This is a sel -a chi ed e sion o an o iginal a icle. This e sion
may di e om he o iginal in pagina ion and ypog aphic de ails.
Au ho (s):
Ti le:
Yea :
Ve sion:
Copy igh :
Righ s:
Righ s u l:
Please ci e he o iginal e sion:
CC BY 4.0
h ps://c ea i ecommons.o g/licenses/by/4.0/
Gene ics o Pe cei ed Family In e ac ion F om 12 o 17 Yea s o Age
© Sp inge Science+Business Media, LLC, pa o Sp inge Na u e 2019
Published e sion
Sil en oinen, Ka i; Su, Jinni; Pulkkinen, Lea; Ba , Pe e ; Rose, Richa d J.; Dick,
Danielle M.; Kap io, Jaakko
Sil en oinen, K., Su, J., Pulkkinen, L., Ba , P., Rose, R. J., Dick, D. M., & Kap io, J. (2019). Gene ics
o Pe cei ed Family In e ac ion F om 12 o 17 Yea s o Age. Beha io Gene ics, 49(4), 366-375.
h ps://doi.o g/10.1007/s10519-019-09960-z
2019
Vol:.(1234567890)
Beha io Gene ics (2019) 49:366–375
h ps://doi.o g/10.1007/s10519-019-09960-z
1 3
ORIGINAL RESEARCH
Gene ics o Pe cei ed Family In e ac ion F om 12 o17 Yea s o Age
Ka iSil en oinen1 · JinniSu2· LeaPulkkinen3· Pe e Ba 2· Richa dJ.Rose4· DanielleM.Dick2,5,6·
JaakkoKap io7,8
Recei ed: 26 Oc obe 2018 / Accep ed: 13 May 2019 / Published online: 24 May 2019
© Sp inge Science+Business Media, LLC, pa o Sp inge Na u e 2019
Abs ac
We analyzed how he e ec s o gene ic and en i onmen al ac o s on he pe cep ions o amily in e ac ion change om ea ly
o la e adolescence. The da a we e collec ed by pos al su eys on Finnish wins (N = 4808) a 12, 14 and 17yea s o age
and analyzed using gene ic win modeling. Addi i e gene ic ac o s explained a modes sha e o he a ia ion in pe cei ed
ela ional suppo (a2 = 0.30 in boys and 0.18 in gi ls) and ela ional ensions (a2 = 0.13 and 0.14, espec i ely) a 12yea s
o age, wi h he p opo ions becoming la ge h ough 17yea s o age (a2 = 0.53 in boys and 0.49 in gi ls o ela ional sup-
po ; a2 = 0.35 in boys and 0.33 in gi ls o ela ional ensions). Simul aneously, he ole o en i onmen sha ed by co- wins
dec eased. These indings sugges ha he associa ions be ween pe cei ed amily in e ac ion and o he ac o s in adul hood
should be in e p e ed wi h cau ion, because hey pa ly e lec gene ic backg ound, whe eas in childhood, hey may p o ide
mo e eliable in o ma ion on pa en al cha ac e is ics.
Keywo ds Family in e ac ion· Adolescen s· Twins· Gene ics
In oduc ion
P e ious s udies ha e shown ha he psychosocial amily
en i onmen has impo an in luences on o sp ing, includ-
ing e ec s on men al heal h (Yap and Jo m 2015), heal h
beha io s such as physical ac i i y (Bee s e al. 2010), and
o he heal h- ela ed ou comes such as body mass index
(BMI) (Sokol e al. 2017). The e is con incing e idence,
howe e , ha he psychosocial amily en i onmen canno
be conside ed as a pu ely en i onmen al ac o , since i also
e lec s he gene ic backg ound o he child. A e iew by
Klah and Bu (2014) demons a ed ha he amily en i on-
men , as epo ed by child en, shows mode a e he i abili y
when es ima ed by using a gene ic win design. The psy-
chosocial amily en i onmen has many aspec s and can be
de ined in di e en ways, bu he quali y o he pa en –child
ela ionship and how o sp ing pe cei e hei amily a mos-
phe e ha e been ega ded as pa icula ly impo an and a e
used especially in he p e ious beha io al gene ic s udies
(Klah and Bu 2014). Thus, a pa o he a ia ion in he
amily en i onmen as measu ed by pa en ing and amily
in e ac ion as pe cei ed by child en is because o in e -indi-
idual gene ic di e ences.
The e a e a leas h ee possible mechanisms explain-
ing his gene ic componen behind he pe cei ed amily
Handling Edi o : Yoon-Mi Hu , PhD.
Elec onic supplemen a y ma e ial The online e sion o his
a icle (h ps ://doi.o g/10.1007/s1051 9-019-09960 -z) con ains
supplemen a y ma e ial, which is a ailable o au ho ized use s.
* Ka i Sil en oinen
ka i.sil [email p o ec ed]
1 Depa men o Social Resea ch, Uni e si y o Helsinki, P.O.
Box18, FIN-00014Helsinki, Finland
2 Depa men o Psychology, Vi ginia Commonweal h
Uni e si y, Richmond, VA, USA
3 Depa men o Psychology, Uni e si y o Jy askyla,
Jy askyla, Finland
4 Depa men o Psychological andB ain Sciences, Indiana
Uni e si y, Blooming on, IN, USA
5 Depa men o Human andMolecula Gene ics, Vi ginia
Commonweal h Uni e si y, Richmond, VA, USA
6 College Beha io al andEmo ional Heal h Ins i u e, Vi ginia
Commonweal h Uni e si y, Richmond, VA, USA
7 Ins i u e o Molecula Medicine (FIMM), Uni e si y
o Helsinki, Helsinki, Finland
8 Depa men o Public Heal h, Uni e si y o Helsinki,
Helsinki, Finland
367Beha io Gene ics (2019) 49:366–375
1 3
en i onmen . Fi s , he e is s ong e idence based on bo h
win and molecula gene ic s udies ha he gene ic makeup
o pa en s a ec s hei beha io and in u n he amily en i-
onmen (Mile a-Sei z e al. 2016). Because child en inhe i
genes om hei pa en s, a co ela ion be ween he geno ype
o child en and hei amily en i onmen esul s, which is
called he passi e gene–en i onmen co ela ion (Ja ee and
P ice 2007). Second, child en can e oke eac ions om hei
pa en s due o hei pe sonali y and o he ac o s ha a e
gene ically in luenced. Fo example, a Swedish win s udy
ound ha ex e nalizing p oblems in child en, which showed
s ong he i abili y, e oked c i icism om hei mo he s
(Na usy e e al. 2011). Finally, child en may expe ience he
same en i onmen al exposu es di e en ly because o di -
e ences in pe sonali y. Since gene ic ac o s a e known o
explain indi idual di e ences in pe sonali y in childhood
(Spengle e al. 2012), di e en pe cep ions o he amily
en i onmen can gene a e gene ic di e ences in he pe -
cei ed amily en i onmen .
The he i abili y es ima es o he pe cei ed amily en i-
onmen may change, howe e , om childhood o adul hood.
Based on he p e ious li e a u e, adolescence is cha ac e -
ized by dec easing dependence on pa en s, widening social
ne wo ks, s onge in luence o pee s and inc eased sen-
sa ion seeking which may lead o inc eased isk beha io
(Ahmed e al. 2015; Kil o d e al. 2016). These changes can
modi y he he i abili y es ima es o he pe cei ed amily
en i onmen h ough changing gene–en i onmen co ela-
ions. A p e ious me a-analysis o mainly c oss-sec ional
s udies on he he i abili y o he amily en i onmen ound
ha when based on child en’s own epo s, he in luence
o en i onmen al ac o s sha ed by co- wins dec eased
and he in luence o en i onmen al ac o s unique o each
win inc eased du ing aging; he e ec o gene ic ac o s
emained oughly s able (Klah and Bu 2014). This esul
is consis en wi h he dec easing in luence o pa en s and
inc easing in luence o pee s du ing adolescence. Howe e ,
he e is also e idence ega ding physiological ai s such as
BMI (Sil en oinen e al. 2016) and many psychological ai s
such as in elligence and men al heal h indica o s (Be gen
e al. 2007) ha he in luence o gene ic ac o s inc eases
du ing adolescence. A possible eason is ha he in luence
o he passi e gene–en i onmen co ela ion dec eases and
is pa ly eplaced by he ac i e o e oca i e gene–en i on-
men co ela ions when child en ac i ely shape hei en i-
onmen o e oke eac ions om o he pe sons such as hei
pee s (Ma ceau e al. 2016). This p ocess is demons a ed
by mul iple s udies showing ha en i onmen al exposu es
show mode a e he i abili y (Kendle and Bake 2007). Fu -
he , because o inc easing independence, pe sonali y ac-
o s may mo e s ongly shape how adolescen s in e p e
en i onmen al exposu es as compa ed o younge child en.
Thus, i could also be expec ed ha he in luence o gene ic
ac o s on how adolescen s expe ience he amily en i on-
men inc eases du ing his de elopmen al pe iod.
We a e awa e o wo longi udinal win coho s which ha e
examined he ole o gene ic ac o s on he pe cei ed amily
en i onmen o e his ansi ional pe iod, bu hese s udies
ha e p oduced somewha inconsis en esul s. Bo h o he
s udies we e based on o sp ing a ings o he amily en i-
onmen , bu he measu emen ins umen s used di e ed. A
US win s udy ound ha he ole o gene ic ac o s inc eased
om 11 o 14yea s o age in boys and gi ls using a pa -
en –child ela ionship ques ionnai e (McGue e al. 2005). A
ollow-up s udy on his same coho ound ha he gene ic
a ia ion inc eased and he sha ed en i onmen al a ia ion
dec eased un il 17yea s o age; his inc easing gene ic a i-
a ion was s ongly co ela ed wi h gene ic a ia ion al eady
p esen a 12yea s o age (Ludeke e al. 2013). In con as ,
a UK win s udy using h ee measu es o he amily en i-
onmen (household chaos, pa en al discipline and pa en al
eelings) ound no sys ema ic change in he ole o gene ic
and sha ed en i onmen al ac o s om nine un il 16yea s o
age (Hannigan e al. 2017a). Fu he , in his UK s udy, he
gene ic co ela ions be ween he ages we e low. Thus, mo e
longi udinal esea ch is needed o cla i y how gene ic and
en i onmen al in luences on amily in e ac ion may change
ac oss de elopmen .
In his s udy, we analyzed how he gene ic a chi ec u e o
pe cei ed amily in e ac ion changes om 12 o 17yea s o
age by using a longi udinal Finnish win da ase . Two indica-
o s o amily in e ac ion, ela ional suppo and ela ional
ensions, we e used. We ocused on hese wo dimensions o
amily in e ac ion because hey cap u e posi i e and nega-
i e aspec s o amily in e ac ion and ha e been shown o
be p edic i e o adolescen psychosocial ou comes (La en-
d esse e al. 2010; Sil en oinen e al. 2014). We analyzed
bo h he change in gene ic and en i onmen al a ia ion o
amily in e ac ion and how gene ic and en i onmen al ac-
o s explained he s abili y o he pe cep ion o amily in e -
ac ion om ea ly o la e adolescence.
Da a andme hods
The s udy coho was de i ed om he longi udinal
FinnTwin12 s udy co e ing all Finnish wins bo n in
1983–1987 (Kap io e al. 2002). The names and pos al
add esses o he wins and hei pa en s we e ecei ed
om he Finnish popula ion egis y by iden i ying ami-
lies wi h wo child en bo n o he same mo he on he
same day (3136 win amilies, supplemen a y Fig.1). In
esponse o an in i a ion o ake pa in he s udy, 86% o
amilies indica ed hei willingness and e u ned a ques ion-
nai e (on he bi h, childhood and ea ly school yea s o he
wins) and consen o m. A e e u n o he ques ionnai e,
368 Beha io Gene ics (2019) 49:366–375
1 3
u he baseline ques ionnai es we e sen o he pa en s and
wins. The baseline win ques ionnai e was sen o bo h co-
wins indi idually in he au umn o he yea hey eached
he age o 11–12yea s (mean age: 11.42yea s; ange:
11.41–11.43yea s). Zygosi y was de e mined based on
ques ionnai e i ems on he physical simila i y and con us-
abili y o appea ance a school age and, i needed, was sup-
plemen ed by school pho og aphs and ques ions o pa en s.
The eliabili y o his me hod was alida ed in a sample o
295 same-sex win pai s using DNA; zygosi y was con-
i med among 97% o he pai s, showing good eliabili y
o his me hod (Jelenko ic e al. 2011). A e emo ing 264
wins wi h unknown zygosi y and 226 wins who we e no
a ailable o did no espond o he su ey, we had 4920 alid
esponses (49% gi ls), including 2456 comple e win pai s
om which 34% we e monozygo ic (MZ), 33% same-sex
dizygo ic (SSDZ) and 33% opposi e-sex dizygo ic wins
(OSDZ). This ep esen s 78% o all Finnish wins in hese
bi h coho s. Simul aneously when he i s ques ionnai e
o wins was sen , a ques ionnai e was sen o hei pa en s
asking o beha io al a ings o he wins and amily in e -
ac ions. This amily ques ionnai e yielded alid esponses
om 2320 amilies (74% o all amilies). A second su ey,
a he mean age o 14.0yea s ( ange: 13.9–14.9yea s; 4523
alid esponses; 72% o all wins), was sen o hose wins
who esponded o he i s su ey and a hi d su ey, a he
mean age o 17.6yea s ( ange 17.2–19.5yea s; 4041 alid
esponses; 64% o all wins), was sen o hose wins who
esponded o he second su ey.
I ems in he amily in e ac ion ques ionnai e asked
whe he he amily was (1) wa m, ca ing; (2) c ea i e, sup-
po i e; (3) us ing, unde s anding; (4) open; (5) s ic ; (6)
unjus ; (7) con lic ed; and (8) indi e en . A 5-poin scale
(1 = i s comple ely, 2 = mainly, 3 = somewha , 4 = mainly
no , 5 = no a all) was used (Na usk and Pulkkinen 1994).
The a ings we e e-coded such ha o all a iables highe
alues indica e be e amily in e ac ions. The same ques-
ions we e used in he ques ionnai es sen o wins a 12,
14 and 17yea s o age as well as o pa en s when hei win
child en we e 12yea s o age. Twins we e ins uc ed o ill
in he ques ionnai e independen ly, bu pa en s we e asked
o do i oge he . In mos o hese amilies, he wins’ mo he
(60%) o mo he and a he join ly (35%) comple ed his
amily ques ionnai e. Thus, in nea ly all cases, he mo he
had an impo an ole in he a ings, bu he a he may ha e
also con ibu ed o hese a ings. The pa en s we e asked
o conside all child en when making he a ings; hus, he
pa en al a ings a e he same o bo h co- wins.
The ini ial ac o analyses showed a 2- ac o solu ion
(supplemen a y Table1). A all ages and in bo h boys and
gi ls, he eigen alues dec eased less han one o he hi d
ac o ; oge he , he i s wo ac o s explained 54–65% o
he o al a ia ion. When we analyzed he ac o loadings
o he un- o a ed solu ions, we ound ha i ems 5–8 loaded
nega i ely on he i s ac o , indica ing ha hese i ems c e-
a e ano he dimension (supplemen a y Table2). The ac o
loadings we e la gely simila a all ages, o bo h boys and
gi ls and when using ei he o sp ing o pa en al epo s. To
c ea e mo e in e p e able esul s, in he u he analyses we
conduc ed wo sepa a e 1- ac o solu ions: o i ems 1–4 we
in e p e o indica e ela ional suppo and 5–8 o indica e
ela ional ensions. This s a egy allowed us o analyze he
co ela ions be ween hese wo dimensions o amily in e -
ac ion. When using hese wo 1- ac o solu ions, he ac o s
explained 55–73% o he a ia ion; i em 5 (s ic ) did no i
well on ei he measu e educing he explained a ia ion by
10–14%, and was hus excluded om u he analyses (sup-
plemen a y Table3).
We hen calcula ed he ac o sco es o ela ional sup-
po and ela ional ensions sepa a ely by using he 1- ac-
o solu ions such as also in he p e ious s udies using his
ques ionnai e (La end esse e al. 2010; Sil en oinen e al.
2014). The C onbach α- alues a ied be ween 0.71 and
0.87 o ela ional suppo and be ween 0.57 and 0.69 o
ela ional ensions, sugges ing good in e nal consis ency o
hese measu es (supplemen a y Table3). The sys ema ically
lowe C onbach α- alues o ela ional ensions was a ec ed
by a smalle numbe o i ems han was used o ela ional
suppo because o he emo al o one i em (s ic ). In he
desc ip i e analyses, we used sum a iables o show how
he pe cei ed amily in e ac ion changed om 12 o 17yea s
o age. The sco es a ied om 4 o 20 o ela ional sup-
po and 3–15 o ela ional ensions. Howe e , o make he
esul s compa able, we ans o med bo h o a y be ween
0 and 100 whe e highe alues indica e be e suppo and
less ensions. All wins ha ing missing o any i em we e
emo ed om he analyses o ela ional suppo (i ems 1–4)
and ela ional ensions (i ems 6–8). The numbe o alid
measu es in wins dec eased om 4799 o ela ional en-
sions and 4808 o ela ional suppo a 12yea s o age o
3997 and 3988, espec i ely, a 17yea s o age (supplemen-
a y Fig.1). The numbe o comple e win pai s ha ing hese
measu es a each age by sex and zygosi y a e a ailable in
supplemen a y Table4.
The da a we e analyzed using gene ic win modeling
based on he compa isons o simila i y be ween MZ and
dizygo ic (DZ) wins. MZ wins ha e i ually he same gene
sequence whe eas DZ wins sha e, on a e age, 50% o hei
gene ic a ia ion (Pos huma e al. 2003). Uni a ia e models
we e i s used o decompose he ai a ia ion in pe cei ed
amily in e ac ions in o h ee a iance componen s: addi i e
gene ic ac o s (A) including he e ec s o all loci on he ai
(co ela ion 1.0 wi hin MZ and 0.5 wi hin DZ co- wins),
sha ed en i onmen (C) including he e ec s o all en i-
onmen al ac o s making co- wins simila (co ela ion 1.0
wi hin bo h MZ and DZ co- wins) and unique en i onmen
369Beha io Gene ics (2019) 49:366–375
1 3
(E) including he e ec s o all en i onmen al ac o s mak-
ing co- wins di e en (co ela ion 0 bo h wi hin MZ and
DZ co- wins), along wi h measu emen e o (Fig.1a). We
ound s a is ically signi ican co ela ions be ween age and
ela ional suppo a 14yea s o age ( = − 0.05; p = 0.021),
whe eas he co ela ion was ma ginally signi ican o ela-
ional suppo a 12yea s (p = 0.078) and ela ional en-
sions a 14yea s o age (p = 0.054). Howe e , o make he
esul s sys ema ic, we adjus ed he esul s o exac age a all
measu emen s because o he wise he age e ec would ha e
been modeled as a pa o sha ed en i onmen al a ia ion.
DZ co ela ions we e sys ema ically highe han hal o he
MZ co ela ions, sugges ing he p esence o sha ed en i-
onmen al ac o s (supplemen a y Table4). Thus, we used
an addi i e gene ic/sha ed en i onmen al/unique en i on-
men al (ACE) model in he analyses. The model i s a is ics
o he gene ic win models a e p esen ed in supplemen a y
Table5. The assump ions o gene ic win models applied
well, as seen in he good i o he ACE models as compa ed
wi h he sa u a ed models; he model i di e ence was only
s a is ically signi ican o ela ional ensions a 17yea s o
age (p = 0.013), bu his can also be due o mul iple es -
ing since i is la ge han he Bon e oni co ec ed p alue
(p = 0.008 o 6 es s). We only ound s a is ically signi ican
sex-speci ic gene ic e ec s o ela ional suppo (p = 0.046)
and ensions (p = 0.009) a 12yea s o age, bu he OSDZ
co ela ions we e also somewha lowe han he SSDZ co e-
la ions a he o he ages, suppo ing he sex-speci ic gene ic
e ec (supplemen a y Table4). Thus, o ha e he in e nally
consis en esul s, we allowed sex-speci ic gene ic e ec s a
all ages. Di e ences be ween boys and gi ls we e highly s a-
is ically signi ican excep o ela ional ensions a 12yea s
o age. Thus, we s a i ied all u he analyses by sex.
Following he uni a ia e models, we decomposed he
co a ia ion be ween he wo dimensions o amily in e -
ac ion a same ages as well as o each measu e be ween
di e en ages by using bi a ia e Cholesky decomposi ion
(Fig.1b). This me hod makes no assump ions abou he
unde lying gene ic a chi ec u e bu decomposes he a i-
ance and co- a iance in he da a in o a se ies o unco ela ed
gene ic and en i onmen al ac o s. Using his me hod, we
calcula ed gene ic and en i onmen al co ela ions be ween
ela ional ensions and ela ional suppo a each age (c oss-
ai co ela ions) as well as be ween hese measu es a di -
e en ages (c oss-age co ela ions). Fu he , we calcula ed
how much sha ed gene ic and en i onmen al ac o s explain
he o al co ela ions.
The gene ic win models we e i ed using he OpenMx
package, e sion 3.0.2, o R s a is ical so wa e (Neale e al.
2016). The maximum likelihood es ima ion was used o es i-
ma e he pa ame e s and hei 95% con idence in e als (CI).
Fo desc ip i e s a is ics, s a is ical es s we e pe o med
using linea eg ession models by S a a/SE 13.1 o Win-
dows s a is ical so wa e (S a aCo p, College S a ion, TX,
USA). In he s a is ical es s, he e ec o in a-pai co -
ela ions on s anda d e o s (i.e. sampling win pai s a he
han independen indi iduals) was aken in o accoun by he
clus e op ion in S a a (Williams 2000).
Resul s
Table1 p esen s he desc ip i e s a is ics o amily in e -
ac ion by sex and zygosi y. The o sp ing a ings o ela-
ional suppo declined om 12 o 17yea s o age by 8.3
poin s (95% CI 7.4–9.3) in boys and by 10.8 poin s (95%
CI 9.8–11.7) in gi ls. A he same ime, he s anda d de ia-
ions (SD) o ela ional suppo inc eased in boys and gi ls.
Fo ela ional ensions, he esul s we e less sys ema ic:
while in gi ls he a ings dec eased by 4.2 poin s (95% CI
3.2–5.3), in boys he di e ence was no s a is ically signi i-
can (p = 0.80). Some dec ease in SD o ela ional ensions
we e seen in boys and gi ls. Pa en s epo ed less ela ional
ensions [7.8 poin s (95% CI 7.0–8.7) in boys and 5.3 poin s
(95% CI 4.4–6.2) in gi ls] bu also epo ed ela ional sup-
po o be lowe [2.0 poin s (95% CI 1.2–2.7) in boys and
3.0 poin s (95% CI 2.2–3.8) in gi ls] han hei win chil-
d en epo ed a 12yea s o age. MZ wins expe ienced
A
C
E
ACEACE
RT1RS1
(A)
(B)
=1 MZ / 0.5 DZ
=1 MZ and DZ
=0 MZ and DZ
ace
ACE
ace
ACE
RT1RT2
Fig. 1 Analy ical models: a Uni a ia e addi i e gene ic (A), sha ed
en i onmen (C) and unique en i onmen (E) model o ela ional en-
sions (RT1 o i s and RT2 o second win); b Bi a ia e Cholesky
decomposi ion (p esen ed o one win only) o addi i e gene ic co -
ela ion ( A), sha ed en i onmen al co ela ion ( C) and unique en i-
onmen al co ela ion ( E) be ween RT and ela ional suppo (RS)

370 Beha io Gene ics (2019) 49:366–375
1 3
sligh ly be e ela ional suppo and less ela ional en-
sions han DZ wins. The di e ences we e s a is ically
signi ican (0.71–1.3 poin s; p- alues 0.002–0.039 when
adjus ed o sex) excep o ela ional ensions a 12yea s
o age (p = 0.065). Gi ls epo ed be e ela ional suppo
(1.7 poin s 95% CI 0.94–2.52) and less ela ional ensions
(2.5 poin s 95% CI 1.6–3.5) a 12yea s o age han boys.
Howe e , his di e ence disappea ed a 14yea s o age,
and a 17yea s o age, gi ls epo ed less ela ional suppo
(1.3 poin s 95% CI 0.2–2.4) and mo e ela ional ensions
(2.1 poin s 95% CI 1.1–3.1) han boys. Only mino di e -
ences we e ound in SD be ween MZ and DZ wins. In he
compa isons o SSDZ and OSDZ wins, bo h means and
SDs we e e y simila (da a no shown bu a e a ailable
on eques ). The pa en al a ings o amily in e ac ion when
hei o sp ing we e 12yea s o age co ela ed modes ly
wi h he o sp ing a ings a he same age [ = 0.32 (95%
CI 0.29–0.34) o ela ional suppo and = 0.24 (95% CI
0.22–0.27) o ela ional ensions]. These co ela ions some-
wha dec eased when using o sp ing a ings a 14 [ = 0.27
(95% CI 0.25–0.30) and = 0.20 (95% CI 0.18–0.23), espec-
i ely] and 17yea s o age [ = 0.20 (95% CI 0.17–0.23) and
= 0.17 (95% CI 0.13–0.20), espec i ely], bu he dec ease
was no subs an ial.
Table2 p esen s he esul s o uni a ia e modeling
(Fig.1a). We allowed he means o di e be ween MZ
and DZ wins in he gene ic win models because o he
abo emen ioned zygosi y mean di e ences. In ela ional
suppo , addi i e gene ic, sha ed en i onmen al and unique
en i onmen al ac o s explained oughly equal sha es o
he a ia ion a 12yea s o age in boys, whe eas in gi ls,
sha ed en i onmen al ac o s we e mo e impo an and
explained nea ly hal o he a ia ion. The ole o gene ic
ac o s became mo e impo an o e ime, explaining a ound
hal o he a ia ion o ela ional suppo a 17yea s o age
in boys and gi ls. In ela ional ensions o bo h boys and
gi ls, he ole o gene ic ac o s was less impo an han
sha ed and unique en i onmen al ac o s a 12yea s o
age. Howe e , gene ic ac o s became mo e impo an o e
ime and a 17yea s o age explained oughly one- hi d o
he a ia ion. The e was some o e lap in he 95% CIs o
pa ame e s, bu gene ally he di e ences be ween he ages
we e s a is ically highly signi ican o ela ional suppo
(Δ-2LL = 63.4, Δd. . = 12, p < 0.00001) and ela ional en-
sions (Δ-2LL = 50.4, Δd. . = 12, p < 0.00001).
Finally, we analyzed he co ela ions be ween he o -
sp ing a ings o ela ional suppo and ela ional ensions
(c oss- ai co ela ions) as well as how hese dimensions
o amily in e ac ion co ela e be ween ages (c oss-age co -
ela ions) (Table3). As expec ed, he c oss-age co ela ions
we e highes be ween he closes ages (12 s. 14yea s and
14 s. 17yea s) bu emained mode a e ( ≤ 0.42). Gene ally,
he co ela ions we e simila in boys and gi ls. The co ela-
ions be ween ela ional suppo and ela ional ensions we e
lowes a 12yea s o age, inc easing un il 17yea s o age o
0.54 in boys and 0.65 in gi ls. When we decomposed hese
co ela ions in o gene ic and en i onmen al co ela ions
using Cholesky decomposi ion (Fig.1b), he highes co -
ela ions we e gene ally ound o sha ed en i onmen al ac-
o s. Mos o hese exceeded 0.60, and some exceeded 0.90.
Sha ed en i onmen al ac o s also explained an impo an
sha e o he c oss- ai and c oss-age co ela ions. Addi i e
gene ic co ela ions we e mos ly signi ican ly posi i e and
explained a pa o hese ai co ela ions. Howe e , mos
o hem we e lowe han sha ed en i onmen al co ela ions.
One o he addi i e gene ic co ela ions was nega i e leading
o he nega i e p opo ion o explained a ia ion, bu i was
no s a is ically signi ican . Gene ally, unique en i onmen al
c oss-age co ela ions we e small, and a ound hal o hem
we e no s a is ically signi ican . Fo c oss- ai co ela ions,
Table 1 Means and s anda d
de ia ions (SD) o amily
in e ac ion by age, sex and
zygosi y
a The scales a e ans o med o ange be ween 0 and 100
Boys Gi ls
All MZ wins DZ wins All MZ wins DZ wins
Mean SD Mean SD Mean SD Mean SD Mean SD Mean SD
Rela ional suppo a
12yea s o age 84 14 85 13 84 14 86 14 87 14 85 14
14yea s o age 80 15 80 15 79 15 79 18 80 17 78 18
17yea s o age 76 17 77 17 76 17 75 19 77 19 74 19
Pa en al epo a age 12 82 13 82 12 82 13 83 13 84 12 82 13
Rela ional ensionsa
12yea s o age 81 18 82 17 81 18 84 16 85 16 83 16
14yea s o age 79 17 81 17 79 17 80 16 82 16 79 16
17yea s o age 82 16 83 15 81 16 80 16 81 15 79 16
Pa en al epo a age 12 89 12 89 11 89 12 89 12 90 12 89 12
371Beha io Gene ics (2019) 49:366–375
1 3
unique en i onmen al co ela ions we e la ge han hose
o c oss-age co ela ions, bu hey we e s ill smalle han
sha ed en i onmen al and gene ic c oss- ai co ela ions.
Discussion
In his s udy based on a longi udinal Finnish win da ase ,
we ound ha gene ic ac o s explained an inc easing sha e
o he a ia ion o pe cei ed amily in e ac ion om 12 o
17yea s o age measu ed as ela ional suppo and ela ional
ensions. The esul s on he in luence o gene ic ac o s on
amily in e ac ion a e no su p ising, and u he mo e, a
gene ic componen has been ound no only o amily en i-
onmen (Klah and Bu 2014) bu also o li e e en s and
social suppo (Kendle and Bake 2007), which ha e been
adi ionally ega ded as pa o one’s en i onmen . We
ound mode a e gene ic co ela ions be ween he a ings o
amily in e ac ion om 12 o 17yea s o age. The US s udy
based on he Minneso a Twin Coho also ound inc eas-
ing gene ic a ia ion and gene ic co ela ions om 11 o
17yea s o age when hey analyzed pa en –child ela ion-
ships (Ludeke e al. 2013; McGue e al. 2005). In his US
coho , bo h he i abili y es ima es and gene ic co ela ions
we e, howe e , highe han wha we ound o pa en al en-
sions, which was due o a lesse ole o sha ed en i onmen al
ac o s. The UK win s udy using he measu es o household
chaos, pa en al discipline and pa en al eelings ound ha
gene ic ac o s explained a smalle and sha ed en i onmen-
al ac o s explained a la ge p opo ion o a ia ion han
in ou s udy (Hannigan e al. 2017a). This UK s udy di -
e ed om bo h ou esul s and he esul s o he US s udy
because no sys ema ic inc ease in he ole o gene ic ac o s
was obse ed and he gene ic co ela ions we e gene ally
o small magni ude. Ou esul s also di e om he esul s
o p e ious me a-analysis o mainly c oss-sec ional s udies
inding ha he e ec o gene ic ac o s emained oughly
s able du ing adolescence (Klah and Bu 2014). Howe e ,
i is no ewo hy ha di e en measu es o amily en i on-
men we e used in hese h ee longi udinal coho s and se -
e al di e en measu es o amily en i onmen we e used in
he me a-analysis o c oss-sec ional s udies. I is unce ain
whe he hese di e ences a e because o he di e ences in
he measu es o amily en i onmen , con ex ual di e ences
o whe he hey e lec o he di e ences be ween he coho s.
In addi ion o gene ic ac o s, sha ed en i onmen al ac-
o s also explained an impo an sha e o he a ia ion in
amily in e ac ion, especially a 12yea s o age. This sug-
ges s ha especially among child en, he a ings o amily
in e ac ion e lec he cha ac e is ics o pa en ing and less
so he child en’s own pe cep ions o pa en ing cha ac e -
is ics. This conclusion is u he suppo ed by ou esul s
ha sha ed en i onmen al co ela ions we e subs an ial and
explained in gene al a la ge sha e o c oss-age co ela ions
han gene ic ac o s. Ou esul s on he impo ance o he
sha ed en i onmen in he s abili y o e ages somewha
Table 2 Rela i e p opo ions
o addi i e gene ic, sha ed
en i onmen al and unique
en i onmen al ac o s wi h
95% con idence in e als (CI)
explaining he a ia ion in
amily in e ac ion om 12 o
17yea s o age
a Fac o sco es based on wo 1- ac o models
Addi i e gene ic ac o s Sha ed en i onmen al
ac o s
Unique en i onmen al
ac o s
a295% CI c295% CI e295% CI
Rela ional suppo a
Boys
12yea s o age 0.30 0.14, 0.47 0.34 0.19, 0.47 0.36 0.31, 0.42
14yea s o age 0.21 0.02, 0.41 0.36 0.19, 0.51 0.43 0.37, 0.50
17yea s o age 0.53 0.31, 0.65 0.06 0.00, 0.25 0.41 0.35, 0.48
Gi ls
12yea s o age 0.18 0.02, 0.34 0.46 0.31, 0.59 0.36 0.32, 0.42
14yea s o age 0.29 0.13, 0.47 0.38 0.21, 0.51 0.33 0.29, 0.39
17yea s o age 0.49 0.29, 0.68 0.16 0.00, 0.33 0.35 0.30, 0.41
Rela ional ensionsa
Boys
12yea s o age 0.13 0.01, 0.32 0.39 0.23, 0.50 0.48 0.41, 0.55
14yea s o age 0.20 0.01, 0.43 0.27 0.08, 0.44 0.52 0.45, 0.61
17yea s o age 0.35 0.20, 0.48 0.05 0.00, 0.18 0.60 0.52, 0.68
Gi ls
12yea s o age 0.14 0.01, 0.28 0.53 0.40, 0.64 0.33 0.29, 0.38
14yea s o age 0.26 0.07, 0.46 0.33 0.15, 0.49 0.41 0.35, 0.47
17yea s o age 0.33 0.05, 0.53 0.23 0.06, 0.46 0.43 0.37, 0.50
372 Beha io Gene ics (2019) 49:366–375
1 3
Table 3 T ai co ela ions o amily in e ac ion om 12 o 17yea s o age and co ela ions be ween addi i e gene ic, sha ed en i onmen al and unique en i onmen al a iance componen s
explaining hese ai co ela ions wi h 95% con idence in e als (CI)
T ai co ela ion Addi i e gene ic co ela ion Sha ed en i onmen al co ela ion Unique en i onmen al co ela ion
T ai 1 T ai 2 95% CI A95% CI % o he ai co e-
la ion explained
C95% CI % o he ai co e-
la ion explained
E95% CI % o he ai
co ela ion
explained
C oss-age co ela ions
Boys
Suppo 12 Suppo 14 0.42 0.38, 0.45 0.14 − 0.61, 0.53 9 0.96 0.71, 1.00 77 0.15 0.05, 0.26 14
Suppo 12 Suppo 17 0.25 0.21, 0.29 0.06 − 0.25, 0.30 9 1.00 0.64, 1.00 85 0.04 − 0.06, 0.15 6
Suppo 14 Suppo 17 0.41 0.37, 0.44 0.68 0.38, 1.00 56 0.82 0.23, 1.00 41 0.03 − 0.09, 0.14 3
Tensions12 Tensions14 0.33 0.29, 0.36 0.64 0.22, 1.00 40 0.63 0.36, 1.00 57 0.02 − 0.07, 0.12 3
Tensions12 Tensions17 0.17 0.13, 0.22 − 0.30 − 1.00, 0.07 − 24 0.52 0.30, 1.00 81 0.14 0.05, 0.22 43
Tensions14 Tensions17 0.32 0.28, 0.36 0.30 − 1.00, 1.00 16 0.82 0.49, 1.00 58 0.15 0.04, 0.25 26
Gi ls
Suppo 12 Suppo 14 0.39 0.35, 0.43 0.43 0.00, 0.91 29 0.67 0.46, 0.92 66 0.06 − 0.05, 0.16 5
Suppo 12 Suppo 17 0.28 0.24, 0.32 0.09 − 0.45, 0.94 9 0.78 0.38, 1.00 78 0.10 − 0.01, 0.21 13
Suppo 14 Suppo 17 0.42 0.38, 0.45 0.34 0.09, 0.59 34 0.97 0.52, 1.00 52 0.17 0.06, 0.26 14
Tensions12 Tensions14 0.31 0.27, 0.34 0.43 0.00, 1.00 28 0.54 0.26, 0.79 74 − 0.02 − 0.13, 0.09 − 2
Tensions12 Tensions17 0.19 0.15, 0.23 0.41 0.14, 0.92 38 0.35 0.17, 0.65 75 − 0.06 − 0.16, 0.03 − 13
Tensions14 Tensions17 0.32 0.28, 0.36 0.14 − 0.77, 0.44 13 0.71 0.45, 1.00 64 0.17 0.07, 0.28 23
C oss- ai co ela ions
Boys
Suppo 12 Tensions12 0.41 0.38, 0.45 0.64 0.29, 0.96 37 0.62 0.46, 0.82 52 0.12 0.02, 0.21 11
Suppo 14 Tensions14 0.46 0.42, 0.49 0.59 0.10, 1.00 28 0.78 0.53, 1.00 51 0.20 0.10, 0.30 21
Suppo 17 Tensions17 0.54 0.50, 0.57 1.00 0.71, 1.00 62 0.38 − 0.42, 0.96 8 0.31 0.22, 0.40 30
Gi ls
Suppo 12 Tensions12 0.44 0.41, 0.47 0.98 0.70, 1.00 45 0.51 0.38, 0.62 53 0.02 − 0.07, 0.11 2
Suppo 14 Tensions14 0.61 0.58, 0.63 0.79 0.49, 1.00 35 0.76 0.59, 0.97 44 0.35 0.26, 0.44 21
Suppo 17 Tensions17 0.65 0.63, 0.68 0.90 0.68, 1.00 51 0.90 0.56, 1.00 27 0.36 0.28, 0.45 22
373Beha io Gene ics (2019) 49:366–375
1 3
di e om a p e ious e iew on gene ic s udies o beha io-
al p oblems inding ha he s abili y is mainly in luenced
by gene ic ac o s (Hannigan e al. 2017b). Na u ally, amily
in e ac ion is no independen o sha ed genes, bu is in lu-
enced by pa en ing which also has a gene ic backg ound
(Mile a-Sei z e al. 2016). We ound modes co ela ions
be ween pa en al and o sp ing a ings o amily in e ac ion,
which also includes he passi e gene–en i onmen co ela-
ion. Sha ed en i onmen al e ec s become smalle om 12
o 17yea s o age when gene ic ac o s become mo e impo -
an . This esul is simila o ha ound o many psychologi-
cal ai s, such as in elligence and men al heal h indica o s
(Be gen e al. 2007), and e en o BMI (Sil en oinen e al.
2016). Since adolescence is cha ac e ized by dec easing
dependence on pa en s and inc easing in luence o pee s
(Ahmed e al. 2015; Kil o d e al. 2016), we could specula e
ha his inc easing gene ic e ec e lec s he ole o an ac i e
gene–en i onmen co ela ion when child en ac i ely shape
hei en i onmen pa ly based on hei geno ype. Inc eas-
ing independence may also lead child en o in e p e amily
in e ac ions mo e h ough hei own cha ac e is ics, such as
pe sonali y, which a e a ec ed by gene ic ac o s (Spengle
e al. 2012).
Unique en i onmen al ac o s also explained a sha e o
a ia ion o amily in e ac ion, and a some ages, hey we e
mo e impo an han gene ic o sha ed en i onmen al ac-
o s. Howe e , in con as o gene ic and sha ed en i on-
men al ac o s, he unique en i onmen al co ela ions we e
small in size especially when conside ing he measu emen s
be ween ages. Mode a e unique en i onmen al co ela ions
we e, howe e , ound be ween he wo dimensions o amily
in e ac ion measu ed a he same ages. This sugges s ha
unique expe iences ela ed o amily in e ac ion a e ime-
speci ic. This small e ec o unique en i onmen al ac-
o s on he s abili y o amily in e ac ion cha ac e is ics is
suppo ed by a p e ious win s udy which ound e y low
unique en i onmen al co ela ions o pa en –child in e ac-
ion, e en when measu ed on consecu i e days (Bu e al.
2015). This may sugges ha pa en s do no sys ema ically
a o o disc imina e agains he same child o e ime, and
hus unique en i onmen al ac o s e lec ansien e ec s o
p obably e y mino en i onmen al in luences, changes in
mood o possible co ela ed measu emen e o s since ela-
ional suppo and ela ional ensions we e measu ed using
he same ques ionnai e.
In addi ion o he gene ic a chi ec u e o he amily in e -
ac ion, ou da a allowed us o s udy how he pe cei ed am-
ily in e ac ion changes om ea ly o la e adolescence as well
as di e be ween pa en s and child en. P e ious s udies ha e
sugges ed ha pa en –child ela ionships de e io a e om
childhood o adolescence (Hadiwijaya e al. 2017; Kim e al.
2001; Loebe e al. 2000). Ou esul s only pa ly suppo ed
his because we ound de e io a ion in ela ional suppo
om 12 o 17yea s o age, bu when s udying ela ional
ensions, he only di e ence was less ela ional ensions
a 12yea s o age in gi ls. Howe e , he age pa e n may
depend on how pa en ing and amily in e ac ion ha e been
assessed. In his s udy, he amily-in e ac ion ques ionnai e
was used (Na usk and Pulkkinen 1994), and based on his
ques ionnai e, he wo dimensions we e calcula ed such as
also in he p e ious s udies using his same measu emen
ins umen (La end esse e al. 2010; Sil en oinen e al.
2014). The decision o use wo dimensions was ini ially
based on he esul s o he explo a i e ac o analyses sys-
ema ically sugges ing he wo- ac o solu ion a all ages, in
boys and gi ls as well as in pa en s and hei o sp ing bu
was u he suppo ed by ou empi ical esul s. The dec eas-
ing sco es o ela ional suppo may be ela ed o he inc eas-
ing ac i i ies o child en ou side home when some child en
hink ha hey do no anymo e ge , o e en need, suppo
om hei pa en s. This may also explain why he a ia-
ion o ela ional suppo inc eases a he same ime. On he
o he hand, he inc easing independence does no inc ease
ensions in he amily and seems ac ually o le el o di -
e ences be ween child en as indica ed by he dec easing
a ia ion o ela ional ensions. The p esence o wo dimen-
sions is also suppo ed by compa isons be ween pa en al
and o sp ing a ings when he o sp ing we e a he age
o 12. Pa en s a ed less ela ional ensions bu lowe ela-
ional suppo han did o sp ing. This may e lec pa en s’
sel -c i ical a i udes owa d amily in e ac ion while ha ing
mo e posi i e a i udes owa d hei child en, a he han ice
e sa. Thus, ou esul s sugges ha ela ional-suppo and
ela ional- ensions a e co ela ed bu s ill pa ly independen
dimensions o amily-in e ac ion.
Ou s udy has bo h s eng hs and weaknesses. The main
s eng h is he longi udinal measu es o amily in e ac ion
using he same ques ions a h ee ages co e ing he c i ical
phase o li e om ea ly o la e adolescence in a la ge se
o wins, allowing us o s udy he ole o gene ic and en i-
onmen al ac o s o e ime. Using he Finnish popula ion
egis e , we we e able o iden i y all wins in he selec ed
bi h coho s. Though he esponse a es in ou win su eys
we e e y high (88–95%), he epea ed su eys dec eased
e en ion o e ime, such ha by 17yea s o age 64% o all
pai s we e s ill in he s udy. None heless, ou s udy coho
can be ega ded as ep esen a i e. Fu he , he pa en s o
wins also independen ly a ed amily in e ac ion using he
same ques ions as hei o sp ing allowing us o s udy he
associa ions be ween he a ings o pa en s and o sp ing.
The main limi a ion o ou da a is ha amily in e ac ion
was assessed wi h a limi ed numbe o ques ions. A la ge
numbe and g ea e b ead h o ques ions would ha e p ob-
ably educed he measu emen e o .
In conclusion, gene ic ac o s play an impo an ole in
pe cei ed amily in e ac ion in la e adolescence, whe eas