Cover Page
The handle http://hdl.handle.net/1887/51103 holds various files of this Leiden University dissertation
Author: Goemans, A.
Title: The development of children in foster care Issue Date: 2017-06-27
103
)267(5&+,/'5(1·6
36<&+262&,$/'(9(/230(17
$1')267(53$5(176·675(66
7(67,1*$75$16$&7,21$/
MODEL
6XEPLWWHGIRUSXEOLFDWLRQ
*RHPDQV$9DQ*HHO0 9HGGHU3)RVWHUFKLOGUHQ·VSV\FKRVRFLDOGHYHORSPHQWDQGIRVWHU
parents’ stress: Testing a transactional model.
5
104 ABSTRACT
The goal of this three-wave longitudinal study was to analyze foster parents’ stress and foster children’s internalizing and externalizing behavior in a transactional framework. In a sample of 237 children in foster care in the Netherlands we examined concurrent, prospective unidirectional and bidirectional relations between foster children and their foster parents by using cross-lagged structural equation modeling. In contrast to our hypothesis, we found no bidirectional relations for the entire group of foster children. There were unidirectional prospective pathways from foster children’s internalizing and externalizing problems to foster SDUHQWV· VWUHVV EXW QR VLJQLÀFDQW SURVSHFWLYH SDWKZD\V IURP IRVWHU SDUHQWV· VWUHVV WR IRVWHU
children’s psychosocial functioning. The lack of bidirectional relations is surprising given the presence of transactional relations in biological parent-child dyads. Foster parents’ stress seem QRW WR LQÁXHQFH WKHLU IRVWHU FKLOGUHQ SV\FKRVRFLDO GHYHORSPHQW DQG WKH TXHVWLRQ LV ZKHWKHU
foster parents are, in more general terms, able to pervade their foster children’s minds and OLYHVDQGZKHWKHUIRVWHUFKLOGUHQDUHDEOHWREHQHÀWIURPWKHLULPSURYHGKRPHHQYLURQPHQW
105
5
INTRODUCTION
Foster care provides a second chance for a good life for children who no longer can EHFDUHGIRUE\WKHLUSDUHQWV<HWVRPHWLPHVFKLOGUHQ·VGHYHORSPHQWDQGDGDSWLYHIXQFWLRQLQJ
GRHV QRW WDNH D SRVLWLYH FRXUVH LQ IRVWHU FDUH %XUQV HW DO *RHPDQV 9DQ *HHO 9HGGHU0DDVNDQWHWDO9DQVFKRRQODQGW9DQGHUIDHLOOLH9DQ+ROHQ'H0DH\HU
5REEHUHFKWV6RPHIRVWHUFKLOGUHQDUHFKDUDFWHUL]HGE\SUREOHPDWLFSV\FKRVRFLDO
functioning and their problems do not easily decrease during their stay in foster care (see IRUDQRYHUYLHZ*RHPDQV9DQ*HHO 9HGGHU+HQFHIRVWHUFDUHLVQRWDOZD\VDQ
effective intervention. Furthermore, foster children’s problem behaviors, especially externalizing problems, are a major reason for placement breakdown (Oosterman et al., 2007). To improve the psychosocial development of children in foster care and to reduce the risk of breakdown, it is important to longitudinally study conditions and processes that affect foster children’s SV\FKRVRFLDOGHYHORSPHQW-DFNVRQHWDO7KHLQWHUDFWLRQEHWZHHQSDUHQWDOVWUHVVDQG
FKLOGGHYHORSPHQWPD\EHRQHRIWKHLPSRUWDQWSURFHVVHV6WXGLHVLQWKHJHQHUDOSRSXODWLRQ
have repeatedly found that parental stress is related to children’s psychosocial outcomes (Crnic /RZ'HDWHU-Deckard, 1998). This relation is not assumed to be unidirectional: both FKLOGUHQ DQG SDUHQWV LQÁXHQFH HDFK RWKHU %HOO 'HDWHU'HFNDUG .DUUDNHU
&ROHPDQ2·&RQQRU6DPHURII+RZHYHUQRVWXG\WRGDWHKDVWHVWHGWKH
mutual associations with respect to parental stress and children’s psychosocial development in foster care. Moreover, studies examining these associations and underlying processes by three-term interactions (i.e., child-parent-child or parent-child-parent) are lacking. The aim of the current study is therefore to examine in a three-wave longitudinal study the psychosocial GHYHORSPHQWRIIRVWHUFKLOGUHQIURPDVRFDOOHGWUDQVDFWLRQDOSHUVSHFWLYH6DPHURII
The focus is on foster parents’ stress and children’s internalizing and externalizing behavior problems.
Parental Stress and Children’s Psychosocial Development
Foster parents take care of children who often have broken attachments from their SDUHQWVLQLWLDOO\FDXVHGE\VXERSWLPDOSDUHQWLQJQHJOHFWRUHYHQDEXVH'XEQHU 0RWWD
*UHHVRQ HW DO 2VZDOG HW DO ,I WKH DWWDFKPHQW WR WKHLU SDUHQWV VWLOO
had positivity, but family functioning together with the parents was no longer possible for other reasons, the attachment is still broken by the placement in a foster family. The eventual attachment problems and the consequences for children’s behavior and for the interaction between foster children and their substitute parents create important challenges for a foster family. This is even the case although many foster parents deliberately chose to be a foster parent (Rodger et al., 2006), and received training prior to placement of the foster child (Dorsey et al., 2008). These challenges are likely to place a burden on the foster parents and PLJKWLQFUHDVHSDUHQWDOVWUHVV)DUPHUHWDO+XUOEXUWHWDO-RQHV 0RUULVVHWWH
0F&DUWK\-DQHZD\ *HGGHV3DUHQWDOVWUHVVPD\QHJDWLYHO\LPSDFWRQIRVWHU
SDUHQWV· PRWLYDWLRQ WR FRQWLQXH IRVWHULQJ DQG PD\ OHDG WR IRVWHU SDUHQW EXUQRXW %URZQ
106
%HGQDU)DUPHUHWDOZKLFKLQWXUQPLJKWKDYHDQHJDWLYHHIIHFWRQIRVWHU
children’s psychosocial development.
Previous studies showed that higher levels of stress in foster parents were related to higher levels of internalizing and externalizing behavior problems in foster children (Kelley HW DO 0XUUD\ HW DO 1LOVHQ 5RUN +RZHYHU WKH FRUUHODWLRQDO
nature of these studies does not allow for conclusions about the directionality of the relations.
Longitudinal studies that focused exclusively on child-to-parent effects showed that increased behavior problems are related to an increase of parent reported stress (Hurlburt et al.,
9DQGHUIDHLOOLH HW DO $OWKRXJK WKHVH ORQJLWXGLQDO VWXGLHV KDYH LPSURYHG RXU
knowledge about the unidirectional effects from foster children to their foster parents, it is generally emphasized that the interactions between parenting and child development are ELGLUHFWLRQDO %RUQVWHLQ 1HHFH HW DO 6WRQH HW DO DQG QHHG WR EH
VWXGLHGLQDWUDQVDFWLRQDOIUDPHZRUN6DPHURII,QWKLVWUDQVDFWLRQDOIUDPHZRUNFKLOG
development is considered as a product of the continuous dynamic interactions of children and WKHLUHQYLURQPHQWV%RUQVWHLQ6DPHURII7KHFXUUHQWVWXG\LVWKHÀUVWWRWHVWWKH
transactional relations between foster children and foster parents’ stress.
Current Study
7KLVVWXG\DLPVWRH[DPLQHIRVWHUFDUHLQDWUDQVDFWLRQDOSHUVSHFWLYH%RUQVWHLQ
6DPHURII:HZLOOH[DPLQHWKHLQWHUSOD\EHWZHHQIRVWHUFKLOGUHQ·VLQWHUQDOL]LQJDQG
externalizing behaviors and their foster parents’ stress by using cross-lagged structural equation PRGHOLQJ6WDWLVWLFDOPRGHOLQJRIWUDQVDFWLRQDOSURFHVVHVUHTXLUHVPRQLWRULQJRIIDFWRUVLQWKH
child and context over time to provide a basis for determining when these factors affect and DUH FKDQJHG E\ HDFK RWKHU :H VWXG\ FKLOGUHQ LQ IRVWHU FDUH LQD ORQJLWXGLQDO GHVLJQ ZLWK
three waves. This allows to study two-term unidirectional relations (i.e., child affects parents or parent affects child), as well as three-term reciprocal relations (i.e., the child changes the parent and is in turn changed by the changed parent, or the parent changes the child and in WXUQLVFKDQJHGE\WKHFKDQJHGFKLOG%RUQVWHLQ7KLVVWXG\VKRXOGVKHGOLJKWRQWKH
question whether there are transactional relations between foster children’s behavior problems and foster parents’ stress, and as a consequence whether interventions and additional support VKRXOGEHDLPHGDWIRVWHUFKLOGUHQIRVWHUSDUHQWVRUERWK:HWKHUHIRUHWHVWHGDQGFRPSDUHG
three models (depicted in Figure 5.1) on the concurrent, unidirectional and bidirectional UHODWLRQV EHWZHHQ IRVWHU FKLOGUHQ·V SV\FKRVRFLDO IXQFWLRQLQJ DQG IRVWHU SDUHQWV· VWUHVV :H
performed separate analyses for internalizing and externalizing behavior problems, because previous studies showed that the two may have divergent relations with parental stress with externalizing behaviors being more stressful for parents and consequently related to placement breakdown (Oosterman et al., 2007). In line with the results from studies on the general SRSXODWLRQ1HHFHHWDO6WRQHHWDOZHK\SRWKHVL]HFRQFXUUHQWSURVSHFWLYH
unidirectional and bidirectional positive relations between foster children’s internalizing and externalizing behavior problems and foster parents’ stress. In addition, we expect that the ELGLUHFWLRQDOPRGHOZLOOÀWWKHGDWDEHWWHUWKDQDPRGHOZLWKRQO\FRQFXUUHQWUHODWLRQVRUD
model with either parent-to-child or child-to-parent relations. As a last step, we examined ZKHWKHUWKHUHVXOWVZHUHVWDEOHDFURVVGLIIHUHQWVXEJURXSVRIIRVWHUFKLOGUHQ%HFDXVHLWKDV
107
5
been previously shown that age (Oosterman et al., 2007), placement history (Oosterman et al., 2007) and duration of the placement (Goemans et al., 2015) are important variables with respect to foster children’s functioning, we performed multigroup analyses on these variables to study the stability of our results across different groups of foster children.
METHOD
Participants
Participants in this three-wave longitudinal study were 237 foster parents of foster children residing in regular foster care in the Netherlands. The foster children were between 4 and 17 years old (M = 10.69, SD DW WKH ÀUVW PHDVXUHPHQW DQG LQFOXGHG
JLUOVDQGER\V7KHIRVWHUFKLOGUHQKDGRQDYHUDJHH[SHULHQFHG
previous foster placements (SD = 1.37) and the mean time in the current foster placement was 62 months (SD PRQWKVDWWKHÀUVWZDYHRIGDWDFROOHFWLRQ7KHPDMRULW\RIWKHIRVWHU
FKLOGUHQUHVLGHGLQPDQGDWHGIRVWHUFDUHDQGZLWKQRQNLQVKLSIRVWHUIDPLOLHV
,QRYHUKDOIRIWKHIRVWHUIDPLOLHVWKHUHZHUHQRRWKHUIRVWHUFKLOGUHQSUHVHQW)RVWHU
Figure 5.1. Structural equation model for testing foster children’s psychosocial functioning and foster parents’ stress in a transactional perspective.
Foster parents:
Parental stress
Foster parents:
Parental stress
Foster parents:
Parental stress
Foster child:
Internalizing/
Externalizing
Foster child:
Internalizing/
Externalizing
Foster child:
Internalizing/
Externalizing
Time 1 Time 2 Time 3
108
IDPLOLHV ZHUH PRVWO\ WZRSDUHQW IDPLOLHV $OPRVW RQH ÀIWK RI PRWKHUV FRPSOHWHG
SULPDU\RUMXQLRUKLJKVFKRROEXWPRVWRIWKHPFRPSOHWHGVHQLRUKLJKVFKRRO RUDXQLYHUVLW\RIDSSOLHGVFLHQFHV/HVVWKDQWHQSHUFHQWJUDGXDWHGIURPDQ
academic university. The educational backgrounds of fathers resembled those of mothers with
ZKRFRPSOHWHGSULPDU\RUMXQLRUKLJKVFKRROZKRFRPSOHWHGVHQLRUYRFDWLRQDO
KLJKVFKRRODQGDXQLYHUVLW\RIDSSOLHGVFLHQFHV,QDGGLWLRQRIIDWKHUVKHOGDQ
DFDGHPLF XQLYHUVLW\ GHJUHH 6LPLODU SHUFHQWDJHV UHJDUGLQJ HGXFDWLRQ FDQ EH IRXQG LQ WKH
JHQHUDOSRSXODWLRQ6WDWLVWLFV1HWKHUODQGV0RVWIRVWHUPRWKHUVDSSUR[LPDWHO\
KDGDMREZKHUHDVWKLVZDVWKHFDVHIRURIIRVWHUIDWKHUV)RVWHUPRWKHUVZRUNHGIHZHU
hours a week than foster fathers.
7KHLQLWLDOVDPSOHWKDWSDUWLFLSDWHGLQ:DYH,FRQVLVWHGRIIRVWHUIDPLOLHV)RVWHU
children that fell outside the age range 4-17 (N = 30), for whom we had no psychosocial functioning scores (N = 43), or who resided in part-time foster care (N = 45) were excluded IURPWKLVVWXG\7KHÀQDOVDPSOHFRQVLVWHGRIIRVWHUIDPLOLHV2IWKHIRVWHUIDPLOLHV
WKDWSDUWLFLSDWHGLQ:DYH,IRVWHUIDPLOLHVDOVRSDUWLFLSDWHGLQ:DYH,,DWWULWLRQUDWH
DQGLQ:DYH,,,DWWULWLRQUDWH5HDVRQVIRUDWWULWLRQZHUHPRVWO\XQNQRZQ
(N EXWUHXQLÀFDWLRQZLWKELUWKSDUHQWVN = 17), placement change to another foster family (N = 9), placement change to residential or group care (N = 8), and leaving foster care because of independent living of the foster child (N = 5), were communicated as reasons for dropout. Little’s (1988) Missing Completely at Random (MCAR) test indicated that the missing data were missing completely at random (ʖ2 (47) = 56.542, p 7KHÀQDOVDPSOH
of foster children (N = 237) was compared to the foster children that dropped out after :DYH , DQG GLG QRW SDUWLFLSDWH LQ :DYH ,,, HLWKHU N = 194) on demographic variables (age, gender, kinship/non-kinship, duration placement, placement history, legal framework) and variables substantively relevant to the study (internalizing, externalizing and prosocial behavior, parenting, parental stress). T-tests and chi-square tests revealed only one difference between the two groups: there were fewer voluntary placements in the sample of remaining foster children (ʖ2 (1) = 8.241, p < .004). Later in this text (see Analyses) we explain how we DFKLHYHGDÀQDOVDPSOHVL]HRI
Instruments
Sociodemographics and foster care characteristics. Foster parents provided information about the foster child (e.g., age, gender, placement history, duration of the current placement), foster family (e.g., kinship or non-kinship, single or double-parent household) and foster placement (e.g., legal framework, parental visiting).
Psychosocial functioning. To measure the psychosocial development of the foster FKLOGUHQ WKH 'XWFK YHUVLRQ 9DQ :LGHQIHOW HW DO RI WKH 6WUHQJWKV DQG 'LIÀFXOWLHV
4XHVWLRQQDLUH6'4IRUSDUHQWVZDVXVHG*RRGPDQ7KH6'4FRQVLVWVRILWHPV
which can be answered on a 3-point Likert scale ranging from 0 (not true) to 2 (very true). As SUHYLRXVO\VXJJHVWHG*RRGPDQ/DPSLQJ 3ORXELGLVWKHLWHPVZHUHFRPELQHG
in three subscales: internalizing behavior problems, externalizing behavior problems, and prosocial behavior. Only the internalizing and externalizing subscale are used in this study.
The subscale internalizing behavior problems consists of ten items covering emotional and
109
5
SHHU SUREOHPV 6DPSOH LWHPV DUH ¶KDV PDQ\ ZRUULHV RU RIWHQ VHHPV ZRUULHG· DQG ¶SLFNHG
on or bullied by other children’. The subscale externalizing behavior problems is formed by combining the ten items for conduct and hyperactivity problems. Items are for example ‘often OLHVRUFKHDWV·DQG¶UHVWOHVVRYHUDFWLYHFDQQRWVWD\VWLOOIRUORQJ·7KH6'4LVDZHOOYDOLGDWHG
VFUHHQLQJLQVWUXPHQW$FKHQEDFKHWDO9DQ:LGHQIHOWHWDODQGWKHVXEVFDOHV
have been shown to have good convergent and discriminant validity (Goodman et al., 2010).
6WXGLHVXVLQJWKH'XWFKYHUVLRQRIWKH6'4IRXQGDFFHSWDEOHWRJRRGSV\FKRPHWULFSURSHUWLHV
0XULVHWDO9DQ:LGHQIHOWHWDO,QWKHFXUUHQWVWXG\&URQEDFK·VDOSKDVUDQJHG
from .747 to .792 for internalizing and from .823 to .854 for externalizing problems and are reported in Table 5.1.
Parenting stress. Parenting stress was measured with the abbreviated version of the 1LMPHHJVH 2XGHUOLMNH 6WUHVV ,QGH[ YHUNRUW 126,. 'H %URFN 9HUPXOVW *HUULV $ELGLQ
ZKLFK LV EDVHG RQ WKH 3DUHQWLQJ 6WUHVV ,QGH[ 36, $ELGLQ ZKLFK KDV EHHQ
SUHYLRXVO\ XVHG LQ UHVHDUFK ZLWK IRVWHU SDUHQWV 0DDVNDQW 9DQ 5RRLM 2YHUEHHN 2RUW +HUPDQQV0XUUD\HWDO1LOVHQ7LPPHU8UTXL]D =HEHOO9DQ
$QGHO 7KH 126,. FRQVLVWV RI LWHPV ZKLFK DUH UDWHG RQ D SRLQW /LNHUW VFDOH
ranging from 1 (totally disagree) to 6 (totally agree). A sample item is: ‘The child does things WKDWERWKHUPHDJUHDWGHDO·)XUWKHUPRUHWKH126,.KDVEHHQIRXQGWRKDYHKLJKLQWHUQDO
FRQVLVWHQFLHV 'H %URFN HW DO +DVNHWW $KHUQ :DUG $OODLUH 7KH LQWHUQDO
consistencies in the current study varied between .952 and .961 (see Table 5.1).
Procedure
7KHVWXG\ZDVDSSURYHGE\WKH/HLGHQ8QLYHUVLW\(WKLFV5HYLHZ%RDUG$OOIRVWHUFDUH
agencies in the Netherlands (N = 28) were asked to participate in a three-wave longitudinal VWXG\ RQ WKH GHYHORSPHQW RI FKLOGUHQ LQ IRVWHU FDUH 6HYHQ DJHQFLHV DJUHHG WR
participate. The main reason for foster care agencies to not participate was that they already participated in other foster care related research projects and wanted to prevent a research overload for their foster families. For those foster care agencies that agreed to participate, foster parents were informed about the goal of the study and consent was asked by the foster care agencies. The researchers received the contact information for those foster parents who gave consent.
Children were followed for twelve months throughout their stay with their foster parents.
'XULQJ WKHVH WZHOYH PRQWKV ZH HVWDEOLVKHG RQH EDVHOLQH PHDVXUHPHQW :DYH , 2FWREHU
DQGWZRVXEVHTXHQWPHDVXUHPHQWVVHSDUDWHGE\VL[PRQWKLQWHUYDOV:DYH,,$SULO
:DYH,,2FWREHU)RUWKHÀUVWZDYHIRVWHUIDPLOLHVZHUHLQYLWHGE\WKH
authors to participate in the study and to complete a questionnaire about the foster placement.
For each of the three waves, foster parents were asked to complete an online questionnaire containing questions about the foster child, the foster family and the foster placement. Each ZDYH WZR UHPLQGHUV WR ÀOO RXW WKH TXHVWLRQQDLUH ZHUH VHQW RQ D WZRZHHN LQWHUYDO 7KH
online questionnaire was closed three weeks after the last reminder. All foster parents who SDUWLFLSDWHGLQ:DYH,N UHVSRQVHUDWHZHUHLQYLWHGWRSDUWLFLSDWHLQERWK
:DYH ,, DQG :DYH ,,, )RU ZDYH ,,, WKH\ ZHUH LQYLWHG HYHQ LI WKH\ KDG QRW SDUWLFLSDWHG LQ
:DYH ,, 7KH VDPH IRVWHU SDUHQW FRPSOHWHG WKH PHDVXUHV DW HYHU\ WLPH SRLQW %RWK IRVWHU
110
parents could be primary caregivers so either of the two foster parents could complete the questionnaire. Research has shown that there is strong agreement in reporting between foster PRWKHUVDQGIRVWHUIDWKHUV0F$XOH\ 7UHZ6WDQJHU /HZLV
Analyses
6WUXFWXUDOHTXDWLRQPRGHOLQJXVLQJ(46%HQWOHUZDVXVHGWRWHVWDSDUDOOHO
VHWRIWKUHHVWUXFWXUDOHTXDWLRQPRGHOVVHH)LJXUH%DVHOLQHPRGHOVUHIHUUHGWRDVPRGHO
1) were tested to investigate the concurrent relations between parental stress and children’s SV\FKRVRFLDO IXQFWLRQLQJ 6WDELOLW\ HIIHFWV LH UHJUHVVLRQ OLQHV EHWZHHQ WKH VDPH FRQVWUXFWV
over time) were included for child behavior problems over time as well as for parental stress RYHUWLPH,QWKHVHFRQGPRGHOHLWKHUFURVVSDWKVIURPSDUHQWWRFKLOGPRGHOD)LJXUH
GRWWHGOLQHVRUFKLOGWRSDUHQWPRGHOE)LJXUHGDVKHGOLQHVZHUHDGGHGWRWHVWWKH
longitudinal one-way effects of parental stress to foster children’s psychosocial functioning, or YLFHYHUVD,QWKHÀQDOPRGHOPRGHOERWKSDUHQWWRFKLOGDQGFKLOGWRSDUHQWFURVVSDWKV
were added to test the bidirectional relations between parental stress and foster children’s internalizing and externalizing behavior problems. Multi-group analyses were performed for age (4-10 years vs. 11-17 years), placement history (no previous placements vs. one or more SUHYLRXVSODFHPHQWVDQGGXUDWLRQRIWKHSODFHPHQW\HDUVYV\HDUV:HWHVWHGZKLFK
PRGHOÀWVEHVWIRUERWKJURXSVVLPXOWDQHRXVO\6XEVHTXHQWO\ZHDQDO\]HGZKHWKHUUHJUHVVLRQ
loadings were similar for the different subgroups by constraining the regression loadings to be HTXDO,IWKLVUHVXOWHGLQDGHFUHDVHLQPRGHOÀWWKLVLQGLFDWHGWKDWWKHELGLUHFWLRQDOUHJUHVVLRQ
ORDGLQJVGLIIHUHGEHWZHHQWKHVXEJURXSVRIIRVWHUFKLOGUHQ,IWKHPRGHOÀWUHPDLQHGHTXDOO\
good, this indicated that the bidirectional regression loadings were similar.
,Q RUGHU WR DVVHVV WKH PRGHO ÀW (46 SURYLGHV JRRGQHVVRIÀW VWDWLVWLFV ZKLFK DUH
reported for each individual model in Table 5.3. In line with the suggestions of Hooper,
&RXJKDQ DQG 0XOOHQ WKH IROORZLQJ JRRGQHVVRIÀW VWDWLVWLFV UHSUHVHQW D VHOHFWLRQ
ZKLFK DUH ZLGHO\ UHJDUGHG DV WKH PRVW LQIRUPDWLYH )LUVW <XDQ%HQWOHU &KLVTXDUH YDOXH
(Y-Bʖ2DUHVFDOHGFKLVTXDUHVWDWLVWLFIRUQRQQRUPDOPLVVLQJGDWD<XDQ %HQWOHULV
UHSRUWHGDQGVKRXOGEHQRQVLJQLÀFDQW6HFRQGWKH&RPSDUDWLYH)LW,QGH[&),UHSUHVHQWV
ZKHWKHUWKHK\SRWKHVL]HGPRGHOÀWVWKHGDWDEHWWHUWKDQDPRUHUHVWULFWHGEDVHOLQHPRGHO
&),YDOXHV!DUHFRQVLGHUHGDVJRRG+X %HQWOHU7KLUGWKH5RRW0HDQ6TXDUH
(UURURI$SSUR[LPDWLRQ506($UHÁHFWVZKHWKHUWKHDSULRULPRGHOVXFFHVVIXOO\UHSURGXFHV
GDWDSDWWHUQV506($YDOXHVOHVVWKDQLQGLFDWHDFFHSWDEOHPRGHOÀWDQGYDOXHVEHORZ
LQGLFDWHJRRGÀW+X %HQWOHU)RXUWKWKH$NDLNH,QIRUPDWLRQ&ULWHULRQ$,&FRQFHUQV
WKHLVVXHRISDUVLPRQ\LQWKHDVVHVVPHQWRIPRGHOÀWDQGFDQEHXVHGLQFRPSDULVRQRIWZRRU
PRUHPRGHOVZLWKVPDOOHUYDOXHVUHSUHVHQWLQJDEHWWHUÀWRIWKHK\SRWKHVL]HGPRGHO%\UQH
7RH[DPLQHZKHWKHUWKHJRRGQHVVRIÀWVWDWLVWLFVLPSURYHGIRUWKHFRQVHFXWLYHPRGHOV
WKH&KLVTXDUHGLIIHUHQFHWHVWFRUUHFWHGIRUWKHXVHRI<XDQ%HQWOHU&KLVTXDUH%\UQH
1HZVRP6DWRUUD %HQWOHUZDVSHUIRUPHG$VLJQLÀFDQW&KLVTXDUHGLIIHUHQFH
test means that the consecutive nested model explains the data better than the previous, PRUHSDUVLPRQLRXVPRGHO,QFDVHRIDQRQVLJQLÀFDQWFKLVTXDUHGLIIHUHQFHWHVWWKHPRUH
SDUVLPRQLRXVPRGHOLVSUHIHUUHG%HQWOHU 0RRLMDDUW,QDGGLWLRQWKHGLIIHUHQFHLQ&),
DQG506($EHWZHHQWKHPRGHOVDUHUHSRUWHG,IWKH&),LQFUHDVHVPRUHWKDQWKLVPHDQV
WKDWWKHUHLVVXEVWDQWLDOLQFUHDVHLQPRGHOÀW
111
5
$VLQPDQ\ORQJLWXGLQDOVWXGLHVRQIRVWHUFKLOGUHQ-DFNVRQHWDODWWULWLRQZDV
D SUREOHP LQ WKH FXUUHQW VWXG\ ,QWHUQDOL]LQJ DQG H[WHUQDOL]LQJ SUREOHPV ERWK KDG
DQG PLVVLQJ LQ WKH VHFRQG DQG WKLUG ZDYH UHVSHFWLYHO\ )RU SDUHQWDO VWUHVV WKLV ZDV
DQGPLVVLQJIRUWKHWKUHHFRQVHFXWLYHZDYHV,QOLQHZLWKSUHYLRXVVWXGLHV
'RRUQZDDUG%LFNKDP5LFK7HU%RJW 9DQGHQ(LMQGHQ.HLMVHUVHWDO6XOLN
%ODLU 0LOOV-.RRQFH %HUU\ *UHHQEHUJ DQG DV UHFRPPHQGHG (QGHUV %DQGDORV
*UDKDP-HOLĀLþ3KHOSV /HUQHUWKHFXUUHQWVWXG\XVHG)XOO,QIRUPDWLRQ
Maximum Likelihood (FIML) estimation to not only include foster children who participated in each of the three waves (N EXWDOVRIRVWHUFKLOGUHQZKRSDUWLFLSDWHGLQERWK:DYH,
DQG:DYH,,EXWQRWLQ:DYH,,,N RULQERWK:DYH,DQG:DYH,,,EXWQRWLQ:DYH,,
(N 7KLVUHVXOWHGLQDÀQDOVDPSOHRIIRVWHUFKLOGUHQ),0/HVWLPDWLRQXVHVDOORIWKH
available data to identify the parameter values that have the highest probability of producing WKHVDPSOHGDWD%DUDOGL (QGHUV%\UQH(QGHUV*UDKDP$OO
statistics reported in this article were estimated using robust FIML.
RESULTS
Descriptive statistics and correlations for all study variables are presented in Table 5.1.
The mean total behavior problem (internalizing and externalizing behaviors) of our sample (MWaveI = 12.63, SDWaveI MWaveII = 12.83, SDWaveII MWaveIII = 12.17, SDWaveIII
= 6.89) fell within the borderline range following the Dutch norm cut-off scores (Goedhart, 7UHIIHUV 9DQ:LGHQIHOW2IDOOIRVWHUFKLOGUHQDQGIHOOZLWKLQ
WKHQRUPDOUDQJHUDQJHIRUWKHWKUHHFRQVHFXWLYHZDYHVZKHUHDVDQG
IHOOZLWKLQWKHERUGHUOLQHUDQJHUDQJHDQGDQGZLWKLQ
WKH FOLQLFDO UDQJH UDQJH ZLWK UHJDUG WR WKHLU 6'4 WRWDO EHKDYLRU SUREOHPV VFRUH
%DVHGRQWKHQRQFOLQLFDO'XWFKQRUPFXWRIIVFRUHVRIWKH126,.IRUPRWKHUV'H%URFNHW
al., 1992), the mean total parental stress scores fell within the average range (range: 43-61).
3DUHQWDOVWUHVVZDVEHORZDYHUDJHUDQJHIRUDQGRIWKHSDUHQWV
IRUWKHWKUHHZDYHVUHVSHFWLYHO\ZKHUHDVDQGRIWKHIRVWHUSDUHQWV
VFRUHG ¶DYHUDJH· OHYHOV RI SDUHQWDO VWUHVV DQG DQG VFRUHG DERYH
average (range: 62-150) for the three consecutive waves respectively. Cross-time correlations were .76 and .77 for internalizing, .85 and .79 for externalizing behavior, and .76 and .76 for parenting stress.
112
Parental Stress and Internalizing Problems
The baseline model (model 1) to investigate the concurrent relations between parental VWUHVVDQGFKLOGUHQ·VLQWHUQDOL]LQJEHKDYLRUVZDVH[DPLQHGDQGÀWWHGGDWDZHOO<%ʖ2 (df = 6) = 8.005, p &), 506($ &,>@7KHFRQVHFXWLYH
models wherein either parent-child (model 2a) or child-parent (model 2b) effects, and the ELGLUHFWLRQDO UHODWLRQV PRGHO ZHUH WHVWHG DOVR ÀWWHG WKH GDWD ZHOO 7KH JRRGQHVVRIÀW
VWDWLVWLFVDUHUHSRUWHGLQ7DEOH'HVSLWHRYHUDOOJRRGPRGHOÀWVLJQLÀFDQWFURVVODJJHG
HIIHFWVZHUHRQO\SUHVHQWIRUFKLOGUHQ·VLQWHUQDOL]LQJSUREOHPVDW:DYH,,WRSDUHQWDOVWUHVVDW
:DYH,,,ɴ = .13, p < .05), and not for parental stress to children’s internalizing problems.
6WDQGDUGL]HGFRHIÀFLHQWVIRUHDFKPRGHODUHUHSRUWHGLQ7DEOH,QRUGHUWRWHVWZKHWKHU
WKHELGLUHFWLRQDOPRGHOÀWWKHGDWDEHWWHUWKDQWKHPRGHOZLWKRQO\FRQFXUUHQWUHODWLRQVRUWKH
models with either parent-to-child or child-to-parent effects, difference tests were performed for the consecutive models. The results are reported in Table 5.2. It appeared that the model ZLWKFURVVODJJHGSDWKVIURPFKLOGWRSDUHQWPRGHOEÀWEHWWHUWKDQWKHEDVHOLQHPRGHO
¨<%ʖ2 (df = 2) = 7.308, p 7KH ELGLUHFWLRQDO PRGHO PRGHO GLG QRW ÀW EHWWHU
WKDQPRGHOE¨<%ʖ2 (df = 2) = 1.288, p )RUUHDVRQVRISDUVLPRQ\%HQWOHU Mooijaart, 1989), model 2b is preferred over the bidirectional model. This model is depicted in Figure 5.2.
To test whether these results were stable across different subgroups of foster children, we performed several multigroup analyses. As was true for the entire group, parent-to-child HIIHFWVZHUHDOVRDEVHQWIRUWKHGLIIHUHQWVXEJURXSVDQGPRGHOEÀWWHGEHVWIRUHDFKRIWKH
subgroups. Unidirectional regression loadings from children to parents were similar for foster Table 5.1. Pearson correlations between the SDQ and NOSIK for each wave (T1, T2, T3).
M (SD) 1 2 3 4 5 6 7 8 9
76'4LQWHUQDOL]LQJ 5.03 (3.83) .767
76'4H[WHUQDOL]LQJ 7.60 (4.64) .387** .854
7126,. 56.38 (24.46) .468** .583** .952
76'4LQWHUQDOL]LQJ 5.13 (4.00) .771** .323** .384** .792
76'4H[WHUQDOL]LQJ 7.52 (4.44) .339** .847** .521** .368** .823
7126,. 56.84 (26.12) .338** .533** .780** .419** .597** .961
76'4LQWHUQDOL]LQJ 5.02 (3.80) .764** .283** .403** .811** .293** .346** .757
76'4H[WHUQDOL]LQJ 7.14 (4.40) .353** .784** .453** .374** .850** .512** .410** .839
7126,. 57.46 (26.44) .411** .505** .773** .471** .542** .797** .465** .590** .961
Note. Cronbach’s alphas are reported on the diagonal.
*p < .05, **p < .01.
113
5
children with different placement histories (no previous placements vs. one or more previous placement) and for foster children with different lengths of stay in the foster family (<4 years vs. >4 years). Unidirectional regression from children to parents’ loadings appeared to be different for younger (4-10 years) and older (11-17 years) foster children. For older children, VLPLODUWRWKHHQWLUHJURXSRIIRVWHUFKLOGUHQFKLOGUHQ·VLQWHUQDOL]LQJSUREOHPVDW:DYH,ZHUH
QRWVLJQLÀFDQWO\UHODWHGWRIRVWHUSDUHQWV·VWUHVVDW:DYH,,)RU\RXQJHUFKLOGUHQWKHUHZDVD
VLJQLÀFDQWFKLOGWRSDUHQWHIIHFWIURP:DYH,WR:DYH,,ɴ = .18, p < .05).
Figure 5.2. Structural equation model (2b-model) with cross paths between children’s internalizing behavior problems and foster parents’ stress.
*p < .05
Parental stress Parental stress Parental stress
Internalizing Internalizing Internalizing
Time 1 Time 2 Time 3
.35*
.37*
.76*
.01*
.47*
.78*
.52*
.47*
.34* .13* .20*
114
1RWH<%ʖ2 <XDQ%HQWOHUFKLVTXDUHdf GHJUHHVRIIUHHGRP&), FRPSDUDWLYHÀWLQGH[
506($ URRWPHDQVTXDUHHUURURIDSSUR[LPDWLRQ&, FRQÀGHQFHLQWHUYDO$,&
= Akaike Information Criterion.
1Difference tests were performed for the consecutive models. The 2a/2b-model were compared ZLWKWKHPRGHO7KHPRGHOZDVFRPSDUHGZLWKWKHEHVWÀWWLQJPRGHO
Table 5.2. Structural equation models on the concurrent (model 1), parent-to-child (model 2a), child-to-parent (model 2b) and transactional relations (model 3) between parental stress and internalizing and externalizing behavior problems.
Model Y-Bʖ2 df p CFI RMSEA AIC
Internalizing behavior problems
Model 1: concurrent 8.005 6 .238 1.000 .000 (.000, .081) -4.00 Model 2a: parent-to-child 6.635 4 .156 1.000 .019 (.000, .102) -1.37 Model 2b: child-to-parent 1.973 4 .741 1.000 .000 (.000, .060) -6.03 Model 3: transactional 0.874 2 .648 1.000 .000 (.000, .086) -3.13 Externalizing behavior problems
Model 1: concurrent 11.986 6 .068 .999 .047 (.000, .105) -.01 Model 2a: parent-to-child 9.326 4 .053 .999 .058 (.000, .125) 1.33 Model 2b: child-to-parent 2.398 4 .663 1.000 .000 (.000, .061) -5.60 Model 3: transactional 1.657 2 .437 1.000 .000 (.000, .107) -2.34
115
5
¨&), ¨506($ ¨<%ʖ2 ¨df p
.000 .019 1.260 2 .533
.000 .000 7.308 2 .026
.000 .000 1.288 2 .525
.000 .011 2.614 2 .217
.001 .047 11.295 2 .005
.002 .000 .972 2 .615
Difference tests1
116
Parental Stress and Externalizing Problems
As was true for the baseline model for internalizing problems, the baseline model RQWKHFRQFXUUHQWUHODWLRQVEHWZHHQSDUHQWDOVWUHVVDQGH[WHUQDOL]LQJEHKDYLRUSUREOHPVÀW
WKH GDWD ZHOO <%ʖ2 (df = 6) = 11.986, p &), 506($ &,
>@7KHFRQVHFXWLYHVPRGHOZKHUHLQWKHFURVVODJJHGSDWKVEHWZHHQHLWKHUSDUHQW
child (model 2a) or child-parent (model 2b), and the transactional model, with cross-lagged SDWKVEHWZHHQERWKSDUHQWFKLOGDQGFKLOGSDUHQWDOVRÀWWHGWKHGDWDZHOO7DEOHVKRZV
WKH JRRGQHVVRIÀW VWDWLVWLFV IRU HDFK PRGHO 6LJQLÀFDQW FURVVODJJHG HIIHFWV ZHUH DJDLQ
RQO\SUHVHQWIRUFKLOGUHQ·VH[WHUQDOL]LQJSUREOHPVWRSDUHQWDOVWUHVVIURP:DYH,WR:DYH,,
(ɴ = .11, pDQGIURP:DYH,,WR:DYH,,,ɴ = .12, p < .05), but not for parental stress to FKLOGUHQ·VH[WHUQDOL]LQJSUREOHPV7DEOHJLYHVDQRYHUYLHZRIWKHVWDQGDUGL]HGFRHIÀFLHQWV
of each model. In line with the model on internalizing behaviors, chi-square difference tests LQGLFDWHGWKDWWKHPRGHOZLWKFURVVODJJHGSDWKVIURPFKLOGWRSDUHQWPRGHOEÀWWKHGDWD
EHWWHUWKDQWKHEDVHOLQHPRGHO¨<%ʖ2 (df = 2) = 11.295, p = .005), and better than the third model in terms of parsimony. Model 2b is depicted in Figure 5.3.
0XOWLJURXSDQDO\VHVZHUHSHUIRUPHGWRWHVWWKHVWDELOLW\RIWKHÀQGLQJV,QOLQHZLWK
the results of the entire group of foster children, for both younger and older children and for children with different lengths of stay in the foster family there were no parent-to-child HIIHFWVDQGPRGHOEÀWWHGEHVW)XUWKHUPRUHXQLGLUHFWLRQDOUHJUHVVLRQORDGLQJVIURPFKLOGUHQ
to parents were similar for these subgroups of foster children. However, for foster children who differed in their experience of placement history, the multi-group analyses showed two GLIIHUHQFHV)LUVWRIDOOWKHELGLUHFWLRQDOPRGHOPRGHOÀWWHGEHWWHUIRUWKHVXEJURXSRI
IRVWHUFKLOGUHQZKRKDGH[SHULHQFHGRQHRUPRUHSODFHPHQWV$OWKRXJKWKHUHZDVQRVLJQLÀFDQW
SDUHQWWRFKLOGHIIHFWIURP:DYH,WR:DYH,,WKHUHZDVDVLJQLÀFDQWSDUHQWWRFKLOGHIIHFW
IURP :DYH ,, WR :DYH ,,, ɴ = .20 p < .05) for foster children with a placement history.
6HFRQGO\XQLGLUHFWLRQDOUHJUHVVLRQORDGLQJVIURPFKLOGUHQWRSDUHQWVIRUPRGHOEGLIIHUHG
IRUIRVWHUFKLOGUHQZLWKGLIIHUHQWSODFHPHQWKLVWRULHV,QFRQWUDVWWRWKHÀQGLQJVRIWKHHQWLUH
JURXSFKLOGWRSDUHQWHIIHFWIURP:DYH,WR:DYH,,ZHUHQRWSUHVHQWIRUWKHJURXSZKRGLG
QRWH[SHULHQFHSUHYLRXVSODFHPHQWV6LPLODUWRWKHUHVXOWVRIWKHHQWLUHJURXSFKLOGWRSDUHQW
HIIHFWVIURP:DYH,,WR:DYH,,,ZHUHSUHVHQWIRUERWKJURXSV
117
5
Figure 5.3. Structural equation model (2b-model) with cross paths between children’s externalizing behavior problems and foster parents’ stress.
*p < .05
Parental stress Parental stress Parental stress
Internalizing Internalizing Internalizing
Time 1 Time 2 Time 3
.39*
.74*
.85*
.11*
.59*
.71*
.66*
.42*
.40* .12* .40*
118
Table 5.3. 6WDQGDUGL]HGFRHIÀFLHQWVIRUWKHVWUXFWXUDOHTXDWLRQPRGHOVRQWKHFRQFXUUHQWPRGHO parent-to-child (model 2a), child-to-parent (model 2b) and transactional relations (model 3) between parental stress and internalizing (int) and externalizing (ext) behavior problems.
Stability effects
T1 parental stress ї T2 parental stress T2 parental stress ї T3 parental stress T1 parental stress ї T3 parental stress T1 int/ext ї T2 int/ext
T2 int/ext ї T3 int/ext T1 int/ext ї T3 int/ext Concurrent effects
T1 parental stress ў T1 int/ext T2 parental stress ў T2 int/ext T3 parental stress ў T3 int/ext Cross-lagged effects
T1 parental stress ї T2 int/ext T2 parental stress ї T3 int/ext T1 int/ext ї T2 parental stress T2 int/ext ї T3 parental stress
*p < .05
119
Model 1 Model 2a Model 2b Model 3 Model 1 Model 2a Model 2b Model 3
.77* .78* .78* .78* .77* .78* .71* .72*
.50* .51* .47* .47* .47* .49* .42* .44*
.38* .37* .36* .35* .40* .39* .39* .38*
.77* .75* .76* .74* .84* .82* .85* .84*
.51* .49* .52* .51* .63* .60* .66* .64*
.38* .37* .37* .37* .25* .24* .24* .23*
.47* .47* .47* .47* .59* .59* .59* .59*
.34* .34* .34* .33* .40* .40* .40* .40*
.22* .22* .20* .20* .40* .41* .40* .39*
.04 .05 .04 .02
.03 .02 .07 .04
-.01 -.01 .11* .10
.13* .12* .12* .11*
Externalizing behavior problems Internalizing behavior problems
5
120
DISCUSSION
Children in foster care may experience internalizing and externalizing behavior SUREOHPV%XUQVHWDO&ODXVHQHWDO*RHPDQV9DQ*HHO 9HGGHU
These problems do not only interfere with their developmental trajectory and well-being, they also place foster children at increased risk for placement breakdown (Oosterman et al., 2007). In addition, the problematic development of foster children may strain foster parents, ZKRDVDUHVXOWH[SHULHQFHSDUHQWDOVWUHVV)DUPHUHWDO+XUOEXUWHWDO-RQHV
0RUULVVHWWH0F&DUWK\HWDO3UHYLRXVVWXGLHVLQWKHJHQHUDOSRSXODWLRQKDYH
VXJJHVWHGDWUDQVDFWLRQDOPRGHO6DPHURIIRIFRQWLQXRXVLQWHUDFWLRQVEHWZHHQFKLOGUHQ·V
SV\FKRVRFLDOIXQFWLRQLQJDQGSDUHQWDOVWUHVV%RUQVWHLQ1HHFHHWDO6WRQHHW
al., 2016). No studies to date have examined transactional relations between parental stress and children’s psychosocial development in foster care. The primary goal of this study was to test foster parents’ stress and foster children’s internalizing and externalizing behavior in a transactional framework by using cross-lagged structural equation modeling.
7KH UHVXOWV RI WKLV VWXG\ SUHVHQW D FOHDU SLFWXUH )RVWHU FKLOGUHQ LQÁXHQFH WKHLU
IRVWHUSDUHQWVEXWIRVWHUSDUHQWVGRQRWLQÁXHQFHWKHLUIRVWHUFKLOGUHQ%RWKIRVWHUFKLOGUHQ·V
LQWHUQDOL]LQJDQGH[WHUQDOL]LQJEHKDYLRUSUREOHPVDUHVLJQLÀFDQWO\UHODWHGWRIRVWHUSDUHQWV·
stress, with higher levels of behavior problems related to higher levels of parental stress. In other words: there is a unidirectional prospective relation from foster children’s internalizing DQG H[WHUQDOL]LQJ SUREOHPV WR IRVWHU SDUHQWV· VWUHVV ,Q FRQWUDVW WKHUH ZHUH QR VLJQLÀFDQW
prospective pathways from foster parents’ stress to foster children’s psychosocial functioning.
This does not mean that foster parents did not experience parental stress. In comparison with a VDPSOHRI'XWFKSDUHQWVIURPWKHJHQHUDOSRSXODWLRQ*RHPDQV9DQ*HHO9HGGHU %UDGOH\
2016), the foster parents in the current study clearly showed elevated stress levels. Although DSSUR[LPDWHO\RIWKHIRVWHUSDUHQWVLQWKHFXUUHQWVWXG\UHSRUWHGWRKDYHORZOHYHOVRI
parental stress, the same percentage of foster parents in our sample experienced stress levels DERYHWKHDYHUDJH1HYHUWKHOHVVIRVWHUSDUHQWV·VWUHVVDSSDUHQWO\GRHVQRWH[HUWDQLQÁXHQFH
on foster children’s psychosocial functioning. This result is in contrast with our hypothesis which was based on previously found transactional relations in the general population, wherein prospective pathways from parental stress to children’s behavioral functioning have been UHSRUWHG1HHFHHWDO6WRQHHWDO7UDQVDFWLRQDOSURFHVVHVFKDUDFWHULVWLFRI
children and their biological parents are apparently not characteristic of foster children and their foster parents with respect to parental stress and children’s behavioral functioning. A lack of transactional relations in foster care might be explained by processes characteristic of either the foster parents, the foster child, or maybe even both.
From the perspective of processes characteristic of the foster parents, it might be that foster parents simply do not express their parental stress in ways that affect their foster children.
Although foster parents in our study experienced more stress when compared to parents IURPWKHJHQHUDOSRSXODWLRQ*RHPDQV9DQ*HHO9HGGHU %UDGOH\IRVWHUSDUHQWV
might have the strength and ability to handle these levels of stress because they deliberately chose to be a foster parent, mostly for reasons which are intrinsic and altruistic (Rodger et al.,
121
5
2006), and because of the training they receive prior to the placement (Dorsey et al., 2008).
The fact that foster parents seemed not to express their parental stress in ways that affected WKHLUIRVWHUFKLOGUHQPLJKWUHÁHFWSURIHVVLRQDOLVPLQIRVWHUSDUHQWVDOORZLQJWKHPHYHQZKHQ
they experience stress, to realize a level of self-control or detachment that shelters children form the negative consequences of parental stress. Furthermore, this study only involved stable ORQJWHUPIRVWHUSODFHPHQWVDQGLWFRXOGEHWKDWWKLVSODFHPHQWVWDELOLW\UHÁHFWVVWUHVVOHYHOV
which can be controlled by foster parents and which are below levels connected to placement EUHDNGRZQ:KHUHDVZLWKRXWSURSHULQWHUYHQWLRQWKHVWUHVVH[SHULHQFHGE\ELRORJLFDOSDUHQWV
can develop toward ongoing levels of unbearable stress which are clearly evident for their FKLOGUHQDQGDOVRKDYHDGYHUVHHIIHFWVRQFKLOGUHQ·VGHYHORSPHQW%DNHUHWDO&UQLF
*D]H +RIIPDQ1HHFHHWDOIRVWHUSDUHQWVGRQRWQHHGWROHWWKLVKDSSHQ
7KH\KDYHWKHRSWLRQRIHQGLQJWKHIRVWHUSODFHPHQWDVDÀQDOHVFDSHIURPWKHLUVWUHVV)LVKHU
6WRROPLOOHU,WLVDOVRSRVVLEOHWKDWIRVWHUFKLOGUHQDUHQRWVXVFHSWLEOHWRIRVWHUSDUHQWV·
stress. Many children in foster care have previously experienced childhood adversities and KDYHKLVWRULHVRIQHJOHFWDQGDEXVH'XEQHU 0RWWD*UHHVRQHWDO2VZDOGHW
al., 2010). As a consequence, it might be that foster children are, in contrast to children from the general population, not affected by relatively minor sources of distress such as the stress expressed by their foster parents.
7KHRYHUDOOÀQGLQJVRIWKLVVWXG\VXSSRUWLQJWKHSUHVHQFHRIFKLOGWRSDUHQWHIIHFWV
and the absence of parent-to-child effects, are fairly stable across different subgroups of IRVWHU FKLOGUHQ +RZHYHU WKHUH DUH D IHZ GLIIHUHQFHV VXFK DV WKH ÀQGLQJ WKDW IRU \RXQJHU
foster children (<10 years) there were unidirectional prospective relations from internalizing EHKDYLRUVWRSDUHQWDOVWUHVVIURPERWK:DYH,WR:DYH,,DQGIURP:DYH,,WR:DYH,,,)RU
WKHHQWLUHJURXSXQGHUVWXG\WKHVHUHODWLRQVZHUHRQO\SUHVHQWIURP:DYH,,WR:DYH,,,,W
seems that the child-to-parent effects for children’s internalizing behaviors on foster parents’
stress are generally stronger for younger children. It might be that internalizing behaviors in younger children quickly cause additional stress for foster parents because these behaviors LQGLFDWHWKHYXOQHUDELOLW\DQGFRPSOH[VHUYLFHQHHGVRIWKHLU\RXQJIRVWHUFKLOG9LJ&KLQLW] 6KXOPDQ7KHVHEHKDYLRUVFDQQRWEHLJQRUHGRUVLPSO\SHUFHLYHGDVDJHDSSURSULDWH
as could be done with older foster children. Multi-group analyses also showed differences between foster children with different placement histories. Externalizing behaviors of foster children with one or more previous placements showed stronger child-to-parent effects than internalizing behaviors. This may be explained by the relatively strong stress inducing effects of externalizing problems on foster parents, which are even related to placement breakdown (Oosterman et al., 2007). Another interesting result for foster children with different placement histories is that the bidirectional model (the model with both child-to-parent and parent-to-child HIIHFWVÀWWHGEHWWHUIRUFKLOGUHQZLWKRQHRUPRUHSUHYLRXVSODFHPHQWVWKDQWKHXQLGLUHFWLRQDO
model with only the child-to-parent effects. It was found that for children with one or more SUHYLRXVSODFHPHQWVWKHUHZDVDVLJQLÀFDQWXQLGLUHFWLRQDOUHODWLRQIURPIRVWHUSDUHQWV·VWUHVV
DW:DYH,,WRIRVWHUFKLOGUHQ·VH[WHUQDOL]LQJEHKDYLRUVDW:DYH,,,7KLVVHHPVWRFRQFXUZLWK
a previous study showing that behavior problems can be both a cause and consequence RI SODFHPHQW GLVUXSWLRQ 1HZWRQ /LWURZQLN /DQGVYHUN )RVWHU FKLOGUHQ ZKR KDYH
previously experienced placement breakdown more often show behavior problems (Newton
122
et al., 2000). The current study showed that foster children’s externalizing behaviors results in increased stress in their foster parents, which might in turn place them at risk for another SUHPDWXUHO\HQGHGIRVWHUSODFHPHQW7KHVHÀQGLQJVVWUHVVWKHLPSRUWDQFHRIHDUOLHUYRLFHG
warnings for professionals to be alert to foster parents’ stress and the risk of breakdown in case of a placement of a foster child with a history of previous placements.
Despite the general lack of support for the bidirectional model, this study did show moderate concurrent relations between foster parents’ stress and foster children’s behavioral problems. Furthermore, not only were parental stress levels moderately stable over time, this was also true for foster children’s behavioral problems. This suggests that on average parental stress and children’s psychosocial functioning do not necessarily improve over time in foster families (see also Goemans et al., 2015). The good news is that this means that behavior problems and parental stress neither get worse, but given the fact that the mean level of behavior problems fell within the borderline range, caution is still needed. Foster care professionals should make sure that in case they signal either children’s behavior problems or caregiver stress, they organize support in order to ensure the wellbeing of both foster children DQGIRVWHUSDUHQWV6XFKLQLWLDWLYHPLJKWQHHGWRVWDUWZLWKDFRPSUHKHQVLYHDVVHVVPHQW/HVOLH
HWDO5RPDQHOOLHWDOWRDGGUHVVWKHSUREOHPRIWKHSUHYLRXVO\UHSRUWHGXQPHW
QHHGV IRU VXSSRUW DQG WUDLQLQJ IRU IRVWHU SDUHQWV DQG IRVWHU FKLOGUHQ %XUQV HW DO
Murray et al., 2011). Comprehensive assessment is an important step to meet these needs and WRLQGLFDWHWDUJHWHGVXSSRUWVHUYLFHVDQGHYLGHQFHEDVHGLQWHUYHQWLRQV)LVKHU 6WRROPLOOHU
/LQDUHV0RQWDOWR/L 2]D6SUDQJ7LPPHUHWDO9DQ$QGHOHW
DO:KHQDQ2[ODG /XVKLQJWRQ
Limitations and Implications for Future Research
This study longitudinally examined the development of children in foster care, though several limitations should be noted. Information for this study is derived from the reports of the foster parent. As a consequence, same method variance might result in an overestimation of the DVVRFLDWLRQVEHWZHHQWKHYDULDEOHVRILQWHUHVW.HLMVHUVHWDO2·&RQQRU)XWXUH
UHVHDUFKVKRXOGWKHUHIRUHPDNHDQHIIRUWWRLQFOXGHIRVWHUFKLOGUHQ·VSHUVSHFWLYHV-RKQVRQ
<RNHQ 9RVV DQG RWKHU LQIRUPDQWV VXFK DV WHDFKHUV DQG SURIHVVLRQDOV WR H[DPLQH
whether the results of the current study can be replicated. Another limitation is that the scores on psychosocial functioning and parental stress did not appear to change over time and which PD\FDXVHGLIÀFXOWLHVLQLGHQWLI\LQJDORQJLWXGLQDOHIIHFW$GGLWLRQDOPHDVXUHVHVSHFLDOO\WKRVH
XVLQJRWKHUPHWKRGVHJLQWHUYLHZVRUREVHUYDWLRQVDQGPRUHVSHFLÀFPHDVXUHVIRUFDSWXULQJ
WKHG\QDPLFQDWXUHRISRWHQWLDOWUDQVDFWLRQDOUHODWLRQVDUHUHFRPPHQGHG0DFFRE\
5RWKEDXP :HLV]VXFKDVSDUHQWFKLOGREVHUYDWLRQVDQGVWDWHVSDFHJULGV*UDQLF
+ROOHQVWHLQ+ROOHQVWHLQ*UDQLF6WRROPLOOHU 6Q\GHU7KLVLGHDOO\UHTXLUHV
longitudinal studies with more measurement points and over longer periods of time, such as performed by Neece et al. (2012).
Another limitation is that this study solely focused on foster parents’ stress and foster children’s psychosocial functioning, and did not examine possible moderating and mediating factors. Although there are several factors which might have an effect on the relation between parental stress and children’s behavioral problems (e.g., foster children’s attachment,
123
5
susceptibility, or foster parents’ parenting style), social support has been frequently found to be related to parental stress and children’s development, also for foster children and foster SDUHQWV&RROH\)DULQHDX 0XOOLV)DUPHUHWDO.HOOH\HWDO5RUN
6ROLGD\0F&OXVNH\)DZFHWW 0HFNDQGDOVRLQORQJLWXGLQDOVWXGLHV*OHHVRQ+VLHK
&U\HU&RXSHW,WKDVEHHQVXJJHVWHGWKDWVXSSRUWKHOSVIRVWHUSDUHQWVWRGHDOZLWK
WKHVWUHVVH[SHULHQFHGE\WKHGLVUXSWLYHEHKDYLRUVRIWKHLUIRVWHUFKLOG&RROH\HWDO
Farmer et al., 2005). It should be noted that foster children in our study were on average more WKDQÀYH\HDUVLQWKHLUFXUUHQWSODFHPHQWZKLFKVXJJHVWVDFHUWDLQOHYHORIVWDELOLW\%HFDXVH
higher levels of reported strain were found to be related with placement instability (Farmer et al., 2005), it might be that our sample received relatively high levels of support which can be an explanation for the lack of reciprocal relations between foster parents’ stress and foster children’s internalizing and externalizing behaviors. However, future research should examine the role of social support, also in newly placed foster children, in order to get a better insight ZKHWKHUVRFLDOVXSSRUWEXIIHUVDJDLQVWWKHLPSDFWRISDUHQWLQJVWUHVVDQGFKLOGUHQ·VGLIÀFXOW
behavior.
A third limitation is the considerable amount of attrition. Despite the use of reminders and incentives, we could not prevent dropout of more than half of our original sample. Attrition is a common problem in longitudinal research, and longitudinal studies on foster care are not DQH[FHSWLRQ-DFNVRQHWDO6WUDWHJLHVIRUORQJLWXGLQDOUHVHDUFKZLWKIRVWHUFKLOGUHQDV
GHVFULEHGE\-DFNVRQHWDODUHKHOSIXOLQWKHGHVLJQRIORQJLWXGLQDOVWXGLHVDLPLQJIRU
fewer attrition. Furthermore, researchers should be transparent in reporting about their missing data and apply modern methods to handle missing data (Graham, 2009), such as multiple imputation or FIML estimation.
Conclusion
This study showed the absence of reciprocal relations between foster parents’ stress and IRVWHUFKLOGUHQ·VLQWHUQDOL]LQJEHKDYLRUSUREOHPV)RVWHUFKLOGUHQLQJHQHUDOVHHPWRLQÁXHQFH
their foster parents, but not vice versa. This can be considered as something positive: Although increased foster children’s behavioral problems do heighten foster parents’ stress, this has no continuation in a downward spiral: elevated levels of parents’ stress do not further increase children’s internalizing and externalizing behaviors. In other words, foster parents’ stress does QRW LQÁXHQFH WKHLU IRVWHU FKLOGUHQ·V SV\FKRVRFLDO IXQFWLRQLQJ $QRWKHU SRVLWLYH ÀQGLQJ LV WKDW
DSSUR[LPDWHO\RIWKHIRVWHUFKLOGUHQGLGQRWKDYHSUREOHPDWLFOHYHOVRISV\FKRVRFLDO
problems. However, the lack of transactional relations between foster children and foster parents is surprising given the presence of transactional relations in biological parent-child G\DGV)RVWHUSDUHQWVVHHPQRWWRLQÁXHQFHWKHLUIRVWHUFKLOGUHQZKHQLWFRPHVWRUHJXODWLQJ
problem behavior and the question is whether foster parents, more generally can make their IRVWHU FKLOGUHQ EHQHÀW IURP WKHLU LPSURYHG KRPH HQYLURQPHQW *LYHQ WKH ÀQGLQJ WKDW PDQ\
IRVWHUFKLOGUHQVWLOOH[SHULHQFHEHKDYLRUDOGLIÀFXOWLHVIXWXUHORQJLWXGLQDOVWXGLHVDUHQHHGHGWR
examine how to boost foster children’s psychosocial development.