Content uploaded by Alexandra Zapko-Willmes
Author content
All content in this area was uploaded by Alexandra Zapko-Willmes on Feb 08, 2018
Content may be subject to copyright.
QUASI-CAUSAL EFFECTS OF DIFFERENTIAL PARENTING ON RWA 1
Unraveling Quasi-Causal Environmental Effects via Phenotypic and Genetically
Informed Multi-Rater Models: The Case of Differential Parenting and
Authoritarianism
Alexandra Zapko-Willmes1,2, Rainer Riemann², and Christian Kandler1
1MSB Medical School Berlin, 2Bielefeld University
Uncorrected preprint version of a paper accepted for publication in
European Journal of Personality
February 6th, 2018
The authors received support from the Deutsche Forschungsgemeinschaft (DFG) KA
4088/2-1. The content of the paper had been presented as invited talk of the symposium
“Patterns and Sources of Personality Development Across the Life Span” at the biennial
Meeting of the German Psychological Association (DGPs) section Differential Psychology,
Personality Psychology and Diagnostics (DPPD) 2017 in Munich, Germany.
Correspondence concerning this article should be addressed to Alexandra Zapko-
Willmes, MSB Medical School Berlin, Calandrellistr. 1–9, 12247, Berlin, Germany.
E-mail: alexandra.zapko@medicalschool-berlin.de
QUASI-CAUSAL EFFECTS OF DIFFERENTIAL PARENTING ON RWA 2
Abstract
This study investigated the association between different experiences of parenting and
individual right-wing authoritarianism (RWA) using twin family data comprising self and
informant reports. We applied a design that allowed us to examine whether the link between
retrospective assessments of parenting and current RWA is effectively environmental or
whether the association is attributable to genetic influences. We hypothesized that an
authoritarian parenting style (low responsiveness and high demandingness) provided by the
parents is associated with higher offspring’s RWA, and that this association is similar for
both twin siblings as a function of their genetic relatedness and shared familial experiences –
that is, genotype-environment correlation. A sample of 875 twins as well as 319 mothers and
268 fathers completed a questionnaire on twins’ parental environment and their own
authoritarian attitudes. Additionally, 1,322 well-informed peers assessed twins’ RWA.
Applying structural equation modeling, we found twins’ experiences of parental
responsiveness and demandingness to be positively associated with self- and peer-reported
RWA. The correlation between responsiveness and RWA was similar for both twins due to
their genetic similarity, whereas twin differences in demandingness were positively
associated with twin differences in RWA, indicating quasi-causal environmental effects.
Implications for the interdependence between parenting and RWA are discussed.
Keywords: right-wing authoritarianism, parenting, multi-rater twin family study,
quasi-causality, genotype-environment correlation
QUASI-CAUSAL EFFECTS OF DIFFERENTIAL PARENTING ON RWA 3
Unraveling Quasi-Causal Environmental Effects via Phenotypic and Genetically Informed
Multi-Rater Models: The Case of Differential Parenting and Authoritarianism
Investigating the influence of family environments on complex human traits, such as
the impact of parenting on right-wing authoritarianism, is a difficult endeavor in light of the
highly replicable result that all behavioral traits are heritable (Turkheimer, 2000). Since the
genetic make-up (i.e., the genotype) and the environment are highly interwoven with each
other (Kandler & Zapko-Willmes, 2017), findings that link certain environmental factors to
an observable trait (i.e., the phenotype) may be confounded by genetic factors. For example,
individual experiences can be driven through heritable traits that affect the selection,
avoidance, or creation of certain environments or evoke specific responses from the social
environment. Moreover, the parental genetic make-up can shape the offspring’s environment,
fostering heritable behaviors by providing a matching and stimulating environment (Scarr &
McCartney, 1983). Furthermore, systematic measurement errors in the form of heritable
response biases, for example acquiescence and social desirability (Kandler, Riemann,
Spinath, & Angleitner, 2010), or response tendencies associated with the investigated trait
itself, for example a favorable assessment of authority figures’ behaviors by highly
authoritarian people (Frenkel-Brunswik, 1950), may confound the results. This is precarious
when the information on both environment and outcome variable are provided by the same
rater.
Genetically informative studies that consider several rater perspectives, in other words
genetically informed multi-rater studies, are useful to control for genetic influences and can
test whether the link between an environmental variable and a complex human trait is
effectively environmental. In this paper, we focus on differential parenting, defined as
differences in the experience of parental treatment between siblings. Using a twin family
design that incorporates several rater perspectives, we investigated whether differences in
QUASI-CAUSAL EFFECTS OF DIFFERENTIAL PARENTING ON RWA 4
retrospectively assessed parenting can act as an environmental factor affecting twin sibling
differences in right-wing authoritarianism (RWA), or whether the association between
parenting and RWA is rater-specific and (or) confounded by genetic influences, suggesting
alternative explanations.
Sources of Individual Differences in RWA
Based on the work concerning the “authoritarian personality” by Adorno, Frenkel-
Brunswik, Levinson, and Sanford (1950), Altemeyer (1988, 1996) conceptualized and
defined RWA as a core motivational value orientation. RWA comprises a strong tendency to
adhere to perceived societally established and legitimate authorities and their advocated
social conventions and norms. This includes negative attitudes and a pronounced level of
aggressiveness towards persons deviating from these directives. Both Adorno et al. and
Altemeyer argued that the development of authoritarianism is substantially influenced by the
rearing environment. Altemeyer (1988) reported a significant positive correlation between
parental RWA scores and those of their biological children (r = .40) as well as their adopted
children (r = .55), indicating an environmental transmission from parents to offspring
independent of their genetic relatedness.
Other studies provided a different picture. Twin studies, for example, have shown that
the environment shared between twin siblings is of little importance for individual differences
in RWA, with genetic and environmental factors not shared between twin siblings being more
important (e.g., Bouchard & McGue, 2003; Funk et al., 2013; Lewis & Bates, 2013; Scarr &
Weinberg, 1981). In a study on RWA and rearing environment, McCourt, Bouchard, Lykken,
Tellegan, and Keyes (1999) applied a four-group twin design comprising monozygotic and
dizygotic twins reared together and apart. They found that twins’ RWA scores were
significantly correlated with family moral-religious emphasis, organization, and control for
individuals that were reared by biological relatives but not for adoptees. They concluded that
QUASI-CAUSAL EFFECTS OF DIFFERENTIAL PARENTING ON RWA 5
the association between family environment and RWA is genetically mediated and, thus,
variance in rearing environments does not affect individual differences in RWA beyond
genetic contributions. McCourt et al. conceptualized parenting as an environmental factor
objectively shared between twin siblings reared together. But do twin siblings actually
receive the same parental treatment and do they perceive it equally?
Differential Parenting and Offspring’s Individuality: Genetically Linked?
Most studies have implemented parenting as an environmental factor shared between
siblings. In their review, Collins, Maccoby, Steinberg, Hetherington, and Bornstein (2000)
criticized this and noted that the offspring’s individuality and the inter-connectedness of
genetic and environmental factors are often not taken into account. Siblings might experience
the same family environment more or less alike and subsequently show more or less
congruent behaviors, based on their genetic relatedness. In addition, parents might treat their
offspring in accordance with their own genetic make-up and the offspring’s idiosyncratic
behavior. Parenting may thus act as a function of senders’ and recipients’ individuality, as
Scarr (1985) put it:
[…] what people experience cannot be indexed by observations of environments to
which they are exposed. What people experience in any given environment depends
on what they attend to, how much they learn, how much reinforcement they feel they
get for what behaviors. And what they experience in any given environment is a
function of genetic individuality and developmental status. (p. 510)
A link between the offspring’s parental treatment (i.e., their family environment) and
the offspring’s trait might thus be affected by the offspring’s genotype. The nonrandom link
between genetic and environmental factors has been termed as genotype-environment
correlation (see Figure 1A; Scarr & McCartney, 1983). Parents usually provide both the
genetic make-up and the rearing environment of their offspring. Since the rearing
QUASI-CAUSAL EFFECTS OF DIFFERENTIAL PARENTING ON RWA 6
environment is influenced by the parental genetic make-up (Klahr & Burt, 2014), the
offspring’s genetic make-up and their parental treatment might be correlated, resulting in
passive genotype-environment correlation. In line with this, Kandler, Bell, and Riemann
(2016) reported a significant contribution of passive genotype-environment correlation (16%)
to individual differences in RWA and discussed the potentially mediating role of
authoritarian parenting on generational transmission of RWA (see below). Moreover, siblings
might evoke different parental behaviors based on their heritable dispositions to individual
behaviors – for example compliant or deviant behaviors. In other words, the offspring’s
genotypes might evoke similar or differential parenting as a function of their genetic
relatedness. This nonrandom link between genetic and environmental factors has been termed
as evocative genotype-environment correlation (Klahr and Burt, 2014).
Individuals may thus be exposed to, evoke, or experience parenting as a function of
their genetic makeup. As a consequence, in order to investigate the effects of individual
differences in the parental treatment on individual differences in RWA, it is worthwhile to
conceptualize parenting as an environmental factor that may be shared to some degree
between siblings, but may also act individually.
Authoritarian and Authoritative Parenting and the Offspring’s RWA
Parenting has been most prominently defined by the level of responsiveness – that is,
emotional support, acceptance and warmth offered to the child – and the level of
demandingness – the provided structure, control and restrictiveness. Typically, four parenting
styles are discerned (Baumrind, 1971; Maccoby & Martin, 1983), of which the authoritarian
and authoritative parenting style have been linked to RWA. Both parenting styles feature high
demandingness, although Baumrind (2013) declared authoritarian parenting to be comparably
more restrictive. More importantly, both styles differ in their level of responsiveness, with
QUASI-CAUSAL EFFECTS OF DIFFERENTIAL PARENTING ON RWA 7
authoritative parenting involving high responsiveness and authoritarian parenting low
responsiveness.
In a study on the influence of the parenting type on the offspring’s competence,
conformity, and problem behavior in 124 families, Baumrind (1991) found children of
authoritative parents to be the most competent, socially adapted and individuated. In contrast,
children who were raised by authoritarian parents showed less prosocial and socially
responsible behavior, were less autonomous and self-regulated, tended to conform easily and
showed more aggressive behavior outside the home. These results have been replicated (e.g.,
Weiss & Schwarz, 1996) and indicate that parents high on RWA provide an authoritarian
parenting style (e.g. Manuel, 2006; Peterson, Smirles, & Wentworth, 1997) that is associated
with higher levels of RWA in their offspring.
Duriez, Soenens, and Vansteenkiste (2007) investigated cross-sectional associations
between parenting and RWA as well as prospective effects of parental styles and goals on
adolescent RWA. They found that both parental support (i.e., responsiveness) and parental
regulation (i.e., demandingness) were positively associated with RWA in two cross-sectional
studies, indicating that authoritative rather than authoritarian parenting is related to RWA,
while their longitudinal data yielded different results: Not parenting styles, but parenting
goals – that is, the parental promotion of extrinsic vs. intrinsic as well as conservation goals
through their parenting – positively predicted RWA. When interpreting their findings, they
pointed to the important role of rater perspectives: When parenting is captured by offspring’s
reports, authoritative parenting can be expected to be positively associated with the raters’
RWA. Offspring high on RWA would probably not cast their parents in a negative light,
since parents represent authorities, which are positively evaluated by individuals high on
RWA. Consequently, while the assessment of high demandingness would probably be
unbiased, as it is not necessarily perceived as detrimental, high responsiveness is socially
QUASI-CAUSAL EFFECTS OF DIFFERENTIAL PARENTING ON RWA 8
desirable. The direction of causation underlying the association between parenting and RWA
would then run into the opposite direction: Individual differences in RWA would affect
individual differences in the perception or assessment of parenting.
Measuring Differential Parenting in Light of Response Biases
When investigating the association between parenting and RWA, it is worthwhile to
consider several rater perspectives. First, as stated above, offspring’s RWA might bias their
parenting rating, consequently inflating (or deflating) the association. One strategy to control
for such specific rater biases might be the use of parenting assessments based on parents’
reports. However, parents may also be inclined to respond socially desirable, especially with
respect to their parenting (e.g., Morsbach & Prinz, 2006). In their meta-analysis, Avinun and
Knafo (2014) reported different heritability estimates and different estimations of
environmental influences when comparing observational, offspring’s, and parental reports.
They also noted that parents are less likely to report differential parenting than their
offspring. As previous studies found that parents’ and offspring’s perception of the same
family environment showed low intercorrelation (Kraemer et al., 2003; Riemann, Kandler, &
Bleidorn, 2012), it may well be that parents and offspring may provide different,
complementary perspectives on the same issue, such as parenting.
Apart from such specific response biases, global response biases such as
acquiescence, severity, and leniency, might confound the association when assessments are
provided by a single rater. To estimate the impact of specific response biases – in other
words, the influence of the rater’s RWA on their parenting ratings – independently of global
response biases that pertain to all assessments, offspring’s self-rated RWA should be
complemented with other measures, such as informant reports. When supplemented by both
parents’ and offspring’s reports on parenting, this would allow to control for both specific
and global responses biases and get a better and more accurate insight into the underlying
QUASI-CAUSAL EFFECTS OF DIFFERENTIAL PARENTING ON RWA 9
sources of the association between differential parenting and authoritarianism (see Figure
1B). Since rater biases might be heritable to some degree (Kandler et al., 2010), genetic
influences on the association between differential parenting and offspring’s RWA might not –
at least not exclusively – reflect a genotype-environment correlation, but genetic differences
in rater tendencies.
Vertical and Horizonal Effects of the Parental RWA
Since the parental RWA was found to be linked with both the parenting style and the
offspring’s RWA (e.g., Altemeyer, 1988; Peterson, Smirles, & Wentworth, 1997), it is
worthwhile to consider the parental RWA as a factor contributing to the association (see
Figure 1C). Apart from genetic transmission, the parental RWA might directly affect the
offspring’s RWA through parental right-wing authoritarian behaviors that can be observed,
imitated and adopted, which might be in turn rewarded (genotype-environment correlation).
To reflect this direct intergenerational transmission (independent of concrete genetic and
environmental contributions), we termed it the vertical effect of the parental RWA on the
offspring’s RWA. Parental authoritarianism could also indirectly affect the offspring’s RWA
and associated behaviors through more or less warm and demanding parenting. To
differentiate this indirect intraindividual effect of the parental RWA on the parenting style
from the direct vertical effect, we termed it the horizontal effect of parental authoritarianism
on their parenting style.
As a consequence, the parenting style might function as a mediator between the
parent’s and offspring’s RWA. Since right-wing authoritarianism is moderately heritable,
such a mediation might be genetically driven and reflect passive genotype-environment
correlation. When genotypes and shared family environments are correlated, twins’ similarity
increases irrespective of their genetic relatedness as a function of shared environmental
factors (Briley, Livengood, & Derringer, 2018).
QUASI-CAUSAL EFFECTS OF DIFFERENTIAL PARENTING ON RWA 10
By taking the parental RWA into account, it is possible to further differentiate
whether findings pointing to a genetic effect indicate a direct genetic transmission, a passive
or even an evocative genotype-environment correlation.
Modeling Framework in a Nutshell: Assumptions and Parameters
Measurement of both the environmental variable and the trait can be decomposed into
the true score (i.e., the latent environmental variable or latent trait), systematic measurement
error and unsystematic measurement error. Since RWA is heritable (a) and since individuals
may be exposed to, evoke or seek environments in accordance with their genetic make-up,
the individual’s genotype and environment may be correlated (Figure 1A; ba). The
association between experienced parental treatment and RWA can be confounded by a shared
genetic basis in the form of a passive or evocative genotype-environment correlation, that is,
due to the offspring being raised in a parenting environment in accordance with their
genotype (a passive genotype-environment correlation), or the parents reacting to the
offspring’s genetically driven behavior (an evocative genotype-environment correlation).
Further, partially heritable response biases both pertaining to a measurement method, such as
self-reports (in the following global response biases; m), and/or due to the specific trait of
interest, namely RWA (in the following specific response biases; atp), can inflate the
systematic error variance and subsequently confound the association between parenting and
RWA (Figure 1B).
If the family environment in question is provided by genetically related individuals,
such as the parents, the offspring’s trait can be linked to the parents’ traits as a consequence
of genetic transmission (in the following vertical effects; v) and/or due to the parents
providing the rearing environment in accordance with their trait level (in the following
horizontal effects; h; Figure 1C).
QUASI-CAUSAL EFFECTS OF DIFFERENTIAL PARENTING ON RWA 11
With the help of twin family data (including informant reports) and structural equation
modeling, we could explore whether the association between twins’ parental treatment and
their RWA is genetically confounded or effectively environmental, and to what extent
response biases and the parental phenotypic trait contribute to the link.
The Present Multi-Rater Twin Family Study
The current twin family study investigated the association between individual
differences in retrospectively reported parenting – assessed by parents and offspring – and
individual differences in offspring’s current RWA – captured by self-reports and ratings of
well-informed observers. In line with Baumrind’s (2013) suggestions, from a strictly
phenotypic point of view, we hypothesized that the offspring’s RWA would be negatively
predicted by parental responsiveness (Hypothesis 1.1) and positively predicted by parental
demandingness (Hypothesis 1.2). Yet, in accordance with previous findings (McCourt et al.,
1999), we expected a genetic contribution to the link between parenting and RWA
(Hypothesis 2) for both responsiveness and demandingness, and this genetic mediation
should reflect evocative genotype-environment correlation (Hypothesis 3).
To examine whether phenotypic associations reflect an at least partially environmental
effect of parenting on RWA as opposed to a complete genetic confounding, we first tested the
hypotheses by applying bivariate structural equation modeling on twin data, termed
genetically informative regression models (Turkheimer & Harden, 2014). To better fathom
and identify potentially underlying structures and mechanisms contributing to the association,
we further ran analyses based on structural equation modeling that did not consider the
genetic relatedness of the twins, which we termed phenotypic semilatent multitrait-
multimethod (MTMM) analyses. These phenotypic semilatent MTMM analyses allowed us to
isolate common and specific variance (1) in offspring’s, mother’s and father’s accounts on
parenting and (2) in self- and peer reports on offspring’s RWA, and to consider (3) self-
QUASI-CAUSAL EFFECTS OF DIFFERENTIAL PARENTING ON RWA 12
reports on maternal and paternal RWA. This analytic strategy allowed us to identify rater-
specific covariance due to influences of response tendencies and influences of parental RWA
on their parenting and offspring’s RWA. Such response tendencies comprise global response
biases, such as acquiescence, leniency, severity, and social desirability pertaining to both the
assessment of parenting and the self-reports on RWA, and specific response biases due to the
rater’s characteristic of interest, in this case biases due to the rater’s RWA. Parental RWA
may have a direct vertical effect on offspring’s RWA due to vertical genetic transmission, as
well as an indirect horizontal effect on offspring’s RWA via their parenting style.
The model estimates were quasi-cross-validated
1
across both twins of a twin pair to
confirm the predictive value of the tested models. This also helped to reduce potential type 1
error inflation, for which we did not control in favor of stabilization of type 2 error. As
retrospective assessments do not allow for causal inferences and retrospective reports on
parenting might be biased by the current RWA as well as other traits and experiences (e.g.,
being a parent oneself), the consideration of rater-specific effects in the MTMM analyses
strengthen the credibility of the findings.
If a significant phenotypic association between parenting and RWA beyond the
aforementioned confounding factors can be found and the association is not solely driven by
genetic factors shared by both twins, the association between parenting and RWA would be
at least partly environmental, supporting the notion that individual differences in RWA are
indeed affected by individual differences in experienced parental treatment – or other
confounding environmental factors. For example, if the association is confounded by
environmental factors shared by the twins, this could reflect parenting influences on RWA
shared by twins or that the association could be due to a third environmental factor
1
This is no direct and full cross-validation, because twin siblings of a pair are not independent and share the
same parents.
QUASI-CAUSAL EFFECTS OF DIFFERENTIAL PARENTING ON RWA 13
influencing both parenting behaviors towards the twins and twins’ RWA in the same vein, for
example the socioeconomic status (Carvacho et al., 2013; Heydari, Teymoori, & Haghish,
2013; Hoff, Laursen, & Tardif, 2002) or parenting goals (Duriez et al., 2007).
A found genetic or shared environmental contribution based on the genetically
informative regression analyses in combination with the results of the phenotypic semilatent
MTMM analyses would allow for four different explanations for the association between
parenting and RWA: (1) passive genotype-environment correlation, if the association is due
to shared environmental factors, valid across rater perspectives, and mainly mediated by
parents’ RWA; (2) evocative genotype-environment correlation, if the association is due to
genetic factors, valid across rater perspectives, but not driven by parents’ RWA; (3)
genetically driven perceptions or assessments of parenting due to raters’ RWA, if the
association is due to genetic factors, accounted for by twins’ RWA affecting twins’ but not
parents’ parenting rating; and finally (4) covariance due to shared rater biases, if the
association is due to genetic or shared environmental factors and explained by rater-specific
perspectives. The explanations are not mutually exclusive. That is, all explanations can
account for the link between assessments of parenting and RWA.
Methods
Participants
We analyzed data from 875 twins reared together and 587 parents of twins from the
Jena Twin Study of Social Attitudes (JeTSSA; Stößel, Kmpfe, & Riemann, 2006). The twins
included 226 monozygotic (MZ) sibling pairs, 168 dizygotic (DZ) pairs (101 same-sex and
67 opposite-sex pairs), and 87 unmatched twins. Twins ranged in age from 17 to 82 years (M
= 34.30; SD = 13.62), of whom 74% were female.
2
The sample’s educational and
2
Even though it is standard to control for age effects, we decided not to control for age differences in RWA and
parenting. Since parenting and authoritarianism, similar to conservatism, may change over time due to zeitgeist
and socio-political developments, age itself might reflect or be linked to factors that act to increase twins’
QUASI-CAUSAL EFFECTS OF DIFFERENTIAL PARENTING ON RWA 14
occupational background was heterogeneous. For a more detailed sample description see
Table 1 and Stößel et al. (2006).
In addition to self-raters, 1,322 well-informed acquaintances of twins provided peer
reports on twins’ RWA. For each twin sibling, different peers provided assessments with
preference given to those peer raters who knew one twin very well but not the co-twin. For
86% of twins at least one peer report and for 66% of twins two peer reports, which were
averaged for analyses, were available (see Table 1).
Measures
RWA. A German version of Altemeyer’s RWA scale (1988; 1996), the 12-item
RWA³D scale (see Funke, 2005) was implemented. Items were rated on a 5-point scale,
ranging from 1 (Strongly disagree) to 5 (Strongly agree). Item descriptions have been
reported elsewhere (Kandler et al., 2016). One item (“People ought to develop their own
moral standards of ‘Good and Bad’ and to put less attention to the Bible and other old
traditional beliefs.”) was omitted due to an insufficient item-total correlation in both the self-
and peer report data. For structural equation modeling, RWA scores were z-standardized. The
internal consistency ranged from (Cronbach’s) α = .675 (father’s report) to α = .743 (twin’s
report) for self-reports and was α = .746 for peer reports. Factor analyses exploring the
unidimensionality of RWA item assessments are provided in online supplement A.
Descriptive statistics are shown in Table 1 and correlations between family members’ self-
reports are displayed in Table 2. In contrast to twins’ self-reports, correlations between twin
siblings based on averaged peer reports were more similar across MZ and DZ twin pairs: r =
.493, n = 176, 95%CI[.372, .597], for MZ twins, and r = .445, n = 134, 95%CI[.298, .571],
similarity in RWA and authoritarian parenting (e.g., older people tend to be more authoritarian and to have
received a more authoritarian parental treatment). We did not want to partial out variance due to those potential
shared environmental effects and thus decided to leave possible age effects statistically uncontrolled.
QUASI-CAUSAL EFFECTS OF DIFFERENTIAL PARENTING ON RWA 15
for DZ twins. These differences could be due to response tendencies, for example
acquiescence or social desirability, and/or due to additional valid information inaccessible to
the peers, such as inner thoughts and feelings. If these error or trait variance components are
genetically influenced, correlations between self-reports of monozygotic twins will be higher
than DZ twin correlations (Kandler et al., 2010). For a thorough discussion of different rater
perspectives on ideological attitudes, see Cohrs, Kämpfe-Hargrave, and Riemann (2012).
Please note that Cohrs et al. relied on the same multi-rater twin data in their study 2.
BEQ. Twins and parents filled out the German version of the Block Environmental
Questionnaire (BEQ; Harmening, 2014; Hur & Bouchard, 1995; Riemann & Wagner, 2000)
– a retrospective measure of family environment. Twins rated maternal and paternal
treatment based on 81 items (concerning maternal and paternal acceptance/rejection, family
cohesion, maternal and paternal intellectual-cultural orientation, and family organization) and
parents rated their own parental behavior based on 54 items on a 5-point scale, ranging from
1 (Strongly disagree) to 5 (Strongly agree).
For the purpose of the current study, we selected BEQ items that – in accordance with
theoretical considerations (Baumrind, 1971, 2013; Maccoby and Martin, 1983) – referred to
specific parental behavior related to either responsiveness or demandingness, in other words
imply shown affection and supportive actions adapted to the child’s needs or monitoring and
restrictive actions aimed at providing structure and order in the child’s environment and
aligning the child’s behavior with social norms. In order to confirm this selection, we took
the following steps: (1) We ran internal consistency analyses for both subscales and excluded
four inconsistent items; (2) we calculated principal axis analyses using varimax rotation in
order to see whether the kept items load on two factors in accordance with the theorized two-
dimensional structure; and (3) finally, since the dimensions are supposed to be independent
(Baumrind, 2013), we computed orthogonal factor scores using the Anderson-Rubin method
QUASI-CAUSAL EFFECTS OF DIFFERENTIAL PARENTING ON RWA 16
(Anderson & Rubin, 1956). At the end, we kept 22 items in the twins’ version, 14 of which
capture responsiveness and 8 demandingness, and 16 items in the parents’ version, 10 items
reflecting responsiveness and 6 measuring demandingness. See online supplement A for more
details regarding factor analyses and psychometric quality. Descriptive statistics of averaged
item scores are provided in Table 1. Correlations between family members’ reported
responsiveness and demandingness are shown in Table 2. All family-dyad correlations were
positive and statistically significant (p < .001). All correlations between parents and twins as
well as between mothers and fathers were comparable across responsiveness and
demandingness. In addition, and in line with previous research (Kraemer et al., 2003), no
correlation was larger than r = .40, indicating substantial differences between parents and
offspring as well as between mothers and fathers in the perception or assessment of
parenting. This might also be due to the parents’ not rating the parenting of each other,
leading to a variance component not shared between offspring’s and parents’ ratings. This
lack of measurement invariance might have contributed to additional incongruence
concerning these correlation analyses. The comparatively low mother-father correlation in
responsiveness and demandingness may indicate substantial differences between mothers and
fathers in the treatment of their children.
Analyses
Structural equation models (SEM) were run based on the statistical software package
IBM SPSS Amos 21.0 (Arbuckle, 2012). All SEM-based estimates were derived via full
information maximum likelihood (FIML) procedures to analyze all available data and handle
missing values due to dropout (Little & Rubin, 2002). See https://osf.io/mecfb/ for Amos
QUASI-CAUSAL EFFECTS OF DIFFERENTIAL PARENTING ON RWA 17
scripts and outputs (the latter as Excel files).
3
Due to the sample size and related issues of
statistical power, we performed the genetically informative and MTMM analyses separately.
Genetically informative regression analyses. Based on data provided by twins
reared together, sources of individual differences in psychological traits can be split into an
additive genetic component (A) and two environmental components due to environmental
influences shared (C) and not shared by twins including measurement error (E). DZ twins
share on average 50% of their segregating alleles, whereas MZ twins share 100% of their
genetic make-up. Both MZ and DZ twins reared together share environmental factors that act
to increase twin pairs’ similarity. Assuming that these shared environmental influences affect
DZ twins’ resemblance to the same degree as they contribute to the similarity of MZ twins,
differences between MZ and DZ twin similarities can be attributed to genetic influences,
whereas within-pair differences are attributable to nonshared environmental influences.
Strong shared environmental influences that act to make twins more similar are indicated in
case of low within-pair differences and low differences between MZ and DZ twin
similarities.
The classic twin model (CTD) cannot take assortative mating into account, which
might lead to an overestimation of the effect of shared environmental factors and an
underestimation of the genetic contribution. Assortative mating of twins’ parents might
inflate DZ twin correlations as a result of the parental genetic similarity. It does not affect
MZ twin correlations, as their perfect genetic similarity cannot be increased. This inflation
consequently diminishes the difference between MZ twin pair correlations and DZ twin pair
correlations, culminating in skewed estimates. Since twin and parental estimates on RWA
3
Note, the JeTSSA dataset is not public domain. Therefore, the data can only be used to reproduce the results
presented in this study. Requests for data use for own research projects should be sent to the principal
investigator Rainer Riemann (rainer.riemann@uni-bielefeld.de).
QUASI-CAUSAL EFFECTS OF DIFFERENTIAL PARENTING ON RWA 18
were available, we could adjust the initial assumption of 50% of shared genes between DZ
and control for parents’ similarity and thereby assortative mating (see Kandler et al., 2016):
As both the heritability [h² = 2 × (rMZ – rDZ)] of RWA taking assortative mating into account
and spouse similarity (µ) in RWA was about .50 (see also Table 2), the genetic correlation
between DZ twins (γ2(DZ) = .50) in the CTD could be corrected as follows: 0.5 + 0.5 × h² × µ
= 0.5 + 0.5 × 0.5 × 0.5 = 0.625 (see Martin et al., 1986, or Stieger, Kandler, Tran, Pietschnig,
& Voracek, 2017, for more details on the correction for assortative mating). As a result, we
specified the genetic correlation of MZ twins at γ2(MZ) = 1 and the genetic correlation of DZ
twins at γ2(DZ)’ = .625 in our CTD of RWA.
4
The CTD also assumes the absence of genotype-environment interplay, although both
genotype-environment correlations and interactions might inflate estimates of A, C, and E
(see Bleidorn, Kandler, & Caspi, 2014, and Bleidorn, Hufer, Kandler, Hopwood, & Riemann,
2018). The applied models allowed us to investigate to what extent genotype-environment
correlations might play a role.
Bivariate ACE twin models allow for genetic and environmental links between two
variables (see Figure 2). In the current study, the bivariate twin model allows for estimations
of regressions from twins’ rated parenting on genetic variance in twins’ RWA (bA), which can
reflect genetically influenced response biases shared between both measures (global response
bias) or due to twins’ RWA (specific response bias), and/or evocative genotype-environment
4
We did not adjust assessments of parenting behaviors for assortative mating primarily due to conceptual
considerations. First, parenting is not a phenotype per se but a feature of twin’s family environment. Second,
fathers’ and mothers’ self-reports on their own parenting are not equivalent to twin reports on their experienced
parental treatment, because twins rated both parents’ behavior, while parents rated only their own behaviors. As
a consequence, it would be unclear, to what extent assortative mating, social homogamy, shared social
background, or spousal interactions (assimilation, accommodation) affect similarity in parents’ parenting.
Adjusting objectively (but not necessarily effectively) shared family environments for the shared component
would artificially result in variation due to effectively not shared environmental influences (i.e., only differences
between maternal and paternal treatment). Thus, adjustment for assortative mating is problematic in case of
parenting.
QUASI-CAUSAL EFFECTS OF DIFFERENTIAL PARENTING ON RWA 19
correlation that act as a function of siblings’ genetic relatedness (Briley et al., 2018). The
model also allows for regressions from twins’ RWA on shared environmental variance in
twins’ rated parenting (bC), which can reflect effects of parenting on RWA shared by twins,
shared environmental factors affecting both parenting and RWA, and/or passive genotype-
environment correlation that act as a function of siblings’ shared family environment (Briley
et al., 2018). Controlling for these confounding genetic (a × bA) and environmental links (cP ×
bC), a regression from RWA on retrospective experiences of parenting (b) would suggest a
quasi-causal effect in terms of non-confoundedness regarding genetic and shared
environmental influences (see Turkheimer & Harden, 2014, for more details). That is, twin
differences in retrospectively reported parenting would predict twin differences in the level of
RWA. These models allow us to infer only quasi-causality, because they cannot exclude
possible confounding factors not shared by twins, for example a differential memory bias of
retrospective information. However, in any case, those confounding factors must be
effectively environmental.
The bivariate twin model analyses were run on the basis of twins’ retrospective
assessments of parental responsiveness and demandingness on the one hand and self- and
peer reports on twins’ RWA on the other, testing Hypotheses 1.1, 1.2, and 2. For all four
associations, the full bivariate twin model was tested against three reduced regression models
nested within the full model: (1) Latent A regression model: bC = 0; (2) Latent C regression
model: bA = 0; and (3) Phenotypic regression model: bA = bC = 0. The phenotypic regression
model, where both latent bA and bC regression paths were fixed to zero, was the most
restrictive model not allowing for confounding genetic and shared environmental factors. It
tested Hypotheses 1.1 (b < 0) and 1.2 (b > 0). The model comparison between the phenotypic
regression model and more complex models tested for the quasi-causal effect from parenting
on RWA. The latent A regression model allowed for genetic factors (partially) accounting for
QUASI-CAUSAL EFFECTS OF DIFFERENTIAL PARENTING ON RWA 20
the association, testing Hypothesis 2 (bA ≠ 0). The latent C regression model controlled for
shared environmental factors (partially) explaining the association. Likewise, the full (latent
A+C regression) model tested for both confounding factors.
The overall model fit was evaluated by the root mean square error of approximation
(RMSEA) and the comparative fit index (CFI). A good model fit would be indicated by
RMSEA < .05 and CFI > .95 (Hu & Bentler, 1999; Steiger, 1990). Nested model
comparisons were run using the χ²-difference test. In other words, the phenotypic regression
model was tested against the latent A regression model or the latent C regression model,
which in turn were tested against the full regression model. In addition, we used the expected
cross-validation index (ECVI; Browne & Cudeck, 1993) and the CFI for descriptive
comparisons of non-nested models. A higher CFI and a smaller ECVI indicate a better model
fit. Thus, the latent A regression model was compared to the latent C regression model in this
regard.
Phenotypic semilatent multitrait-multimethod analyses. In order to further
complement findings from the genetically informative regression models, we inspected the
association for several confounding factors by running phenotypic semilatent MTMM
analyses for both parenting dimensions separately by considering several rater perspectives
for parenting as well as for the offspring’s RWA (see Figure 3). The latent variable parenting
accounts for the correlations between parents’ and twins’ ratings of parenting, whereas the
latent variable twin’s RWA accounts for the common variance in self- and peer reports on
twins’ RWA. By including several rater assessments for both the predictor and the outcome,
rater-specific effects pertaining the self-rater method could be taken into account. These
include global rater biases and specific response biases associated with the criterion (i.e.,
RWA) itself. Accordingly, the model allows for common method variance in twins’ parenting
rating and their RWA self-rating (mt) as well as for the influence of the twin’s latent RWA on
QUASI-CAUSAL EFFECTS OF DIFFERENTIAL PARENTING ON RWA 21
the twin’s parenting rating via regression of twin’s parenting rating on the latent variable
twin’s RWA (atp). If the latent regression of twin’s RWA on parenting (b) remains significant
in the presence of potentially confounding global and specific response biases, despite
concomitant effects possibly counteracting the association’s direction and consequently
suppressing the regression, the association between parenting and twin’s RWA can be
considered to be valid across different rater perspectives. If a significant genetic mediation
between parenting and offspring’s RWA is found (Hypothesis 2), a significant latent
regression b may reflect evocative genotype-environment correlation, in line with Hypothesis
3.
To disentangle different kinds of genotype-environment correlations as potential
accounts of the association between parenting and RWA, the aforementioned model was
extended by including self-rated parental RWA (see Figure 4). This model allows for
horizontal influences of maternal and paternal RWA on parenting (hf, hm), vertical influences
of parental RWA on the offspring’s RWA (vf, vm), and consequently two further potentially
confounding factors of the association between parenting and twin’s RWA, namely mother’s
and father’s RWA, can be taken into account. In addition, the model takes account of global
response biases reflected by residual correlations between parental ratings of their parenting
and own RWA (mf, mm). Thus, this model allowed to rule out further confounding effects due
to rater-specific perspectives, a vertical transmission of RWA and a mediation of this vertical
transmission via parenting. If Hypothesis 2 can be supported, a significant latent regression b
without horizontal effects (i.e., h = 0) may reflect evocative genotype-environment
correlation, whereas a shared environmental mediation in the genetically informative
regression model and a mediation via parental RWA in the phenotypic semilatent MTMM
analysis (i.e., h ≠ 0 and v ≠ 0) would support a passive genotype-environment correlation, in
line with Hypothesis 3.
QUASI-CAUSAL EFFECTS OF DIFFERENTIAL PARENTING ON RWA 22
Both phenotypic model analyses were first run separately for each twin of a pair.
Then, we constrained the model parameters to be equal across twin siblings as quasi-cross-
validation of the estimated effects. Furthermore, the full models were compared with
different reduced models by fixing parameters that turned out to be non-significant in the full
model. The overall model fit was evaluated by the RMSEA and CFI. Nested model
comparisons were based on the χ²-difference tests.
Results
Genetically Informative Regression Analyses
The best fitting models for the association between each parenting dimension and
RWA within twins’ self-reports and across rater perspectives are reported in Tables 3 and 4
and standardized estimates are reported in Figures 5 and 6. For the other three model results
see online supplement B. All models provided good to excellent model fits. Model estimates
yielded significant genetic and nonshared environmental contributions to the variance in self-
reported RWA and significant but smaller genetic and larger nonshared environmental
contributions as well as significant shared environmental influences to the variance in peer-
reported RWA.
Responsiveness and RWA. The phenotypic regression model yielded a significant
positive prediction of twin’s RWA by responsiveness for self-reports (b = .163, SE = .038, p
< .001) and peer-reports (b = .179, SE = .042, p < .001), not confirming Hypothesis 1.1. For
the association between twins’ RWA and responsiveness, the latent A regression model
provided the best model fit (for self-reports: χ² = 16.355, df = 12, p = .175, CFI = .989,
RMSEA = .030, ECVI = .123; for peer reports: χ² = 13.956, df = 12, p = .304, CFI = .994,
RMSEA = .020, ECVI = .117). It yielded a significantly better model fit compared to the
phenotypic regression model (Δχ² =11.279, Δdf = 1, p = .001 for self-reports, and Δχ² =
4.270, Δdf = 1, p = .039 for peer reports) and did not yield a significantly worse model fit
QUASI-CAUSAL EFFECTS OF DIFFERENTIAL PARENTING ON RWA 23
compared to the latent A+C regression model (Δχ² =0.672, Δdf = 1, p = .412 for self-reports,
and Δχ² = 0.007, Δdf = 1, p = .935 for peer reports). Moreover, the model showed a higher
CFI as well as a smaller ECVI than the latent C regression model (self-reports: CFI = .989 vs.
.981; ECVI = .123 vs. .132; peer reports: CFI = .994 vs. 988; ECVI = .117 vs. 122). Hence, in
line with Hypothesis 2, we found that genetic factors accounted for the association between
responsiveness and RWA. That is, twins higher on RWA due to genetic sources also reported
a higher level of experienced responsiveness.
Demandingness and RWA. The phenotypic regression model yielded a significant
positive prediction of twin’s RWA by demandingness for self-reports (b = .161, SE = .035, p
< .001) and a non-significant positive prediction for peer-reports (b = .069, SE = .040, p =
.088), confirming Hypothesis 1.2. Since Hypothesis 1.2 was directed (b > 0), the effect can be
treated as one-tailed significant (p = .044). For the association between the twins’ RWA and
demandingness, the phenotypic regression model represented the best fitting model (for self-
reports: χ² = 24.012, df = 13, p = .031, CFI = .963, RMSEA = .046, ECVI = .138; for peer
reports: χ² = 15.414, df = 13, p = .282, CFI = .987, RMSEA = .022, ECVI = .116). All three
more complex models did not provide significantly better model fits (A+C regression model,
Δχ² = 0.331, Δdf = 2, p = .847 for self-reports, and Δχ² = 1.711, Δdf = 2, p = .425 for peer
reports; latent A regression model, Δχ² = 0.093, Δdf = 1, p = .761 for self-reports, and Δχ² =
1.279, Δdf = 1, p = .258 for peer reports; latent C regression model, Δχ² = 0.004., Δdf = 1, p =
.949 for self-reports, and Δχ² = 1.711, Δdf = 1, p = .191 for peer reports). In consequence,
contrary to Hypotheses 2 and 3, twin differences in retrospectively reported demandingness
were associated with twin differences in twins’ self-reports on RWA beyond genetic and
shared environmental effects, indicating a small but significant quasi-causal positive effect
from demandingness on RWA. That is, twins who reported experiences of more
QUASI-CAUSAL EFFECTS OF DIFFERENTIAL PARENTING ON RWA 24
demandingness compared to their twin-siblings showed higher levels of RWA than their co-
twins.
Phenotypic Semilatent Multitrait-Multimethod Analyses
Responsiveness and RWA. Both the model not including parental RWA and the
model including parental RWA were successfully cross-validated, as reflected by no
significantly worse fit of the models upon the assumption of the same effects for both twin
siblings (for the model not including parental RWA: Δχ2 = 6.291, Δdf = 8, p = .615; for the
model including parental RWA: Δχ2 = 8.139, Δdf = 13, p = .834). As a consequence, results
for model parameters constrained to be equal across twins are reported in the following
sections. See supplement C for results for each twin sibling and supplement D for
correlations between offspring’s and parents’ ratings of parenting and offspring’s self- and
peer-assessed RWA and parental RWA. See Table 5 for unstandardized model parameter
estimates and Figures 7A and 7B for standardized estimates of the model not taking parental
RWA into account and the model including parental RWA. Both models showed an excellent
overall model fit (Figure 7A: χ2 = 9.945, df = 18, p = .934, CFI = 1.000, RMSEA = .000;
Figure 7B: χ2 = 18.031, df = 42, p > .999, CFI = 1.000, RMSEA = .000).
For both models, similar to the results of the genetically informative regression
models, we found the twins’ RWA to be significantly positively predicted by responsiveness
(b = .189–.271, SE = .048–.067, p < .001). Regarding global rater effects, such as
acquiescence or social desirability, the analyses did not yield a significant residual covariance
between the twins’ reports on responsiveness and their RWA rating (mt = .086, SE = .060, p =
.150) for the model not including parental RWA, but did yield it for the model including
parental RWA (mt = .113, SE = .057, p = .048). In any case, correlations were rather small.
Considering the specific rater effect of twins’ RWA on twins’ responsiveness rating (atp =
.099, SE = .071, p = .164, and atp = .095, SE = .068, p = .167), we did not find a significant
QUASI-CAUSAL EFFECTS OF DIFFERENTIAL PARENTING ON RWA 25
association for either model. Thus, specific response biases due to twin’s RWA did not seem
to have an influence on the twin’s assessment of parental responsiveness. Regarding the
association between parental RWA and the residual variance in parents’ own responsiveness
assessment, we did not find a significant association for the twins’ mothers (mm = .074, SE =
.051, p = .144), but did find it for the twins’ fathers (mf = .130, SE = .050, p = .009),
suggesting that global response biases and potentially specific response biases due to the
parent’s RWA had an influence on both paternal assessments. However, again these
correlations were rather small.
For the model not including parental RWA, the model fit did not significantly worsen
when effects of either global or specific response biases were excluded (global: Δχ2 = 1.901,
Δdf = 1, p = .168; specific: Δχ2 = 1.647, Δdf = 1, p = .199), but upon exclusion of both effects
(Δχ2 = 6.297, Δdf = 2, p = .043). For the model including parental RWA, the exclusion of
effects due to the influence of twins’ RWA on twins’ responsiveness rating (i.e., specific
response biases) did not worsen the model fit (Δχ2 = 1.582, Δdf = 1, p = .209), but the
exclusion of effects due to parent’s and twin’s global response biases did (Δχ2 = 14.936, Δdf
= 2, p = .002). Thus, even though the effects were rather small, response biases have to be
taken into account when analyzing the regression of RWA on retrospectively reported
responsiveness, because they artificially act to increase positive correlations, inflating
positive associations between parental responsiveness and RWA to some degree. However,
their general contribution was rather small and the latent regression of twin’s RWA on
responsiveness remained statistically significant and moderate in effect size (β = .252 and β =
.233; see Figures 7A and 7B).
The analyses including parents’ RWA self-reports revealed significant vertical effects
of the parental RWA on offspring’s RWA; both the maternal and paternal RWA positively
predicted the offspring’s RWA, with father’s RWA to a lesser degree. In line with a vertical
QUASI-CAUSAL EFFECTS OF DIFFERENTIAL PARENTING ON RWA 26
genetic transmission from parents to offspring (c.f. Tables 3 and 4), parents’ self-reported
RWA accounted for about 33% of individual differences in offspring’s RWA (see Figures 5
and 7B: vm2 + vf2 + 2 × vm × vf × cova = .457² + .175² + 2 × .457 × .175 × .529 = .325).
Concerning indications for horizontal effects, the maternal RWA did not have a
significant influence on responsiveness (hm = .078, SE = .050, p = .118), while paternal RWA
had a significant negative influence on responsiveness (hf = −.125, SE = .049, p = .011). The
latter indicated that fathers high on RWA showed lower levels of (latent) responsiveness,
contrasting the positive association between fathers’ RWA scores and residual variance in
paternal responsiveness rating, indicating that fathers high on RWA reported higher levels of
responsiveness. However, the horizontal effects of mothers’ and fathers’ RWA on parental
responsiveness were generally small, accounting for less than 3% of the variance in
responsiveness (see Figures 5 and 7B: hm2 + hf2 + 2 × hm × hf × cova = .120² + [–.193]2 + 2 ×
.120 × [–.193] × .529 = .027). Hence, given the complete genetic confounding of the
association between responsiveness and twins’ RWA (see Table 3 and Figure 5A), passive
genotype-environment correlation could only account for a small proportion of the
association between responsiveness and twin’s RWA and tended – if at all – to suppress the
positive correlation (see Figures 5 and 7B: hm × vm + hf × vf + hm × cova × vf + hf × cova × vm =
.120 × .457 + [–.193] × .175 + .120 × .529 × .175 + [–.193] × .529 × .457 = –.014).
5
In sum,
the findings pointed towards an alternative explaining mechanism of the positive genetic link
5
The negative horizontal effect from father’s RWA (compared to the non-significant horizontal effect from
mother’s RWA) on responsiveness indicated diverging and contrasting contributions of specific passive
genotype-environment correlations depending on the specific parent. The paternal influence resulted in a
negative passive genotype-environment correlation (see Figures 5 and 7B: hf × vf + hf × cova × vm = [–.193] ×
.175 + [–.193] × .529 × .457 = -.080), whereas the maternal influence indicated a positive but non-significant
passive genotype-environment correlation (see Figures 5 and 7B: hm × vm + hm × cova × vf = .120 × .457 + .120 ×
.529 × .175 = .066). Thus, the paternal horizontal effect tended to suppress the positive link between
responsiveness and twin’s RWA, whereas the maternal horizontal effect tended to increase the positive link.
However, both contributions were rather small and need to be replicated by future research before drawing
definitive conclusions.
QUASI-CAUSAL EFFECTS OF DIFFERENTIAL PARENTING ON RWA 27
between parental responsiveness and offspring’s RWA, such as evocative genotype-
environment correlation.
Demandingness and RWA. For demandingness, both the model not including
parental RWA (Δχ2 = 7.977, Δdf = 8, p = .436) and the model including parental RWA (Δχ2 =
9.147, Δdf = 13, p = .762) were successfully cross-validated. For estimates of each twin of a
pair, see supplement C; for correlations between offspring’s and parents’ reports on parenting
and offspring’s self- and peer-assessed RWA and parental RWA, see supplement D.
Unstandardized model parameter estimates are reported in Table 6 and standardized estimates
for both models are shown in Figures 8A and 8B. The fit of both models was excellent
(Figure 8A: χ2 = 15.461, df = 18, p = .859, CFI = 1.000, RMSEA = .000; Figure 8B: χ2 =
24.539, df = 42, p = .986, CFI = 1.000, RMSEA = .000).
Again, the model fit worsened for both the model not including parental RWA (Δχ2 =
6.519, Δdf = 2, p = .038) and the model including parental RWA (Δχ2 = 16.627, Δdf = 4, p =
.002), when effects due to global and specific response biases were excluded, suggesting
some influences of both global and specific response biases. More specifically, the direction
of effects indicated that global response tendencies acted to increase within-rater associations
between demandingness and RWA ratings. In other words, higher RWA self-ratings came
along with higher demandingness ratings. In contrast, specific response biases due to twins’
RWA acted to decrease the twins’ demandingness rating. In other words, twins higher on
RWA reported lower levels of demandingness, suppressing a positive association between
parental demandingness and twin’s RWA within self-reports. Hence, global rater biases acted
to increase positive correlations, whereas specific rater biases acted to decrease positive
correlations. Taking the effects due to rater biases into account, the latent regression between
demandingness and offspring’s RWA was significant and even larger than would be expected
on the basis of single rater studies (β = .430 and β = .160; see Figures 8A and 8B).
QUASI-CAUSAL EFFECTS OF DIFFERENTIAL PARENTING ON RWA 28
Whereas demandingness positively predicted twins’ RWA in both models, the
association was found to considerably weaken when parental RWA was taken into account (β
= .430 vs. β = .160). In addition to the already mentioned vertical effects from parental RWA
on offspring’s RWA, we found horizontal effects of the maternal RWA (hm = .183, SE =
.046, p < .001) and the paternal RWA (hf = .189, SE = .046, p < .001) on demandingness:
More right-wing authoritarian parents showed more demandingness. The effect of parents’ –
in particular the mother’s – RWA on both demandingness and the twin’s RWA thus partially
accounted for the association between twins’ RWA and demandingness (see Figures 6 and
8B: hm × vm + hf × vf + hm × cova × vf + hf × cova × vm = .296 × .436 + .306 × .087 + .296 ×
.531 × .087 + .306 × .531 × .436 = .240), indicating that a substantial proportion of the link
between parental demandingness and offspring’s RWA was driven by parental RWA. This
partial explanation by parents’ RWA could not be attributed to passive genotype-environment
correlation, because the genetically informative regression model analysis yielded a quasi-
causal environmental effect from experienced demandingness on RWA: Differential
experiences of demandingness acted to increase differences in RWA within twin pairs. That
is, even though more right-wing authoritarian parents tended to show more demandingness
and tended to have offspring higher on RWA, the association between retrospectively
reported demandingness and twin’s RWA was effectively environmental.
Discussion
Investigating the influence of differential parenting on differences in RWA using a
genetically informative twin family multi-rater design, our analyses yielded heterogeneous
implications for both parenting dimensions. We could confirm a genetic contribution to the
positive association between twins’ retrospectively reported responsiveness and self-reports
as well as peer reports on twins’ RWA, inconsistent with Hypothesis 1.1, but consistent with
Hypothesis 2. In accordance with Hypothesis 3, the link between responsiveness and twins’
QUASI-CAUSAL EFFECTS OF DIFFERENTIAL PARENTING ON RWA 29
RWA may reflect evocative genotype-environment correlation, since it was primarily
genetically mediated, remained significant after controlling for rater biases, and could only be
marginally accounted for by parental RWA. Evocative genotype-environment transaction
mechanisms during development may result in stable individual differences in RWA. In
contrast, we found a quasi-causal environmental effect of individual differences in
retrospectively reported demandingness on individual differences in offspring’s current
RWA, consistent with Hypothesis 1.2, but inconsistent with Hypothesis 2 and 3.
Responsiveness and RWA: Evocative Genotype-Environment Correlation
Although we found some evidence for global response biases such as acquiescence or
severity, our analyses indicate that heritable rater biases cannot explain the genetic link
between experiences of responsiveness and RWA. Similarly, the analyses yielded little
evidence for passive genotype-environment correlation accounting for the positive link
between retrospectively perceived parental responsiveness and offspring’s RWA. As a
consequence, our findings suggest that the positive association was mainly attributable to an
evocative genotype-environment correlation. In other words, parents responded more
positively to offspring with higher levels of RWA. That is, while the parental RWA did not
substantially influence latent responsiveness, the offspring’s partly heritable right-wing
authoritarian behavior, for example compliant, conforming behavior concerning rule
adherence, could have been rewarded with more responsive parental behavior; or conversely,
displayed disobedience may have led to more conflicts and less emotional warmth towards
the offspring. These individual differences in reactive parental responsiveness may have
contributed to the enhancement of initial genetically based individual differences in RWA.
This interpretation is in line with recent molecular genetic findings on the existence of
evocative genotype-environment correlation in the context of parenting behavior (e.g., Hajal
et al., 2015; Kopala-Sibley, Hayden, Singh, Sheikh, Kryski, & Klein, 2017; Pener-Tessler et
QUASI-CAUSAL EFFECTS OF DIFFERENTIAL PARENTING ON RWA 30
al., 2013). For example, Pener-Tessler et al. (2013) showed for boys that the genotype of
Serotonin Transporter Linked Polymorphic Region (5-HTTLPR), a polymorphic region in the
gene that encodes the serotonin transporter, affected the level of positive parental treatment,
partly mediated through the boys’ self-control.
Consistent with previous research implementing offspring’s and parents’ ratings on
responsiveness and related parenting constructs (e.g., Hur and Bouchard, 1995; Kendler,
1996; Plomin, McClearn, Pedersen, Nesselroade, and Bergeman, 1989; Rowe, 1983; for
meta-analyses, see Avinun and Knafo, 2014; Kendler and Baker, 2007; Klahr and Burt,
2014), we found responsiveness to be moderately genetically influenced. In a study taking
parental ratings, twins’ and co-twins’ ratings of the twins’ parenting into account, Kendler
(1996) concluded that this observation is mainly attributable to identical twin siblings
evoking more similar parenting behaviors than fraternal twin siblings, in other words: an
evocative genotype-environment correlation. The presence of evocative genotype-
environment correlation when investigating responsiveness has been repeatedly reported,
especially in early childhood research (see Klahr and Burt, 2014, for a thorough discussion).
Thus, responsiveness might be especially susceptible to certain genetically influenced
dispositions of the offspring in general. Our study provided further support and extends this
picture to individual differences in adult RWA.
Demandingness and RWA
A quasi-causal environmental link. First and foremost, using a genetically
informative regression model, we could exclude genetic and environmental factors shared by
twins as factors potentially confounding the association between demandingness and
offspring’s RWA. Accordingly, we found a positive quasi-causal environmental effect of
differences between twin siblings in experienced demandingness on differences in twin
siblings’ RWA. Interestingly, parental RWA partially accounted for the association, as
QUASI-CAUSAL EFFECTS OF DIFFERENTIAL PARENTING ON RWA 31
revealed by the semilatent multi-rater model analyses. Against the background of the findings
from the genetically informative analyses, this cannot be explained by genotype-environment
correlation, but by nonshared experiences and phenotype-environment correlation.
The semilatent MTMM analyses yielded some contributions of global rater biases in
the form of within-rater correlations that acted to artificially increase single-rater correlations
due to acquiescence or other response styles, and specific rater biases in the form of negative
correlations between perceived parenting and RWA due to the actual outcome characteristic.
That is, being right-wing authoritarian resulted in retrospective reports of lower parental
demandingness, suppressing positive correlations between experienced demandingness and
RWA. This finding supports the idea that offspring high on RWA would probably not cast
their parents in a negative light, since parents represent authorities, which are positively
evaluated by individuals high on RWA (Duriez et al., 2007).
Parenting and parental RWA as effectively nonshared experiences. We found the
association between parental demandingness and offspring’s RWA to be partly attributable to
mothers’ and fathers’ RWA. Parents high on RWA generally showed more demanding
behaviors, which is plausible in light of the held values of social conformity and obedience.
In contrast, more right-wing authoritarian adult children reported less experienced
demandingness, which alludes to a possible relativization or misremembering of their
parenting as a consequence of their inclination to characterize authority figures such as their
parents in a favorable light. In addition, their individual experiences as parents as well as the
peer environment in adulthood primarily not shared by adult twins might have influenced the
interpretation of certain parental behaviors as less or more demanding. In line with this,
Duckitt (2001) argued in his dual process model that a punitive (vs. tolerant) parental
socialization would lead to a personality disposition inclined to be socially conforming,
resulting in a heightened sensitivity to conformity violations posing a threat to societal
QUASI-CAUSAL EFFECTS OF DIFFERENTIAL PARENTING ON RWA 32
structures. Both the disposition and the sensitivity would lead to a salience of the
motivational goal of security and social control, and culminate in heightened RWA.
Other nonshared environmental influences. It should be noted that the link between
parental demandingness and RWA might be attributable to third variables not shared between
twin siblings affecting both differences in perceptions of parental demandingness and
differences in offspring’s RWA, for example peer influences. Fuligni and Eccles (1993)
reported a negative relationship between perceived parental strictness and monitoring and
extreme peer orientation. This might not only affect the development of sociopolitical
attitudes, but a higher conformity with one’s peers might also lead to noncompliance with
parental expectations, rules and norms, potentially further increasing parental demandingness.
Altemeyer (1988) reported a significant correlation between close friends’ RWA (r = .31),
indicating influences due to the interaction (i.e., selection and socialization) with peers. Yet,
to what extent peer socialization affects the development of personality traits in comparison
to and interplay with parental socialization is controversially discussed (e.g., Harris, 1995;
Vandell, 2000).
Evocative phenotype-environment correlation. A number of studies applying a MZ
twin differences design investigated the association between differences in reported and/or
observed parenting and differences in the children’s behavior (e.g., Asbury, Dunn, Pike, and
Plomin, 2003; Mullineaux, Deater-Deckard, Petrill, and Thompson, 2009). For example,
Mullineaux and colleagues (2009) conducted a multi-rater study on the association between
maternal parenting and children’s problematic and adaptive behaviors. They found a positive
bidirectional association between changes in maternal authoritarian parenting and changes in
children’s non-compliance to maternal verbalizations. In addition, they reported a positive
bidirectional association between changes in maternal authoritative parenting and children’s
autonomy, attentiveness and engagement in the task. Asbury, Dunn, Pike, and Plomin (2003)
QUASI-CAUSAL EFFECTS OF DIFFERENTIAL PARENTING ON RWA 33
found differences in parental discipline to account for 5% of the variance for conduct
problems that varied extensively between twins. Cross-lagged twin models also reflected the
bidirectional nature of the association between parenting and offspring’s outcomes, for
example for parental negativity and offspring’s antisocial behavior (e.g., Burt, McGue,
Krueger, and Iacono, 2005; Larsson, Viding, Rijsdijk, & Plomin, 2008). Thus, offspring’s
idiosyncratic behaviors incongruent with parental esteem of obedience and conformity may
evoke a more demanding behavior – especially in parents high on RWA – which more
strongly reinforces their offspring’s obedient and conforming attitudes relative to their other
offspring. As a consequence, small twin differences may increase over time.
Limitations and Future Outlook
While our multi-rater twin design exhibits a number of strengths and our results will
hopefully generate new intriguing questions, several limitations should be mentioned that
also point to directions for future research. First, drawing on multiple perspectives broadens
the analyses, allows for more far-reaching interpretations of certain complex associations,
and is beneficial for research subjects in which self-report measures might considerably bias
the relevant psychological constructs, for example the association between demandingness
and RWA. Multi-generational studies will be helpful to further illuminate dynamics between
parental dispositions, experienced contextual characteristics, and offspring’s dispositional and
behavioral outcomes. The investigation of both parenting dimensions disentangled from each
other as opposed to a categorical manner, might help to overcome the lack of a found
evocative genotype-environment correlation for experienced warmth in observational studies
(Avinun and Knafo, 2014).
Second, as Klahr and Burt (2014) discussed, the implementation of different parenting
measures may result in rather discrepant results. We decided to shorten an existing
questionnaire in favor of a narrower construct of parenting. The development of a balanced
QUASI-CAUSAL EFFECTS OF DIFFERENTIAL PARENTING ON RWA 34
measure for both parenting dimensions is statistically advisable, especially one in which strict
and harsh demandingness is measured separately, as this differentiation might have distinct
dispositional outcomes (e.g., Duckitt, 2001). In addition, scale content differed slightly
between offspring’s and parents’ version, since parents did not assess their spouse’s
parenting. Thus, we latently modeled parenting as a more global family environmental
component that is agreed upon by all family members, as opposed to the sum of agreements
on maternal, paternal and parent-unspecific parenting. On the one hand, this might be
advantageous in the context of this study, since the impact of parent-related effects is hard to
determine when specific aspects, for example time spent together, cannot be taken into
account. On the other hand, it might be more worthwhile to investigate parent-related effects,
which should be investigated in the future (see below). In addition, this might have led to
additional valid variance not considered for the association between parenting and offspring’s
RWA.
Third, we captured offspring’s RWA in adulthood and parenting as retrospective
measure. Even though the found phenotypic links are in line with studies capturing parenting
and offspring’s behavior in childhood and adolescence (Duriez et al., 2007), the retrospective
perspective might be biased by for example “softening” or selective remembrance (see
above), consequently diminishing the “true” link between parenting and RWA in our study,
and additionally might not capture critical age-specific consolidating processes
predetermining future remembrance mechanisms. Yet, Avinun and Knafo (2014) discussed
that retrospective parenting assessments might be advantageous in the sense that they capture
more general parenting behavior assessments as opposed to contemporaneous accounts that
may be subject to recurrent fluctuations and prevailing conflicts. This would explain why
heritability estimates are smaller (Klahr & Burt, 2014) when parenting is reported by children
and adolescents.
QUASI-CAUSAL EFFECTS OF DIFFERENTIAL PARENTING ON RWA 35
Fourth, our data, albeit genetically informed and including multiple rater perspectives,
thus allowing to test phenotypic associations for quasi-causality in terms of non-
confoundedness and to explore diverse explanations, were ultimately cross-sectional, which
does not allow for definitive implications of causality. Since the parenting style – among
others – underlies the influence of the offspring’s behavior (Klahr and Burt, 2014) and even
the offspring’s attitudes (e.g., Degner & Dalege, 2013), longitudinal studies can shed more
light upon the interaction of parenting and attitude or, more generally, personality
development – especially with regard to the dynamic nature of evocative phenotype-
environment and genotype-environment correlations.
Fifth, the relatively small sample size and the concomitant statistical power led to
analytical restrictions. We did not combine the genetically informative regression model and
the semilatent MTMM model into one model analysis – primarily because parents did not
rate their parenting for each twin sibling separately, thus not allowing for a twin-difference
perspective, but also – due to the lack of statistical power for such complex model analyses.
This is also why we decided not to control for a potential type 1 error inflation. In addition,
we could not analyze the impact of offspring’s sex. The investigation of within-sex and cross-
sex associations might offer an even more multifarious pattern, as has been indicated by
previous research (e.g., Klahr and Burt, 2014; Laible and Carlo, 2004; Paulson and Sputa,
1998; Russell et al., 1998). For instance, differential effects of mothers’ and fathers’ RWA on
the association between parenting and offspring’s RWA might depend on offspring’s sex.
Moreover, regarding retrospective child reports, Avinun and Knafo (2014) reported different
genetic and environmental contributions depending on the parent’s sex, with maternal
parenting being more genetically influenced and attributable to nonshared environmental
factors and paternal parenting being more affected by shared environmental factors. This
difference might indicate differential mechanisms involving parent-offspring interactions and
QUASI-CAUSAL EFFECTS OF DIFFERENTIAL PARENTING ON RWA 36
different links between parental and offspring’s trait, as indicated by our phenotypic MTMM
analyses, and should be illuminated by future investigations.
Finally, we relied on a Western sample. Since the impact of different parenting
behaviors and parenting itself might have different effects on offspring’s outcomes depending
on the socio-cultural context (Putnick et al., 2012; Rudy & Grusec, 2001; Sorkhabi, 2005),
future research should focus on the cultural influence on the interplay of parenting and RWA.
Conclusion
The applied design in the current study helped to approximate a quasi-causal inference
in terms of non-confoundedness of the association between differential parenting and RWA.
We tested for potentially underlying factors confounding the association in the form of a
genetic and shared environmental factors and subsequently examined the found association
for more in-depth explanations, for example common method biases in terms of global and
specific rater biases, passive or evocative genotype-environment correlations.
We found a positive association between retrospectively reported responsiveness and
RWA to be completely attributable to common genetic factors, which reflected neither
heritable rater biases nor passive genotype-environment correlation to a substantial degree,
pointing to evocative genotype-environment correlation as explanation: More right-wing
authoritarian, in other words more compliant and obedient, offspring elicited more responsive
behaviors from their parents. Moreover, we found a positive association between
retrospectively reported demandingness and RWA unconfounded by genetic factors and
consequently quasi-causally environmental: Twin siblings who experienced more
demandingness than their co-twins showed higher levels of right-wing authoritarianism.
Undoubtedly, the “true” impact of family environments has engaged researchers for
decades. Findings showing that every trait is heritable and that theoretically shared
environmental factors are negligible do not tell the whole story, especially considering
QUASI-CAUSAL EFFECTS OF DIFFERENTIAL PARENTING ON RWA 37
genotype-environment interplay and interindividually different experiences. Disentangling
genetic effects and identifying possibly idiosyncratic experiences within the “shared”
environment between siblings is vital in order to illuminate the etiology of human complex
traits, such as ideological attitudes. The current study highlights the use of genetically
informative, multi-rater twin family designs – in particular, when investigating effects of
measured family environments on psychological outcomes of the offspring. Such complex
and multifaceted constructs require equal complexity in design and analyses. As Turkheimer
and Waldron (2000) put it:
The limitations of our existing social scientific methodologies ought not provoke us to
wish that human behavior were simpler than we know it to be; instead they should
provoke us to search for methodologies that are adequate to the task of understanding
the exquisite complexity of human development. (p. 93)
QUASI-CAUSAL EFFECTS OF DIFFERENTIAL PARENTING ON RWA 38
References
Adorno, T. W., Frenkel-Brunswik, E., Levinson, D., & Sanford, R. N. (1950). The
authoritarian personality. New York, NY: Harper & Row.
Altemeyer, B. (1988). Enemies of freedom: Understanding right-wing authoritarianism. San
Francisco, CA: Jossey-Bass.
Altemeyer, B. (1996). The authoritarian specter. Cambridge, MA: Harvard University Press.
Altemeyer, B. (1998). The other “authoritarian personality”. In M. P. Zanna (Ed.), Advances
in experimental social psychology, (Vol. 30, pp. 47–92). San Diego, CA: Academic
Press.
Anderson, T. W., & Rubin, H. (1956). Statistical inference in factors analysis. In J. Neyman
(Ed.), Proceedings of the third Berkeley Symposium of Mathematical Statistics and
Probability: Vol. 5. Contributions to econometrics, industrial research, and
psychometry (pp. 111–150). Berkeley, CA: University of California Press.
Arbuckle, J. L. (2012). IBM SPSS Amos 21 User’s Guide. Amos Development Corporation.
Asbury, K., Dunn, J. F., Pike, A., & Plomin, R. (2003). Nonshared environmental influences
on individual differences in early behavioral development: A monozygotic twin
differences study. Child Development, 74(3), 933–943. doi:10.1111/1467-8624.00577
Avinun, R., & Knafo, A. (2014). Parenting as a reaction evoked by children’s genotype: A
meta-analysis of children-as-twins studies. Personality and Social Psychology
Review, 18(1), 87–102. doi:0.1177/1088868313498308
Baumrind, D. (1971). Current patterns of parental authority. Developmental Psychology
Monographs, 4(1, Pt. 2), 1–103. doi:10.1037/h0030372
Bandura, A. (1977). Social learning theory. Englewood Cliffs, NJ: Prentica-Hall.
QUASI-CAUSAL EFFECTS OF DIFFERENTIAL PARENTING ON RWA 39
Baumrind, D. (1991). The influence of parenting style on adolescent competence and
substance use. The Journal of Early Adolescence, 11, 56–95.
doi:10.1177/0272431691111004
Baumrind, D. (2013). Authoritative parenting revisited: History and current status. In R. E.
Larzelere, A. S. Morris, & A. W. Harrist (Eds.), Authoritative parenting: Synthesizing
nurturance and discipline for optimal child development (pp. 11–34). Washington,
DC: American Psychological Association Press.
Bleidorn, W., Hufer, A., Kandler, C., Hopwood, C. J., & Riemann, R. (2018). A nuclear twin
family study of self-esteem. European Journal of Personality, early view
doi:10.1002/per.2136
Bleidorn, W., Kandler, C., & Caspi, A. (2014). The behavioral genetics of personality
development in adulthood – classic, contemporary, and future trends. European
Journal of Personality, 28, 244–255. doi:10.1002/per.1957
Bouchard, T. J., & McGue, M. (2003). Genetic and environmental influences on human
psychological differences. Journal of Neurobiology, 54, 4–45. doi:10.1002/neu.10160
Briley, D., Livengood, J., & Derringer, J. (2018). Behavior genetic frameworks of causal
reasoning for personality psychology. European Journal of Personality, manuscript
submitted for publication in the same special issue.
Browne, M. W., & Cudeck, R. (1993). Alternative ways of assessing model fit. In K. A.
Bollen & J. S. Long (Eds.), Testing structural equation models (pp. 136–162).
Newbury Park, CA: Sage.
Burt, S. A., McGue, M., Krueger, R. F., & Iacono, W. G. (2005). How are parent-child
conflict and childhood externalizing symptoms related over time? Results from a
genetically informative cross-lagged study. Development and Psychopathology, 17(1),
145–165. doi:10.1017/S095457940505008X
QUASI-CAUSAL EFFECTS OF DIFFERENTIAL PARENTING ON RWA 40
Carvacho, H., Zick, A., Haye, A., González, R., Manzi, J., Kocik, C., & Bertl, M. (2013). On
the relation between social class and prejudice: The roles of education, income, and
ideological attitudes. European Journal of Social Psychology, 43(4), 272–285.
doi:10.1002/ejsp.1961
Cohrs, J. C., Kämpfe-Hargrave, N., & Riemann, R. (2012). Individual differences in
ideological attitudes and prejudice: Evidence from peer-report data. Journal of
Personality and Social Psychology, 103, 343–361. doi:10.1037/a0028706
Collins, W. A., Maccoby, E. E., Steinberg, L., Hetherington, E. M., & Bornstein, M. H.
(2000). Contemporary research on parenting: the case for nature and nurture.
American Psychologist, 55, 218–232. doi:10.1037/0003-066X.55.2.218
Darling, N., & Steinberg, L. (1993). Parenting style as context: An integrative model.
Psychological Bulletin, 113, 487–496. doi:10.1037/0033-2909.113.3.487
Degner, J., & Dalege, J. (2013). The apple does not fall far from the tree, or does it? A meta-
analysis of parent-child similarity in intergroup attitudes. Psychological Bulletin, 139,
1270–1304. doi:10.1037/a0031436
Doty, R. M., Peterson, B. E., & Winter, D. G. (1991). Threat and authoritarianism in the
United States, 1978–1987. Journal of Personality and Social Psychology, 61, 629–
640. doi:10.1037//0022-3514.61.4.629
Dunn, J., & Plomin, R. (1986). Determinants of maternal behaviour towards 3-year-old
siblings. British Journal of Developmental Psychology, 4, 127–137.
doi:10.1111/j.2044-835X.1986.tb01004.x
Duckitt, J. (2001). A dual-process cognitive-motivational theory of ideology and prejudice.
Advances in experimental social psychology, 33, 41–113. doi:10.1016/S0065-
2601(01)80004-6
QUASI-CAUSAL EFFECTS OF DIFFERENTIAL PARENTING ON RWA 41
Duriez, B., Soenens, B., & Vansteenkiste, M. (2007). In search of the antecedents of
adolescent authoritarianism: The relative contribution of parental goal promotion and
parenting style dimensions. European Journal of Personality, 21, 507–527.
doi:10.1002/per.623
Frenkel-Brunswik, E. (1950). Parents and Childhood as Seen Through the Interviews. In
Adorno, T. W., Frenkel-Brunswik, E., Levinson, D., & Sanford, R. N. (Eds.), The
authoritarian personality (pp. 337–389). New York, NY: Harper & Row.
Fuligni, A. J., & Eccles, J. S. (1993). Perceived parent-child relationships and early
adolescents' orientation toward peers. Developmental Psychology, 29(4), 622–632.
doi:10.1037/0012-1649.29.4.622
Funk, C. L., Smith, K. B., Alford, J. R., Hibbing, M. V., Eaton, N. R., Krueger, R. F., …
Hibbing, J. R. (2013). Genetic and environmental transmission of political
orientations. Political Psychology, 34, 805–819. doi:10.1111/j.1467-
9221.2012.00915.x
Funke, F. (2005). The dimensionality of right‐wing authoritarianism: Lessons from the
dilemma between theory and measurement. Political Psychology, 26, 195–218.
doi:10.1111/j.1467-9221.2005.00415.x
Hajal, N., Neiderhiser, J., Moore, G., Leve, L., Shaw, D., Harold, G., ... & Reiss, D. (2015).
Angry responses to infant challenges: Parent, marital, and child genetic factors
associated with harsh parenting. Child Development, 86(1), 80–93.
doi:10.1111/cdev.12345
Harmening, A. C. (2014). Eine deutsche Fassung des Block Environmental Questionnaire
(BEQ): Untersuchung der psychometrischen Qualitt (Unpublished master’s thesis).
Bielefeld University, Bielefeld, Germany.
QUASI-CAUSAL EFFECTS OF DIFFERENTIAL PARENTING ON RWA 42
Harris, J. R. (1995). Where is the child's environment? A group socialization theory of
development. Psychological Review, 102(3), 458–489. doi:10.1037/0033-
295X.102.3.458
Heydari, A., Teymoori, A., & Haghish, E. F. (2013). Socioeconomic status, perceived
parental control, and authoritarianism: Development of authoritarianism in Iranian
society. Asian Journal of Social Psychology, 16(3), 228–237. doi:10.1111/ajsp.12027
Hoff, E., Laursen, B., & Tardif, T. (2002). Socioeconomic status and parenting. In M. C.
Bornstein, Handbook of parenting (2nd edition), Volume 2: biology and ecology of
parenting (231–252). Mahwah, NJ: Erlbaum.
Hu, L., & Bentler, P. M. (1999). Cutoff criteria for fit indexes in covariance structure
analysis: Conventional criteria versus new alternatives. Structural Equation
Modeling: A Multidisciplinary Journal, 6, 1–55. doi:10.1080/10705519909540118
Hur, Y. M., & Bouchard, T. J. (1995). Genetic influences on perceptions of childhood family
environment: A reared apart twin study. Child Development, 66, 330–345.
doi:10.2307/1131581
Kandler, C., Bell, E., & Riemann, R. (2016). The structure and sources of right-wing
authoritarianism and social dominance orientation. European Journal of Personality,
30, 406-420. doi:10.1002/per.2061
Kandler, C., Gottschling, J., & Spinath, F. M. (2016). Genetic and environmental parent–
child transmission of value orientations: an extended twin family study. Child
Development, 87, 270–284. doi:10.1111/cdev.12452
Kandler, C., Lewis, G. J., Feldhaus, L. H., & Riemann, R. (2015). The genetic and
environmental roots of variance in negativity toward foreign nationals. Behavior
Genetics, 45, 181–199. doi:10.1007/s10519-014-9700-8
QUASI-CAUSAL EFFECTS OF DIFFERENTIAL PARENTING ON RWA 43
Kandler, C., Riemann, R., Spinath, F. M., & Angleitner, A. (2010). Sources of variance in
personality facets: A multiple-rater twin study of self-peer, peer-peer, and self-self
(dis)agreement. Journal of Personality, 78, 1565–1594. doi:10.1111/j.1467-
6494.2010.00661.x
Kandler, C., & Zapko-Willmes, A. (2017). Theoretical perspectives on the interplay of nature
and nurture in personality development. In: J. Specht (Ed.), Personality Development
across the Lifespan (pp. 101–115). Elsevier. doi:10.1016/B978-0-12-804674-6.00008
Kendler, K. S. (1996). Parenting: A genetic-epidemiologic perspective. The American
Journal of Psychiatry, 153(1), 11–20. doi:10.1176/ajp.153.1.11
Kendler, K. S., & Baker, J. H. (2007). Genetic influences on measures of the environment: A
systematic review. Psychological Medicine, 37(5), 615–626.
doi:10.1017/S0033291706009524
Klahr, A. M., & Burt, S. A. (2014). Elucidating the etiology of individual differences in
parenting: A meta-analysis of behavioral genetic research. Psychological Bulletin,
140, 544–586. doi:10.1037/a0034205
Kopala‐Sibley, D. C., Hayden, E. P., Singh, S. M., Sheikh, H. I., Kryski, K. R., & Klein, D.
N. (2017). Gene‐environment correlations in the cross‐generational transmission of
parenting: Grandparenting moderates the effect of child 5‐HTTLPR genotype on
mothers' parenting. Social Development, 26(4), 724–739. doi:10.1111/sode.12221
Kraemer, H. C., Measelle, J. R., Ablow, J. C., Essex, M. J., Boyce, W. T., & Kupfer, D. J.
(2003). A new approach to integrating data from multiple informants in psychiatric
assessment and research: Mixing and matching contexts and perspectives. American
Journal of Psychiatry, 160, 1566–1577. doi:10.1176/appi.ajp.160.9.1566
QUASI-CAUSAL EFFECTS OF DIFFERENTIAL PARENTING ON RWA 44
Laible, D. J., & Carlo, G. (2004). The differential relations of maternal and paternal support
and control to adolescent social competence, self-worth, and sympathy. Journal of
Adolescent Research, 19(6), 759–782. doi:10.1177/0743558403260094
Larsson, H., Viding, E., Rijsdijk, F. V., & Plomin, R. (2008). Relationships between parental
negativity and childhood antisocial behavior over time: A bidirectional effects model
in a longitudinal genetically informative design. Journal of Abnormal Child
Psychology, 36(5), 633–645. doi:10.1007/s10802-007-9151-2
Lewis, G. J., & Bates, T. C. (2013). Common heritable effects underpin concerns over norm
maintenance and in‐group favoritism: Evidence from genetic analyses of right‐wing
authoritarianism and traditionalism. Journal of Personality, 82, 297–309.
doi:10.1111/jopy.12055
Little, R. J. A., & Rubin, D. B. (2002). Statistical analysis with missing data. New York:
Wiley.
Ludeke, S. G., & Krueger, R. F. (2013). Authoritarianism as a personality trait: Evidence
from a longitudinal behavior genetic study. Personality and Individual Differences,
55, 480–484. doi:10.1016/j.paid.2013.04.015
Maccoby, E. E., & Martin, J. A. (1983). Socialization in the context of the family: Parent-
child interaction. In P. H. Mussen (Series Ed.) & E. M. Hetherington (Vol. Ed.),
Handbook of child psychology: Vol. 4. Socialization, personality, and social
development (4th edition, pp. 1–101). New York, NY: Wiley.
Manuel, L. (2006). Relationship of personal authoritarianism with parenting styles.
Psychological Reports, 98, 193–198. doi:10.2466/PR0.98.1.193-198
Martin, N. G., Eaves, L. J., Heath, A. C., Jardine, R., Feingold, L. M., & Eysenck, H. J.
(1986). Transmission of social attitudes. Proceedings of the National Academy of
Sciences of the United States of America, 83(12), 4364–4368.
QUASI-CAUSAL EFFECTS OF DIFFERENTIAL PARENTING ON RWA 45
McCourt, K., Bouchard, T. J., Lykken, D. T., Tellegen, A., & Keyes, M. (1999).
Authoritarianism revisited: Genetic and environmental influences examined in twins
reared apart and together. Personality and Individual Differences, 27, 985–1014.
doi:10.1016/S0191-8869(99)00048-3
Morsbach, S. K., & Prinz, R. J. (2006). Understanding and improving the validity of self-
report of parenting. Clinical Child and Family Psychology Review, 9(1), 1–21.
doi:10.1007/s10567-006-0001-5
Mullineaux, P. Y., Deater-Deckard, K., Petrill, S. A., & Thompson, L. A. (2009). Parenting
and child behaviour problems: a longitudinal analysis of non-shared environment.
Infant and Child Development, 18(2), 133–148. doi:10.1002/icd.593
Paulson, S. E., & Sputa, C. L. (1996). Patterns of parenting during adolescence: Perceptions
of adolescents and parents. Adolescence, 31(122), 369–382.
Pener-Tessler, R., Avinun, R., Uzefovsky, F., Edelman, S., Ebstein, R. P., & Knafo, A.
(2013). Boys' serotonin transporter genotype affects maternal behavior through self-
control: A case of evocative gene–environment correlation. Development and
Psychopathology, 25(1), 151–162. doi:10.1017/S095457941200096X
Peterson, B. E., Smirles, K. A., & Wentworth, P. A. (1997). Generativity and
authoritarianism: Implications for personality, political involvement, and parenting.
Journal of Personality and Social Psychology, 72(5), 1202–1216. doi:10.1037/0022-
3514.72.5.1202
Plomin, R., DeFries, J., & Fulker, D. (1988). Nature and nurture during infancy and early
childhood. New York, NY: Cambridge University Press.
Plomin, R., McClearn, G. E., Pedersen, N. L., Nesselroade, J. R., & Bergeman, C. S. (1989).
Genetic influence on adults’ ratings of their current family environment. Journal of
Marriage and the Family, 51(3), 791–803. doi:10.2307/352177
QUASI-CAUSAL EFFECTS OF DIFFERENTIAL PARENTING ON RWA 46
Putnick, D. L., Bornstein, M. H., Lansford, J. E., Chang, L., Deater-Deckard, K., Di Giunta,
L., ... & Pastorelli, C. (2012). Agreement in mother and father acceptance-rejection,
warmth, and hostility/rejection/neglect of children across nine countries. Cross-
Cultural Research, 46(3), 191–223. doi:10.1177/1069397112440931
Riemann, R., Kandler, C., & Bleidorn, W. (2012). Behavioral genetic analyses of parent twin
relationship quality. Personality and Individual Differences, 53, 398–404.
doi:10.1016/j.paid.2012.02.022
Riemann, R., & Wagner, R. (2000). Deutsche Version des Block Environmental
Questionnaire nach Hur & Bouchard [German Version of BEQ]. Unpublished
Questionnaire.
Rowe, D. C. (1981). Environmental and genetic influences on dimensions of perceived
parenting: A twin study. Developmental Psychology, 17, 203–208. doi:10.1037/0012-
1649.17.2.203
Rowe, D. C. (1983). Biometrical genetic models of self-reported delinquent behavior: A twin
study. Behavior Genetics, 13(5), 473–489. doi:10.1007/BF01065923
Rudy, D., & Grusec, J. E. (2001). Correlates of authoritarian parenting in individualist and
collectivist cultures and implications for understanding the transmission of values.
Journal of Cross-Cultural Psychology, 32(2), 202–212.
doi:10.1177/0022022101032002007
Russell, A., Aloa, V., Feder, T., Glover, A., Miller, H., & Palmer, G. (1998). Sex-based
differences in parenting styles in a sample with preschool children. Australian Journal
of Psychology, 50(2), 89–99. doi:10.1080/00049539808257539
Sales, S. M. (1973). Threat as a factor in authoritarianism: An analysis of archival data.
Journal of Personality and Social Psychology, 28, 44–57. doi:10.1037/h0035588
QUASI-CAUSAL EFFECTS OF DIFFERENTIAL PARENTING ON RWA 47
Scarr, S. (1985). Constructing psychology: Making facts and fables for our times. American
Psychologist, 40, 499–512. doi:10.1037/0003-066X.40.5.499
Scarr, S., & McCartney, K. (1983). How people make their own environments: A theory of
genotype→ environment effects. Child Development, 54(2), 424–435.
doi:10.2307/1129703
Scarr, S., & Weinberg, R. (1981). The transmission of authoritarianism in families: Genetic
resemblance in social-political attitudes. In Scarr, S. (Ed.), Race, Social Class, and
Individual Differences in I.Q. (pp. 299–347). Hillsdale, NJ: Erlbaum.
Schwartz, S. (1996). Value priorities and behavior: Applying a theory of integrated value
systems. In C. Seligman, J. Olson, & M. P. Zanna (Eds.), The psychology of values:
The Ontario symposium, (Vol. 8, pp. 1–24). Hillsdale, NJ: Erlbaum.
Sorkhabi, N. (2005). Applicability of Baumrind's parent typology to collective cultures:
Analysis of cultural explanations of parent socialization effects. International Journal
of Behavioral Development, 29(6), 552–563. doi:10.1080/01650250500172640
Steiger, J. H. (1990). Structural model evaluation and modification: An interval estimation
approach. Multivariate Behavioral Research, 25, 173–180.
doi:10.1207/s15327906mbr2502_4
Stieger, S., Kandler, C., Tran, U. S., Pietschnig, J., & Voracek, M. (2017). Genetic and
environmental sources of implicit and explicit self-esteem and affect: Results from a
genetically sensitive multi-group design. Behavior Genetics, 47, 175–192.
doi:10.1007/s10519-016-9829-8
Stößel, K., Kämpfe, N., & Riemann, R. (2006). The Jena twin registry and the Jena twin
study of social attitudes (JeTSSA). Twin Research and Human Genetics, 9(6), 783–
786. doi:10.1375/183242706779462615
QUASI-CAUSAL EFFECTS OF DIFFERENTIAL PARENTING ON RWA 48
Turkheimer, E. (2000). Three laws of behavior genetics and what they mean. Current
Directions in Psychological Science, 9, 160–164. doi:10.1111/1467-8721.00084
Turkheimer, E., & Harden, K. P. (2014). Behavior genetic research methods: Testing quasi-
causal hypotheses using multivariate twin data. In H. T. Reis, & C. M. Judd (Eds.),
Handbook of research methods in social and personality psychology (2nd ed., pp.
159–187). New York, NY: Cambridge University Press.
Turkheimer, E., & Waldron, M. (2000). Nonshared environment: A theoretical,
methodological, and quantitative review. Psychological Bulletin, 126(1), 78–108.
doi:10.1037/0033-2909.126.1.78
Weiss, L. H., & Schwarz, J. C. (1996). The relationship between parenting types and older
adolescents' personality, academic achievement, adjustment, and substance use. Child
Development, 67, 2101–2114. doi:10.2307/1131612
QUASI-CAUSAL EFFECTS OF DIFFERENTIAL PARENTING ON RWA 49
Table 1. JeTSSA Sample Information and Descriptive Statistics for RWA, Responsiveness, and Demandingness
Age
RWA
Responsiveness
Demandingness
Rater and sex
Nall
N
M
SD
N
M
Range
SD
N
M
Range
SD
N
M
Range
SD
Self-reports
Female twins
648
638
35.10
13.72
644
2.72
1.00–4.36
0.57
626
3.87
1.00–5.00
0.79
643
2.83
1.13–5.00
0.75
Male twins
227
226
32.04
13.15
226
2.78
1.00–4.45
0.63
220
3.94
2.00–5.00
0.61
224
2.90
1.38–4.75
0.61
Mothers
319
313
56.50
10.64
313
2.95
1.00–4.45
0.59
317
4.52
2.40–5.00
0.41
315
2.97
1.00–5.00
0.65
Fathers
268
249
58.37
10.35
261
3.08
1.09–4.73
0.60
268
4.38
2.00–5.00
0.49
265
2.95
1.00–5.00
0.65
Peer reports
Female twins
986a
557b
2.77
1.14–4.09
0.49
Male twins
336a
190b
2.86
1.18–4.18
0.58
Note. Means, ranges, and SDs are based on mean scores across all items of a scale.
aOverall number of peers rating female or male twins.
bOverall number of peer reports after averaging peer reports for twins for which two peer reports were available.
QUASI-CAUSAL EFFECTS OF DIFFERENTIAL PARENTING ON RWA 50
Table 2. Correlations between Family Members’ Assessments of Parenting and Their RWA
Responsiveness
Demandingness
RWA
Family dyad
n
r
95% CI
n
r
95% CI
n
r
95% CI
MZ Twin 1 – Twin 2
209
.746
[679, .801]
209
.580
[.482, .663]
225
.672
[.593, .738]
DZ Twin 1 – Twin 2
152
.531
[.406, .636]
152
.421
[.281, .544]
166
.414
[.279, .533]
Father – Twin 1
245
.391
[.279, .492]
245
.339
[.223, .445]
258
.331
[.218, .436]
Father – Twin 2
222
.239
[.111, .359]
222
.258
[.131, .377]
235
.258
[.134, .374]
Mother – Twin 1
253
.381
[.270, .482]
253
.359
[.247, .462]
280
.422
[.321, .514]
Mother – Twin 2
281
.328
[.219, .429]
281
.324
[.215, .425]
311
.401
[.303, .490]
Father – Mother
202
.348
[.221, .464]
202
.335
[.206, .452]
218
.512
[.407, .604]
Note. Parenting correlations are based on factor scores uncorrected for age differences; MZ: monozygotic; DZ: dizygotic. All correlations were
significant (p < .001).
QUASI-CAUSAL EFFECTS OF DIFFERENTIAL PARENTING ON RWA 51
Table 3. Results for the Best Fitting Model of the Association between Retrospectively Assessed Responsiveness as Predictor Variable and Self- and
Peer-rated RWA as Outcome Variable: Latent A Regression Model
Self-Report
Peer Report
Model paths
Estimate
SE
p
R²
Estimate
SE
p
R²
Responsiveness
aP:
Additive genetic effects
.547
.104
< .001
.311
.429
.253
.090
.191
cP:
Shared environmental effects
.556
.093
< .001
.321
.560
.107
< .001
.325
eP:
Nonshared environmental effects
.503
.024
< .001
.263
.504
.025
< .001
.263
bA:
ARWA → Responsiveness
.317
.092
< .001
.105
.462
.277
.096
.221
RWA
a:
Additive genetic effects
.858
.102
< .001
.713
.477
.230
.038
.215
c:
Shared environmental effects
.000
.359×107
> .999
.000
.548
.175
.002
.298
e:
Nonshared environmental effects
.546
.026
< .001
.297
.700
.036
< .001
.486
bC:
CResponsiveness → RWA
.000
.000
b:
Responsiveness → RWA
−.060
.074
.415
.003
−.024
.103
.815
.001
Note. Estimates reflect unstandardized path coefficients. Parameter bC was fixed to zero. Significant values are bold-faced.
QUASI-CAUSAL EFFECTS OF DIFFERENTIAL PARENTING ON RWA 52
Table 4. Results for the Best Fitting Model of the Association between Retrospectively Assessed Demandingness as Predictor Variable and Self-
and Peer-rated RWA as Outcome Variable: Phenotypic Regression Model
Self-Report
Peer Report
Model paths
Estimate
SE
p
R²
Estimate
SE
p
R²
Demandingness
aP:
Additive genetic effects
.545
.131
< .001
.301
.547
.131
< .001
.305
cP:
Shared environmental effects
.529
.123
< .001
.284
.525
.124
< .001
.280
eP:
Nonshared environmental effects
.639
.031
< .001
.415
.640
.031
< .001
.415
bA:
ARWA → Demandingness
.000
.000
RWA
a:
Additive genetic effects
.832
.102
< .001
.687
.476
.232
.041
.227
c:
Shared environmental effects
.000
.417×106
> .999
.000
.523
.188
.005
.274
e:
Nonshared environmental effects
.537
.025
< .001
.287
.704
.037
< .001
.496
bC:
CDemandingness → RWA
.000
.000
b:
Demandingness → RWA
.161
.035
< .001
.026
.069
.040
.088
.005
Note. Estimates reflect unstandardized path coefficients. Parameters bA and bC were fixed to zero. Significant values are bold-faced.
QUASI-CAUSAL EFFECTS OF DIFFERENTIAL PARENTING ON RWA 53
Table 5. Results of the Phenotypic Semilatent MTMM Analyses for Responsiveness and RWA
Not including parental RWA
Including parental RWA
Model path
Estimate
SE
p
Estimate
SE
p
b:
Responsiveness
→
Twin’s RWA
.189
.048
< .001
.271
.067
< .001
Horizontal and vertical effects of parent’s RWA
hm:
Mother’s RWA
→
Responsiveness
.078
.050
.118
hf:
Father’s RWA
→
Responsiveness
−.125
.049
.011
vm:
Mother’s RWA
→
Twin’s (latent) RWA
.344
.040
< .001
vf:
Father’s RWA
→
Twin’s (latent) RWA
.131
.043
.002
cova:
Mother’s RWA
↔
Father’s RWA
.529
.030
< .001
Effect of twin's RWA on twin’s responsiveness rating
atp:
Twin’s RWA
→
Twin’s responsiveness rating
.099
.071
.164
.095
.068
.167
Covariance between self-rating residuals
mt:
Twin’s RWA
↔
Twin’s responsiveness rating
.086
.060
.150
.113
.057
.048
mm:
Mother’s RWA
↔
Mother’s responsiveness rating
.074
.051
.144
mf:
Father’s RWA
↔
Father’s responsiveness rating
.130
.050
.009
Factor loadings of twins’ RWA self- and peer reports on latent twin's RWA
at:
Twin’s RWA
→
Self-rated RWA
1.000
1.000
ap:
Twin’s RWA
→
Peer-rated RWA
1.000
1.000
RWA measures’ residual variances
rat:
Residual variance of self-rated RWA
.659
.028
< .001
.652
.026
< .001
QUASI-CAUSAL EFFECTS OF DIFFERENTIAL PARENTING ON RWA 54
rap:
Residual variance of peer-rated RWA
.655
.028
< .001
.661
.026
< .001
ra:
Residual variance of twin’s (latent) RWA
.729
.028
< .001
.596
.029
< .001
Factor loadings of mothers’, fathers’, and twins’ responsiveness rating on latent responsiveness
pt:
Responsiveness
→
Twin’s rating
.589
.055
< .001
.944
.106
< .001
pm:
Responsiveness
→
Mother’s rating
.683
.057
< .001
1.000
pf:
Responsiveness
→
Father’s rating
.624
.057
< .001
1.000
Responsiveness measures’ residual variances
rpt:
Residual variance of twin's responsiveness rating
.792
.036
< .001
.776
.037
< .001
rpm:
Residual variance of mother's responsiveness rating
.756
.045
< .001
.778
.035
< .001
rpf:
Residual variance of father’s responsiveness rating
.811
.040
< .001
.812
.036
< .001
rp:
Residual variance of (latent) responsiveness
.640
.039
< .001
Note. Estimates are unstandardized path coefficients. Significant path coefficients are bold-faced. Parameters at and ap were fixed to 1; Parameters
pm and pf were fixed to 1 for the model considering parental RWA.
QUASI-CAUSAL EFFECTS OF DIFFERENTIAL PARENTING ON RWA 55
Table 6. Results of the Phenotypic Semilatent MTMM Analyses for Demandingness and RWA
Not including parental RWA
Including parental RWA
Model path
Estimate
SE
p
Estimate
SE
p
b:
Demandingness
→
Twin’s RWA
.323
.047
< .001
.194
.094
.039
Horizontal and vertical effects of parent’s RWA
hm:
Mother’s RWA
→
Demandingness
.183
.046
< .001
hf:
Father’s RWA
→
Demandingness
.189
.046
< .001
vm:
Mother’s RWA
→
Twin’s (latent) RWA
.327
.043
< .001
vf:
Father’s RWA
→
Twin’s (latent) RWA
.065
.046
.153
cova:
Mother’s RWA
↔
Father’s RWA
.531
.030
< .001
Effect of twin's latent RWA on twin’s demandingness rating
atp:
Twin’s RWA
→
Twin’s demandingness rating
−.206
.093
.026
−.169
.091
.063
Covariance between self-rating residuals
mt:
Twin’s RWA
↔
Twin’s demandingness rating
.119
.060
.046
.107
.059
.067
mm:
Mother’s RWA
↔
Mother’s demandingness rating
.118
.050
.018
mf:
Father’s RWA
↔
Father’s demandingness rating
.093
.050
.066
Factor loadings of twins’ RWA self- and peer reports on latent twin's RWA
at:
Twin’s RWA
→
Self-rated RWA
1.000
1.000
ap:
Twin’s RWA
→
Peer-rated RWA
1.000
1.000
RWA measures’ residual variances
rat:
Residual variance of self-rated RWA
.664
.027
< .001
.428
.034
< .001
QUASI-CAUSAL EFFECTS OF DIFFERENTIAL PARENTING ON RWA 56
rap:
Residual variance of peer-rated RWA
.650
.028
< .001
.436
.035
< .001
ra:
Residual variance of twin’s (latent) RWA
.679
.032
< .001
.371
.034
< .001
Factor loadings of mothers’, fathers’, and twins’ demandingness rating on latent demandingness
pt:
Demandingness
→
Twin’s rating
.636
.071
< .001
1.002
.136
< .001
pm:
Demandingness
→
Mother’s rating
.689
.057
< .001
1.000
pf:
Demandingness
→
Father’s rating
.595
.056
< .001
1.000
Demandingness measures’ residual variances
rpt:
Residual variance of twin's demandingness rating
.813
.040
< .001
.816
.037
< .001
rpm:
Residual variance of mother's demandingness rating
.745
.046
< .001
.770
.044
< .001
rpf:
Residual variance of father’s demandingness rating
.820
.038
< .001
.798
.035
< .001
rp:
Residual variance of (latent) demandingness
.524
.037
< .001
Note. Estimates are unstandardized path coefficients. Significant path coefficients are bold-faced. Parameters at and ap were fixed to 1;
Parameters pm and pf were fixed to 1 for the model considering parental RWA.
QUASI-CAUSAL EFFECTS OF DIFFERENTIAL PARENTING ON RWA 57
Figure 1. Schematic depiction of b = the unbiased effect of the environment on the
phenotypic trait and (A) genetic contributions to the phenotypic trait (a) and a genotype-
environment correlation (ba) between the measured environment, comprising latent
environment, systematic measurement error and unsystematic measurement error, and the
measured trait, comprising the latent phenotypic trait, systematic measurement error and
unsystematic measurement error; (B) method effects in the form of global response biases (m)
that contribute to systematic measurement errors of the measured environment and measured
trait, and specific response biases (atp) that contribute to systematic measurement errors of the
measured environment via the rater’s phenotypic trait; (C) vertical (v) and horizontal (h)
effects of the parental phenotypic trait on the latent environment and the phenotypic trait. All
pathways can be considered by analyses of twin family data including several rater
perspectives on the environment and the phenotypic trait via structural equation modeling.
QUASI-CAUSAL EFFECTS OF DIFFERENTIAL PARENTING ON RWA 58
QUASI-CAUSAL EFFECTS OF DIFFERENTIAL PARENTING ON RWA 59
Figure 2. Genetically informative model of the regression of RWA on parenting (i.e.,
responsiveness or demandingness). AP1/AP2/aP = additive genetic factors/effects on parenting;
A1/A2/a = additive genetic factors/effects on RWA; CP1/CP2/cP = common environmental
factors/effects on parenting; C1/C2/c = common environmental factors/effects on RWA;
EP1/EP2/eP = unique environmental factors/effects on parenting (including measurement
error); E1/E2/e = unique environmental factors/effects on RWA (including measurement
error); bA = effect on parenting due to correlated genetic factors between RWA and parenting;
bC = effect on RWA due to correlated shared environmental factors between RWA and
parenting; b = effect on RWA controlled for genetic (a × bA) and shared environmental
mediation (cP × bC); γ1 = 1.0 correlation between monozygotic twin siblings, 0.5 between
dizygotic twin siblings; γ2 = 1.0 correlation between monozygotic twin siblings, 0.625
between dizygotic twin siblings after adjustment for assortative mating.
QUASI-CAUSAL EFFECTS OF DIFFERENTIAL PARENTING ON RWA 60
Figure 3. Phenotypic Semilatent Multitrait-Multimethod Model. Parenting and the twin’s
RWA are plotted as latent variables, with loadings of maternal (pm), paternal (pf), and twin’s
(pt) assessments of parenting on latent parenting, with loadings of twins’ self- (at) and peer
ratings (ap) as well as a blended loading (atp) of twin’s parenting rating on latent twin’s RWA,
and with a regression b from twin’s RWA on parenting. All residual variables (RM, RF, RTP,
RTA, RPA, and RA) are uncorrelated, except within-rater residual components, namely twins’
assessments (mt). All variances of exogenous variables and at as well as ap are fixed to 1 for
model identification and in order to freely estimate all other paths coefficients. For more
details on parametrization and model specification see text.
QUASI-CAUSAL EFFECTS OF DIFFERENTIAL PARENTING ON RWA 61
Figures 4. Phenotypic Semilatent Multitrait-Multimethod Model including parental RWA.
Parenting and the twin’s RWA are plotted as latent variables as described in Figure 2. This
model also includes mother’s and father’s self-reports on their own RWA, which is allowed to
correlate between them (cova). All residual variables (RM, RF, RTP, RTA, RPA, RP, and RA) are
uncorrelated, except within-rater residual components, namely twin’s (mt), mother’s (mm), and
father’s (mf) assessments. All variances of exogenous variables and pm, pf, at as well as ap are
fixed to 1 for model identification and in order to freely estimate all other paths coefficients.
For more details on parametrization and model specification see text.
QUASI-CAUSAL EFFECTS OF DIFFERENTIAL PARENTING ON RWA 62
Figure 5. Results for the latent A model of the regression of (A) self-, respectively (B) peer-
rated RWA on responsiveness. Parameters indicate standardized path coefficients. Dashed
lines show nonsignificant paths (p < .05). For parameter descriptions, see Figure 2.
QUASI-CAUSAL EFFECTS OF DIFFERENTIAL PARENTING ON RWA 63
Figure 6. Results for the phenotypic model of the regression of (A) self-, respectively (B)
peer-rated RWA on demandingness. Parameters indicate standardized path coefficients.
Dashed lines show nonsignificant paths (p < .05). For parameter descriptions, see Figure 2.
QUASI-CAUSAL EFFECTS OF DIFFERENTIAL PARENTING ON RWA 64
Figures 7. Results of the phenotypic semilatent MTMM analysis for responsiveness (A)
without parental RWA and (B) considering parental RWA. Parameters are standardized path
coefficients. Dashed lines reflect nonsignificant estimates (p < .05). For parameter
descriptions, see Figures 3 and 4.
QUASI-CAUSAL EFFECTS OF DIFFERENTIAL PARENTING ON RWA 65
Figure 8. Results of the phenotypic semilatent MTMM analysis for demandingness (A)
without parental RWA and (B) considering parental RWA. Parameters are standardized path
coefficients. Dashed lines reflect nonsignificant estimates (p < .05). For parameter
descriptions, see Figures 3 and 4.
