to be linked to gradual reli-
period is sometimes also marked by increased conflicts gious changes occurring during life situations pointing
with parents and by personal uncertainties on behalf of to the importance of socialization of significant others’
the adolescent (e.g., Allen & Land, 1999). Again, how- religious standards (themes of correspondence). Due to
ever, security of attachment is associated with more its exploratory nature, no predictions were formulated
favorable outcomes than insecurity (Allen & Land, 1999; for the question concerning the relative ability of attach-
Kobak, Cole, Ferenz-Gillies, & Fleming, 1993; Kobak & ment to parents and peer attachment to explain unique
Ferenz-Gillies, 1995; Kobak, Ferenz-Gillies, Everhart, & variance in religiosity.
Seabrook, 1994).
Major religious changes and sudden religious conver- METHOD
sions appear to be overrepresented in this period of
Participants and Procedure
attachment transition (e.g., James, 1902; Starbuck, 1899;
see also Hood et al., 1996). Such experiences, unlike The sample consisted of 196 adolescents in Stock-
more gradual and less intense religious changes, are holm, the capital of Sweden. Sweden is a highly secular-
often linked to problematic life situations, such as rela- ized country. Even though approximately 90% of the
tionship problems within or outside the family, personal Swedish population are members of the Lutheran
crises, mental/physical illness, bereavement, and expe- Church of Sweden, less than 10% of the population are
riences of relationship break-ups (e.g., Granqvist, 1998; confessing Christians (Pettersson, 1994). To obtain vari-
Granqvist & Hagekull, 1999; James, 1902; Kirkpatrick & ation on the religiousness variables, part of the sample
Shaver, 1990). However, apostasy, or the often temporary was drawn from a religious population. The sample
decline of religiosity among those raised in a religious included adolescents from the Christian youth organiza-
Granqvist / ATTACHMENT AND RELIGIOSITY IN ADOLESCENCE 263

tion of the Lutheran Church of Sweden (n = 82) and TABLE 1: Measurement Characteristics, Descriptive Statistics, and
Swedish upper secondary school classes (n = 114). There Homogeneity Evaluations
were 37% male respondents at T1 (34% at T2) and the
Number
mean age of the sample at T1 was 16.3 years (SD = 1.2, of Scale Mean Alpha
range = 14-19) (corresponding figures at T2: M = 17.6, Variable Items Range (SD) Coefficient
SD = 1.2, range = 15-20).
At T1, the data collections were made at the end of EBRS 10 1-6 2.35 (1.41) .96
SBRS
lectures or at organization gatherings during the early
Maternal items 10 1-6 4.02 (1.40) .91
fall of 1998. Participants were (also at T2) informed of Paternal items 10 1-6 3.94 (1.38) .89
the voluntary and anonymous premises for participa- Increased religiosity 1 1-6 T1: 2.88 (1.85) —
tion. They also were informed about the possibility of a T2: 2.06 (1.42) —
longitudinal follow-up. The study was presented as a Suddenness/intenseness
of change 1 1-3 T1: 1.80 (.72) —
study on intergenerational similarities and differences
T2: 2.00 (.71) —
with respect to religiosity and attitudes. Questionnaires Themes of compensation 5 1-6 T1: 2.23 (1.10) T1: .75
were filled out during approximately 20 mins and then T2: 2.38 (1.21) T2: .75
handed in to the investigator. A total of 6 adolescents, Themes of correspondence 3 1-6 T1: 2.91 (1.60) T1: .73
present at the occasions of data gathering, chose not to T2: 2.75 (1.16) T2: .43
Decreased religiosity 1 1-6 T1: 1.94 (1.47) —
participate, yielding a response rate of 97% at T1.
T2: 1.95 (1.48) —
Because Swedish youths are inclined to move around Insecurity of attachment
internationally during late adolescence, the short-term to mother 13 1-9 2.27 (1.06) .84
longitudinal follow-up was conducted only 15 months Insecurity of attachment
after T1 to maintain a high response rate at T2. At T2, to father 13 1-9 3.23 (1.90) .94
Peer avoidance 9 1-9 3.90 (1.55) .82
questionnaires, including a prestamped response enve-
Peer anxiety 9 1-9 3.29 (1.44) .81
lope, were mailed home to the participants during the
winter of 1999. They were then offered a cinema check NOTE: EBRS = Emotionally Based Religiosity Scale, SBRS = Socialization-
worth approximately U.S.$6 for participation. A Based Religiosity Scale, T1 = Time 1, T2 = Time 2.
reminder was sent 2 weeks later, and after another 2
weeks, a second reminder (including questionnaire and
potentiate an examination of the empirical structure of
response envelope) was sent out, yielding a final
the instrument, evaluate its homogeneity, and increase
response rate of 72% (N = 142). An attrition analysis was
its measurement precision (cf. Fraley & Waller, 1998;
performed on all variables included. Out of a total of 20
Simpson, Rholes, & Nelligan, 1992).
significance tests, only 1 reached significance—mater-
To examine the empirical structure of these mea-
nal religiousness was higher among T2 participants than
sures, two factor analyses (one per parent), with Varimax
in the attrition group (p < .05).
rotation, were undertaken. Using eigenvalue > 1 as crite-
Measurements and Statistical Analyses rion, these analyses yielded four factors for the maternal
items and one factor for the paternal items. Upon rota-
Questionnaire data containing the following vari- tion, the first maternal factor accounted for 63% of the
ables were utilized in the present study (unless noted, explained variance and 24% of the total variance
items were couched as statements, where a low value (eigenvalue = 4.70; eigenvalues for the remaining factors
indicates strong disagreement and a high value indicates were < 1.35). It was difficult to find theoretically mean-
strong agreement; see Table 1 for measurement charac- ingful contents in the factors. In the unrotated solution,
teristics and descriptive statistics). 10 of the 13 items loaded highest on the first factor. To
Attachment measures. Retrospective ratings of 13 items achieve comparability across attachment targets, a one-
were used to assess perceived maternal and paternal factor solution was therefore chosen for both parents
attachment, respectively. These items were drawn from (total variance accounted for in the maternal and pater-
the three retrospective attachment paragraphs pro- nal solutions = 36% and 58%, respectively). Items were
posed by Hazan and Shaver (1986). Hazan and Shaver’s reversed where appropriate to yield global Insecurity of
measure was intended to tap characteristic parenting Attachment to Mother and Father scales (see above for
behaviors in avoidant (e.g., “she was generally distant”), sample items), where low values denote perceptions of
secure (e.g., “he knew when to be supportive”), and sensitive and responsive parenting and high values
ambivalent (e.g., “she was noticeably inconsistent in her denote distant, rejecting, and inconsistent parenting.
reactions to me, sometimes warm and sometimes not”) Collins and Read’s (1990) Adult Attachment Scale
parent-child dyads. The paragraphs were transformed to (AAS) was used to assess peer attachment. At the time of
a continuous multi-item measure in the present study to T1 measurement in early 1998, Brennan, Clark, and
264 PERSONALITY AND SOCIAL PSYCHOLOGY BULLETIN

Shaver’s (1998) recently constructed and since then Characteristics of religious change were assessed by
widely used Experiences in Close Relationships Scale having participants in the increased religiosity groups
(ECR) was not available to the author. However, when choose one of the following alternatives to describe the
subjected to factor analysis, the AAS and ECR have been Suddenness/Intenseness of Change: (a) “A slow, grad-
shown to yield relatively similar factor solutions. Three ual change over a long period of time”; (b) “A slow, grad-
factors (labeled Depend, Close, and Anxiety by Collins ual change with one or more relatively intense experi-
and Read) have been obtained in some studies, whereas ences and changes”; and (c)”An intense and sudden
other studies (Brennan et al., 1998; cf. Simpson et al., personal experience.” This measure was taken from
1992) have supported a two-factor solution, with factors Kirkpatrick and Shaver (1990).
labeled Avoidance and Anxiety. Other characteristics of religious change were
A factor analysis (varimax rotation) was run on the assessed by having respondents in the increased religios-
AAS. Unlike results in previous studies, a four-factor ity groups mark Granqvist and Hagekull’s (1999)
solution was obtained in the present study (according to Themes of compensation and Themes of correspon-
the eigenvalue criterion). Upon rotation, the first two dence scales, describing empirically derived (Granqvist,
factors accounted for 51% of the explained variance and 1998) life factors associated with increased religiosity.
29% of the total variance (eigenvalues = 5.04 and 2.85, Themes of compensation consists of life factors pointing
respectively; eigenvalues for the remaining factors were to religion as filling a supportive role for a person in
< 1.40). In the unrotated solution, 16 of the 18 items emotional need: problem in love relationship or
loaded highest on one of the first two factors. A two- divorce, relationship problem within family, relation-
factor solution yielded dimensions highly comparable in ship problem with others, mental or physical illness, and
content to those of Simpson et al. (1992) and Brennan personal crisis. Themes of correspondence consists of
et al. (1998). In keeping with the current attachment three items concerning life factors pointing to the
research agenda, a two-factor solution, accounting for importance of significant others’ religious beliefs: close
44% of the total variance, was therefore chosen (items friendships with believers, meetings or discussions with
were reversed where appropriate). The factors were believers, and membership in religious youth
labeled Peer Avoidance (e.g., “Often, love partners want association.
me to be more intimate than I feel comfortable being,” “I At T1 and T2, Decreased Religiosity was assessed with
find it difficult to allow myself to depend on others”) and the following item: “I have experienced a change which
Peer Anxiety (“I often worry that my partner does not meant that religion became less important to me during
really love me,” “I am not sure that I can always depend a period of my life.”
on others to be there when I need them”).1 Partial correlations were utilized to test bivariate rela-
tions between each attachment and religiosity variable.
Religiousness measures. The EBRS (Granqvist &
Standard multiple regression analyses were run to exam-
Hagekull, 1999) was used to tap the affect regulating
ine the relative ability of attachment to parents and peer
functions of turning to and maintaining contact with
attachment to explain unique religiosity variance.2 In
God and religion to obtain/maintain felt security (e.g.,
view of some previous findings showing sex differences
“When I am under mental stress [e.g., during moments
in the associations between attachment and religiosity
of sadness or anxiety] I may feel an urgent need for
(Kirkpatrick, 1998b), the influence of sex was controlled
God’s support,” “I feel most content when I experience a
in all analyses.
close communion with God”).
The SBRS (Granqvist & Hagekull, 1999) was used to
RESULTS
tap the degree of participant adoption of parents’ reli-
gious standards/nonstandards (e.g., “I will probably Relations between the attachment dimensions were
give/I give my children an equally religious/nonreli- in most cases modest in strength (Mdn r = .27, range of rs =
gious upbringing as my mother gave me,” “Religion is .06 to .38). For the sake of describing results in relation to
equally important/unimportant to me as it was to my the socialized correspondence and emotional compen-
father during my childhood”). sation hypotheses, both of which make predictions in
At T1 and T2, Increased Religiosity was assessed with terms of security and insecurity, high scores on the four
the following statement: “I have experienced a change attachment dimensions will be referred to as high inse-
which meant that religion became more important to curity, with low values representing high security.
me during a period of my life” (Granqvist & Hagekull,
Cross-Sectional Relations Between
2000). This measure was adapted from Kirkpatrick and
Attachment and Religiousness at T1
Shaver (1990). Those who agreed with this statement
(i.e., scored > 3) were assigned to increased religiosity Partial correlations were first computed to test the
groups (ns = 78 at T1 and 23 at T2). predictions of concurrent relations between each of the
Granqvist / ATTACHMENT AND RELIGIOSITY IN ADOLESCENCE 265

TABLE 2: Time 1 Correlations Between Attachment and Religiousness Variables for the Total Sample and for the Increased Religiosity Group,
Controlling for Sex

Total Sample (N = 196) Increased Religiosity Group (n = 78)


Decreased Increased Suddenness/ Themes of Themes of
Attachment Variable EBRS SBRS a Religiosity Religiosity Intenseness Compensation Correspondence

Insecurity (mother) .16* –.24*** .08 .22*** .16 .45**** .19


Insecurity (father) .14 –.25**** –.04 .05 .05 .38**** .15
Peer avoidance .15* –.11 –.08 .09 .03 .09 .02
Peer anxiety .21*** –.23*** .05 .12 .02 .25* .11

NOTE: EBRS = Emotionally Based Religiosity Scale, SBRS = Socialization Based Religiosity Scale.
a. Denotes maternal items for correlations with maternal attachment, paternal items for correlations with paternal attachment, and combined pa-
rental items for correlations with peer attachment.
*p < .05. ***p < .005. ****p < . 001.

attachment and religiousness variables. Results from make independent contributions, ps > 10. The equation
these analyses are presented in Table 2, showing (left for SBRS was also significant, F(4, 171) = 5.66, p < .001,
panel) that insecurity of attachment to mother was with an independent contribution made by insecurity of
directly related to EBRS and inversely associated with attachment to mother, t = –2.57, β = –.21, p < .02, whereas
SBRS, which was in line with predictions. Also in line results for insecurity of attachment to father and peer
with predictions, insecurity of attachment to mother was avoidance failed to reach significance, ps > .10. Finally,
positively linked to increased religiosity as occurring the equation for themes of compensation was signifi-
prior to T1, but no association was obtained with cant, F(4, 87) = 8.73, p < .001. Insecurity of attachment to
decreased religiosity. Examining relations between inse- both mother and father made independent contribu-
curity of attachment to mother and characteristics of tions, ts = 3.44 and 2.51, βs = .35 and .25, ps < .001 and .02,
religious change among participants who had experi- respectively, whereas peer anxiety did not, p > .10.
enced increased religiosity prior to T1 (right panel) Longitudinal Predictions of
showed insecurity to be positively associated with themes Religious Change From T1 Attachment
of compensation but to be unrelated to suddenness/
intenseness of change and themes of correspondence. Partial correlations were first computed to test the
The direction of results for insecurity of attachment predictions of longitudinal relations between the attach-
to father was similar, although associations with EBRS ment and religious change variables. Results from these
and increased religiosity failed to reach significance. analyses are presented in Table 3, showing (left panel)
Results for peer attachment also pointed in the same that insecurity of attachment to mother was positively
direction, particularly for peer anxiety. Peer anxiety was linked to decreased religiosity, which was in line with pre-
directly linked to EBRS and themes of compensation, as dictions. However, insecurity of attachment to mother
well as inversely related to SBRS. Peer avoidance was pos- was unrelated to increased religiosity. Examining rela-
itively associated with EBRS but was unrelated to the tions between insecurity of attachment to mother and
remaining religiosity variables.3 characteristics of religious change among participants
Standard multiple regression analyses were run to who had experienced an increased religiosity between
analyze the ability of attachment to parents and peer T1 and T2 (right panel) showed that insecurity was posi-
attachment to explain unique variance in each of the tively linked to suddenness/intenseness of change,
religiosity variables that were significantly associated again in line with predictions. No significant associations
with at least two of the attachment dimensions, that is, on were obtained between insecurity and themes of com-
EBRS, SBRS (combined parental items), and themes of pensation or themes of correspondence. Also, insecurity
compensation. Only the attachment dimensions that of attachment to father and peer avoidance were unre-
displayed significant associations with the religiosity vari- lated to all religious change variables investigated.
ables in question (see Table 2) were included as predic- Peer anxiety was positively related only to decreased
tors in the regression equations. religiosity.4
The regression equation for EBRS was significant, A multiple regression analysis was run on decreased
F(4, 187) = 5.68, p < .001. However, only peer anxiety religiosity with insecurity of attachment to mother and
made an independent (marginally) significant contribu- peer anxiety as predictors. The equation was significant,
tion, t = 1.93, β = .15, p = .055, whereas insecurity of F(3, 133) = 4.83, p < .005. Insecurity of attachment to
attachment to mother and peer avoidance failed to mother explained unique variance in decreased religios-
266 PERSONALITY AND SOCIAL PSYCHOLOGY BULLETIN

TABLE 3: Correlations Between Attachment Variables at Time 1 and Aspects of Religious Change at Time 2 for the Total Sample and for the In-
creased Religiosity Group, Controlling for Sex

Total Sample (N = 143) Increased Religiosity Group (n = 23)


Decreased Increased Suddenness/ Themes of Themes of
Attachment Variable Religiosity Religiosity Intenseness Compensation Correspondence

Insecurity (mother) .26**** .11 .58** .20 .16


Insecurity (father) .04 .06 .06 –.19 .25
Peer avoidance .03 .04 –.14 –.08 .31
Peer anxiety .25**** .15 .07 .01 .18

**p < .01. ****p < .001.

ity, t = 2.18, β = .20, p < .05. Peer anxiety made a margin- Maccoby, 1984; Matas, Arend, & Sroufe, 1978; Richters &
ally significant contribution, t = 1.95, β = .17, p = .052. Waters, 1991).
The cross-sectional results showing that insecurity of
DISCUSSION attachment to parents is linked to emotionally based reli-
giosity, and that insecurity of attachment to mother is
Summary of the Findings related to increased religiosity, also replicate previous
Results of the present study were, as a whole, in line adult findings (Granqvist, 1998; Granqvist & Hagekull,
with our hypotheses. Supporting the socialized corre- 1999; Kirkpatrick & Shaver, 1990). Further support for
spondence predictions, security (i.e., low values on the the hypotheses emerged in the cross-sectional findings
insecurity dimensions) was linked to a religiosity based showing peer attachment anxiety to be negatively linked
on parental socialization and characterized by small fluc- to socialization-based religiosity and anxiety as well as
tuations. Supporting the emotional compensation avoidance to be positively linked to emotionally based
hypothesis, insecurity was related to an unstable religios- religiosity. These findings on emotionally based religios-
ity (both increases and decreases) based on affect regula- ity seem to be in opposition to Kirkpatrick’s (1998b,
tion. When increases in religiosity had been experi- 1999) suggestion of concurrent correspondence
enced, insecurity also was linked to sudden and intense between security and religiousness. However, insofar as
increases occurring during life situations of emotional the EBRS taps an affect regulating strategy of turning to
turmoil. These findings emerged particularly in the God, it could be that this strategy is more closely tied to
cross-sectional analyses, whereas results from the longi- attachment system dynamics than to IWMs. In that case,
tudinal analyses were more mixed. the links between emotionally based religiosity and inse-
Concerning the ability of attachment to parents and curity are in line with Kirkpatrick’s interpretation.
peer attachment to explain unique variance in religios- The emotional compensation hypothesis also was
ity, attachment to parents made six significant independ- supported by the prospective results showing that inse-
ent contributions, of which five were from attachment to curity of attachment to mother and peer attachment
mother. Peer attachment only made two marginally sig- anxiety are linked to decreased religiosity. These results
nificant independent contributions, both from anxiety. represent new findings, which are in opposition to
These results seem to highlight the relative importance Kirkpatrick’s interpretation of longitudinal compensa-
of attachment to parents, particularly to mother, in pre- tion. That interpretation was based, in part, on results
dicting adolescent religiosity. showing ambivalent and fearful attachment (i.e., con-
taining individuals scoring high on the anxiety dimen-
Socialized (vs. Concurrent) Correspondence
sion) (see Brennan et al., 1998) to predict increased reli-
and Emotional (vs. Longitudinal) Compensation
giosity over time. Our findings suggest that insecurity
The cross-sectional findings showing that insecurity may be related to a general instability in religiosity rather
of attachment to parents is associated with less socializa- than simply to prospective increases. Furthermore, the
tion-based religiosity replicate our previous findings on direction of religious change (increase or decrease) for
adults (Granqvist & Hagekull, 1999). These results sug- those high in insecurity may depend on developmental
gest that there are important individual differences with phase issues. The results suggest that adolescence repre-
respect to adolescent apostasy. Relatively secure off- sents a developmental period of general religious insta-
spring seems to be less likely to turn down the influence bility, whereas (early) adulthood is associated with
of parental standards, just as has been shown to be true increased religiousness (Granqvist, 1998; Granqvist &
for young children (e.g., Londerville & Main, 1981; Hagekull, 1999; Kirkpatrick, 1997, 1998b).
Granqvist / ATTACHMENT AND RELIGIOSITY IN ADOLESCENCE 267

The emotional compensation hypothesis further pro- tions due to current affect regulation needs (socialized
poses that increases in religiosity for relatively insecure correspondence). In addition, the IWMs derived from
individuals should be relatively intense and sudden and the attachment relationship subsequently affect the indi-
occur during life situations of emotional turmoil. The vidual’s image of a loving God (mental models corre-
cross-sectional relations between indices of insecurity spondence). Such an integrative hypothesis will be
and themes of compensation lent support to these prop- labeled “two-level correspondence.” The first level
ositions, as did the prospective link between insecurity of denotes a primary mechanism of social learning of
attachment to mother and suddenness/intenseness of parental standards in the context of a secure relation-
change. These results are also in line with Kirkpatrick’s ship and the second level a secondary effect reflecting
(1998b, 1999) suggestion of longitudinal compensation. mental models correspondence between self/other and
In fact, our and Kirkpatrick’s interpretations of the com- God.
pensation effects converge nicely in that both tie the reli-
Correspondence and Compensation in
giosity of insecure individuals to attachment system
Relation to Which Attachment Targets?
dynamics. However, our hypothesis has two advantages.
First, it adds affect regulation as the mechanism respon- In answering which attachment targets (i.e., parents
sible for the compensation effects. Second, it accounts and/or peers) that the correspondence and compensa-
for cross-sectional, not only longitudinal, relations tion hypotheses should refer to, there are empirical,
between insecurity and religiosity. attachment theoretical, and historical factors that need
Unlike Granqvist and Hagekull (1999), who found to be taken into account. Concerning empirical factors,
security of attachment to parents to be associated with the results of the present study seem to suggest that
themes of correspondence, the present study failed to attachment to parents, particularly to mother, should
find any such relations. These discrepant findings could constitute the primary frame of reference for the corre-
be an effect of developmental differences between ado- spondence and compensation hypotheses.
lescents and adults. Also note that the internal consis- The present results were based on participants in
tency of themes of correspondence was low at T2 in this midadolescence, representing a life period during
study. which attachment transition from parents to peers has
Insofar as adolescents high in security of attachment generally been initiated but not concluded (see Fraley &
to mother had experienced increased religiosity Davis, 1997; Hazan & Zeifman, 1999). Therefore, it is
between T1 and T2, these changes were gradual and possible that different results will emerge for older, adult
were not marked by themes of compensation. These samples. However, although results from previous stud-
results are in line with the socialized correspondence ies on adults have not made direct comparisons, the find-
hypothesis. Important to note here is that Kirkpatrick’s ings on attachment to parents (Granqvist, 1998;
suggestion of concurrent correspondence fails to make Granqvist & Hagekull, 1999, 2001; Kirkpatrick & Shaver,
any predictions concerning religious change for secure 1990) have generally tended to be stronger, in terms of
individuals. effect sizes, than those on peer attachment (Granqvist &
Comparing our socialized correspondence hypothe- Hagekull, 2000; Kirkpatrick, 1997, 1998b; Kirkpatrick &
sis with Kirkpatrick’s suggestion of concurrent corre- Shaver, 1992).
spondence leads to the conclusion that both hypotheses Concerning attachment theoretical factors, child-
are necessary. Using the concurrent correspondence parent relations represent a purer version of attachment
hypothesis in isolation fails not only to predict the reli- than do romantic relations. The former are asymmetri-
gious changes of secure individuals but also to account cal (i.e., the attachment figure is stronger and wiser) and
for the theoretically relevant moderating influence of have proximity/protection as the predictable outcomes,
parental religiousness (Granqvist, 1998; Kirkpatrick & whereas romantic relations are symmetrical and have
Shaver, 1990). Using the socialized correspondence reproduction as the most central outcome (see also
hypothesis in isolation fails to account for the links Kirkpatrick, 1998a). With respect to the individual’s rela-
observed between attachment security and some con- tionship with God, a reading of religious literature and
tent aspects of religiousness, such as a loving God image of research in the psychology of religion suggests that it
as opposed to, for instance, a distant God image (Hyde, resembles child-parent attachment more than romantic
2000; Kirkpatrick, 1998b; Kirkpatrick & Shaver, 1992). attachment. The believer-God relationship is asymmetri-
Therefore, it is henceforward hypothesized that the reli- cal (i.e., God is stronger and wiser) and is tied to a sense
giosity of secure individuals originates from social learn- of protection (see Hood et al., 1996) rather than to
ing and partial adoption of parental religious standards. reproduction.
Social learning in the context of secure attachment also Regarding historical factors, the attachment analysis
creates a religiosity that is not prone to intense fluctua- of religion was preceded by classic psychoanalytic writers
268 PERSONALITY AND SOCIAL PSYCHOLOGY BULLETIN

(e.g., Freud, 1927/1964), ego psychologists (e.g., Erik- To move the attachment and religion research further in
son, 1950), and object relationists (e.g., Rizzuto, 1979). a developmental direction, longitudinal research in
These theoreticians all delved into the functional and younger child populations should be conducted with
structural similarities between the individual’s child- measurement procedures such as the Separation Anxi-
hood relationships with the parents and his or her rela- ety Test (Main, Kaplan, & Cassidy, 1985).
tionship with God. Although the perspectives were based A complementary way of handling the potential
on a database of qualitative, biographical, and clinical biases of self-reports would be to study the effects on reli-
case material, marked by post hoc reconstructions of the giousness from experimental activation of participants’
individual’s past (see also Kirkpatrick, 1995), their main attachment systems (cf. Burris, Batson, Altstaedten, &
emphases on child-parent and adult-God relationships Stephens, 1994). Such studies may be revealing in that
converge with the above arguments. they may undermine avoidant/dismissing individuals’
Notwithstanding this preference for attachment to deactivating attachment strategies (e.g., Mikulincer,
parents, there may be indirect links between peer attach- 1998) as well as overcome limitations inherent in some
ment and religiosity. Such links are expected if attach- religiosity self-reports (e.g., Batson, Natifeh, & Pate,
ment orientation in peer relations is empirically linked 1978). They also may help solve the causal ambiguities
to attachment to parents. Relations between peer attach- that have resulted from the sole reliance on
ment and religiosity also may be expected if love rela- correlational designs. Such investigative efforts could
tionship experiences transform the individual’s IWMs, finally test hypothesized effects derived from the norma-
thereby affecting his or her way of relating to God (cf. tive attachment framework.
Kirkpatrick & Shaver, 1992). However, the opposite sce- Results from investigations of relevance to the norma-
nario is also possible, that is, that the relationship with tive attachment framework have been interpreted in
God transforms the IWMs, hence affecting the individ- attachment theory terms by Kirkpatrick (1992, 1994,
ual’s peer attachment orientation. 1995, 1999). Studies also should be conducted that
derive a priori attachment predictions, which are subse-
Methodological Directions and Future Studies quently subjected to empirical trials. One such cross-
The present study had the advantage over all previ- sectional study, on relationship status and religiousness,
ously published studies on religion and attachment to has currently been published (Granqvist & Hagekull,
parents to be based on a prospective longitudinal design 2000). That study needs to be complemented with pro-
instead of only cross-sectional comparisons. However, spective longitudinal investigations that permit within-
subject analyses addressing real-time effects on religios-
the short time span between T1 and T2 resulted in a low
ity from formation of new relationships, separation,
frequency of participants who had experienced
divorce, and the like.
increased religiosity between the measurement occa-
sions. Significant religious changes and conversions
have been shown to peak at ages 15 and 16 (see Hood et NOTES
al., 1996). A recommendation for future studies is there- 1. The Avoidance factor consisted of all of Collins and Read’s
fore to collect T1 data at an earlier age and to increase (1990) Close factor items as well as half of their Depend (D) factor
the time span between measurements. Future studies on items (1, 3, and 5; see their Table 2). The Anxiety factor consisted of all
of their corresponding Anxiety factor items as well as their D2, D4, and
attachment and religious change also should explore life D6.
themes associated with attachment in the context of 2. To avoid the problem of confounding within-group with
decreased religiosity. The emotional compensation between-group variance, all analyses to be presented controlled for the
influence of a dummy-coded subsample (i.e., members of Christian
hypothesis suggests that decreased religiosity occurs for youth organization vs. ordinary upper secondary school students).
insecure individuals in situations where their attach- Attachment made an independent and significant contribution in the
ment needs may have the potential to be met (e.g., dur- same analyses as when only sex was controlled (see Results), with the
exceptions that the cross-sectional links between the Emotionally
ing romantic relationship formation). In contrast, the Based Religiosity Scale (EBRS) on one hand and insecurity of attach-
IWM-level of the correspondence hypothesis suggests ment to mother and peer attachment avoidance on the other turned
that insofar as secure individuals experience decreases out to be nonsignificant. The subsample was so strongly correlated with
EBRS (r = .77) that the variance on the outcome measure may have been
in religiosity, these experiences occur during situations too restricted to allow for attachment to make a contribution.
of attachment deprivation (e.g., following divorce, 3. Partial correlation analyses were performed on attachment and
bereavement). religiosity for low and high parental religiousness groups (cf.
Granqvist, 1998). In general, results remained and were somewhat
Future studies should complement the simple attach- strengthened at low parental religiousness and disappeared at high
ment self-reports with measures (e.g., the Adult Attach- parental religiousness. A more detailed description will be obtained
ment Interview) (George, Kaplan, & Main, 1996) that from the author on request.
4. Given the low n in the increased religiosity group at Time 2 (T2),
are less prone to defensive strategies and more promis- outliers on the attachment and characteristics of religious change vari-
ing for tapping the unconscious components of IWMs. ables were searched according to a criterion of ±2.5 standard score
Granqvist / ATTACHMENT AND RELIGIOSITY IN ADOLESCENCE 269

deviations (Hair, Anderson, Tatham, & Black, 1995). Two outliers were Granqvist, P., & Hagekull, B. (1999). Religiousness and perceived
detected (one each on insecurity to mother and peer attachment anxi- childhood attachment: Profiling socialized correspondence and
ety); excluding the outlier on insecurity to mother resulted in signifi- emotional compensation. Journal for the Scientific Study of Religion,
cant associations between insecurity to mother in relation to themes of 38, 254-273.
compensation and themes of correspondence (partial rs = .49 and .45, Granqvist, P., & Hagekull, B. (2000). Religiosity, adult attachment,
respectively; ps < .05). Remaining results were unchanged. and why “singles” are more religious. International Journal for the
Psychology of Religion, 10, 111-123.
Granqvist, P., & Hagekull, B. (2001). Seeking security in the new age:
On attachment and emotional compensation. Journal for the Scien-
REFERENCES tific Study of Religion, 40, 529-547.
Hair, J. F., Jr., Anderson, R. E., Tatham, R. L., & Black, W. C. (1995).
Ainsworth, M. D. (1985). Attachments across the life-span. Bulletin of Multivariate data analysis with readings (4th ed.). Englewood Cliffs,
the New York Academy of Medicine, 61, 792-812.
NJ: Prentice Hall.
Ainsworth, M.D.S., Blehar, M. C., Waters, E., & Wall, S. (1978). Pat-
Hazan, C., & Shaver, P. R. (1986). Parental Caregiving Style Question-
terns of attachment: A psychological study of the strange situation.
naire. Unpublished document, Cornell University.
Hillsdale, NJ: Lawrence Erlbaum.
Hazan, C., & Zeifman, D. (1999). Pair bonds as attachments: Evalu-
Allen, J. P., & Land, D. (1999). Attachment in adolescence. In
ating the evidence. In J. Cassidy & P. R. Shaver (Eds.), Handbook of
J. Cassidy & P. R. Shaver (Eds.), Handbook of attachment theory and
attachment: Theory, research, and clinical applications (pp. 336-355).
research (pp. 319-335). New York: Guilford.
New York: Guilford.
Armsden, G. C., & Greenberg, M. T. (1987). The inventory of parent
Hood, R. W., Jr., Spilka, B., Hunsberger, B., & Gorsuch, R. L. (1996).
and peer attachment: Individual differences and their relation-
The psychology of religion: An empirical approach (2nd ed.). New York:
ship to psychological well-being. Journal of Youth and Adolescence,
Guilford.
16, 427-453.
Hyde, J. A. (2000). Adult attachment and the representation of God.
Batson, C. D., Natifeh, S. J., & Pate, S. (1978). Social desirability, reli-
Unpublished manuscript, Illinois School of Professional Psychol-
gious orientation, and racial prejudice. Journal for the Scientific
Study of Religion, 17, 31-41. ogy, Chicago.
Bowlby, J. (1969). Attachment and Loss: Vol. 1. Attachment. New York: James, W. (1902). Varieties of religious experience. New York: Longmans,
Basic Books. Green.
Bowlby, J. (1973). Separation anxiety and anges: Vol. 2. Attachment and Kirkpatrick, L. A. (1992). An attachment-theory approach to the psy-
Loss. New York: Basic Books. chology of religion. International Journal for the Psychology of Reli-
Bowlby, J. (1980). Attachment and Loss: Vol. 3. Loss. New York: Basic gion, 2, 3-28.
Books. Kirkpatrick, L. A. (1994). The role of attachment in religious belief
Brennan, K. A., Clark, C. A., & Shaver, P. R. (1998). Self-report mea- and behavior. In K. Bartholomew & D. Perlman (Eds.), Advances in
surement of adult attachment: An integrative overview. In J. A. personal relationships (Vol. 5, pp. 239-265). London: Jessica
Simpson & W. S. Rholes (Eds.), Attachment theory and close relation- Kingsley.
ships (pp. 46-76). New York: Guilford. Kirkpatrick, L. A. (1995). Attachment theory and religious experi-
Bretherton, I. (1985). Attachment theory: Retrospect and prospect. ence. In R. W. Hood, Jr. (Ed.), Handbook of religious experience (pp.
In I. Bretherton & E. Waters (Eds.), Growing points of attachment the- 446-475). Birmingham, AL: Religious Education.
ory and research. Monographs of the Society for Research in Child Develop- Kirkpatrick, L. A. (1997). A longitudinal study of changes in religious
ment, 50(1-2, Serial No. 209), 3-35. belief and behavior as a function of individual differences in adult
Bretherton, I. (1987). New perspectives on attachment relations: attachment style. Journal for the Scientific Study of Religion, 36, 207-
Security, communication, and internal working models. In J. D. 217.
Osofsky (Ed.), Handbook of infant development (2nd ed., pp. 1061- Kirkpatrick, L. A. (1998a). Evolution, pair-bonding, and reproduc-
1100). New York: John Wiley. tive strategies: A reconceptualization of adult attachment. In J. A.
Bretherton, I. (1991). The roots and growing points of attachment Simpson & W. S. Rholes (Eds.), Attachment theory and close relation-
theory. In C. M. Parkes, J. Stevenson-Hinde, & P. Marris (Eds.), ships (pp. 353-393). New York: Guilford.
Attachment across the life-cycle (pp. 9-32). London: Tavistock/ Kirkpatrick, L. A. (1998b). God as a substitute attachment figure: A
Routledge. longitudinal study of adult attachment style and religious change
Burris, C. T., Batson, C. D., Altstaedten, M., & Stephens, K. (1994). in college students. Personality and Social Psychology Bulletin, 24,
“What a friend . . . ”: Loneliness as a motivator of intrinsic religion. 961-973.
Journal for the Scientific Study of Religion, 33, 326-334. Kirkpatrick, L. A. (1999). Attachment and religious representations
Cassidy, J., & Shaver, P. R. (Eds.). (1999). Handbook of attachment: The- and behavior. In J. Cassidy & P. R. Shaver (Eds.), Handbook of attach-
ory, research, and clinical applications. New York: Guilford. ment theory and research (pp. 803-822). New York: Guilford.
Colin, V. L. (1996). Human attachment. New York: McGraw-Hill. Kirkpatrick, L. A., & Shaver, P. R. (1990). Attachment theory and reli-
Collins, N. L., & Read, S. J. (1990). Adult attachment, working mod- gion: Childhood attachments, religious beliefs and conversions.
els, and relationship quality in dating couples. Journal of Personality Journal for the Scientific Study of Religion, 29, 315-334.
and Social Psychology, 58, 644-663. Kirkpatrick, L. A., & Shaver, P. R. (1992). An attachment-theoretical
Erikson, E. (1950). Childhood and society. New York: Norton. approach to romantic love and religious belief. Personality and
Fraley, R. C., & Davis, K. E. (1997). Attachment formation and trans- Social Psychology Bulletin, 18, 266-275.
fer in young adults’ close friendships and romantic relationships. Kobak, R., & Ferenz-Gillies, R. (1995). Emotion regulation and
Personal Relationships, 4, 131-144. depressive symptoms during adolescence: A functionalist per-
Fraley, R. C., & Waller, N. G. (1998). Adult attachment patterns: A test spective. Development and Psychopathology, 7, 183-192.
of the typological model. In J. A. Simpson & W. S. Rholes (Eds.), Kobak, R., Ferenz-Gillies, R., Everhart, E., & Seabrook, L. (1994).
Attachment theory and close relationships (pp. 77-114). New York: Maternal attachment strategies and emotion regulation with ado-
Guilford. lescent offspring. Journal of Research on Adolescence, 4, 553-566.
Freud, S. (1964). The future of an illusion. Garden City, NY: Double Day. Kobak, R. R., Cole, H. E., Ferenz-Gillies, R., & Fleming, W. S. (1993).
(Original work published 1923) Attachment and emotion regulation during mother-teen prob-
George, C., Kaplan, N., & Main, M. (1996). Attachment interview for lem solving: A control theory analysis. Child Development, 64, 231-
adults (3rd ed.). Unpublished manuscript, University of Califor- 245.
nia, Berkeley. Lapsley, D. K., Rice, K. G., & Fitzgerald, D. P. (1990). Adolescent
Granqvist, P. (1998). Religiousness and perceived childhood attach- attachment, identity, and adjustment to college: Implications for
ment: On the question of compensation or correspondence. Jour- the continuity of adaptation hypothesis. Journal of Counseling and
nal for the Scientific Study of Religion, 37, 350-367. Development, 68, 561-565.
270 PERSONALITY AND SOCIAL PSYCHOLOGY BULLETIN

Londerville, S., & Main, M. (1981). Security of attachment, compli- an international perspective: Welfare and values of youths].
ance, and maternal training in the second year of life. Developmen- Statens Offentliga Utredningar, 73. Stockholm: Civildepartementet.
tal Psychology, 17, 289-299. Rice, K. G. (1990). Attachment and adolescence: A narrative and
Maccoby, E. E. (1984). Socialization and developmental change. meta-analytic review. Journal of Youth and Adolescence, 19, 511-538.
Child Development, 55, 317-328. Richters, J. E., & Waters, E. (1991). Attachment and socialization: The
Main, M., Kaplan, N., & Cassidy, J. (1985). Security in infancy, child- positive side of social influence. In M. Lewis & S. Feinman (Eds.),
hood and adulthood: A move to the level of representation. In Social influences and socialization in infancy: Genesis of behavior series
I. Bretherton & E. Waters (Eds.), Growing points of attachment theory (Vol. 6, pp. 185-213). New York: Plenum.
and research. Monographs of the Society for Research in Child Develop- Rizzuto, A.-M. (1979). The birth of the living God: A psychoanalytical
ment, 50(1-2, Serial No. 209), 66-104. study. Chicago: Chicago University Press.
Matas, L., Arend, R. A., & Sroufe, L. A. (1978). Continuity of adapta- Roof, W. C., & Hadaway, C. K. (1979). Denominational switching in
tion in the second year: The relationship between quality of the seventies: Going beyond Stark and Glock. Journal for the Scien-
attachment and later competence. Child Development, 49, 547-556. tific Study of Religion, 18, 363-379.
Mikulincer, M. (1998). Adult attachment style and affect regulation: Roof, W. C., & McKinney, W. (1987). American mainline religion: Its
Strategic variations in self-appraisals. Journal of Personality and changing shape and future. New Brunswick, NJ: Rutgers University
Social Psychology, 75, 420-435. Press.
Moore, D. (1987). Parent-adolescent separation: The construction of Simpson, J. A., Rholes, W. S., & Nelligan, J. S. (1992). Support seeking
adulthood by late adolescents. Developmental Psychology, 23, 298- and support giving within couples in an anxiety provoking situa-
307. tion: The role of attachment styles. Journal of Personality and Social
Palladino Schultheiss, D. E., & Blustein, D. L. (1994). Role of adolescent- Psychology, 62, 434-446.
parent relationships in college student development and adjust- Sroufe, L. A., & Waters, E. (1977). Attachment as an organizational
ment. Journal of Counseling Psychology, 41, 248-255.
construct. Child Development, 48, 1184-1199.
Papini, D. R., & Roggman, L. A. (1992). Adolescent perceived attach-
Starbuck, E. D. (1899). The psychology of religion. New York: Scribner.
ment to parents in relation to competence, depression, and anxi-
ety: A longitudinal study. Journal of Early Adolescence, 12, 420-440.
Pettersson, T. (1994). Svensk mentalitet i ett internationellt per- Received March 29, 2000
spektiv. Ungdomars välfärd och värderingar [Swedish mentality in Revision accepted July 25, 2001