Culture Moderates the Genetic and Environmental Etiologies of Parenting

Abstract

A cultural behavior genetic approach was introduced as a prospective means to describe psychological differences between cultures. We compared genetic and environmental influences on remembered parenting for samples of twins from Japan and Sweden. Data were collected from 720 pairs of young adult Japanese twins and 824 pairs of adult Swedish twins using the Parental Bonding Instrument. In both samples, a very similar phenotypic factor structure was developed for maternal and paternal parenting. However, the genetic and environmental contributions were different. Parenting in Japan showed more genetic influences, whereas parenting in Sweden showed more shared environmental influences. Moreover, covariation among the six dimensions of parenting (i.e., maternal and paternal Warmth, Protectiveness, and Authoritarianism) was due to genetic correlations in Japan and to shared environmental correlations in Sweden. These results are consistent with the cultural psychology argument that parenting practices are child centered in Japan but parent centered in the West.

Keywords

behavior genetics culture and cognition family children

The present study aims to introduce a new perspective on describing cultural differences based on a comparison of traits in terms of underlying genetic and environmental factors. We termed this new approach “cultural behavior genetics” and applied it to parenting by comparing genetic and environmental etiologies of parenting between Japanese and Western cultures.

Japanese parenting has received substantial attention from cultural psychologists because, although Japan’s economy and education are comparable to those in Western countries, it has some distinct characteristics. Psychological and sociological studies comparing Japanese parenting style with Western (particularly American) parenting styles, have identified contrasting features of parenting, for example, indirect versus direct (Azuma, Kashiwagi, & Hess, 1981), indulgent versus authoritative (Azuma, 1986), empathetic versus imperative (Ujiie, 1997), persuasive versus instructive (Conroy, Hess, Azuma, & Kashiwagi, 1980) and affect salient versus Information salient (Bornstein et al., 1992). Contrasts are also seen in sleeping arrangements. Co-sleeping, where parents sleep with their children in order to meet the child’s needs, is typical in Japan, while in the West, parents’ rights precede the child’s demands and sleeping in separate rooms is common (Shweder, 2003).

What brought about these cultural differences in parenting? Several defining studies in cultural psychology comparing the East and West, or Japan and the United States, have identified cultural differences in individual’s social practices (e.g., secondary control vs. primary control; Weisz, Rothbaum, & Blackburn, 1984; and collectivism vs. individualism; Triandis, Bontempo, Villareal, Asai, & Lucca, 1987). Among them, Markus and Kitayama (1991) argued that the East and West differ in their construals of self and others; interdependence is the norm in Asian cultures while independence of the self is encouraged in Western cultures. Consistent with this argument, Morling, Kitayama, and Miyamoto (2002) pointed out that in many Asian societies individuals are more inclined to adjust their behavior in accordance with surrounding people or communities. In contrast, there is a social expectation in many Western societies that behavior should be guided by the individual’s own internal attitudes and values.

Given the differences in parenting behavior and social practices, we would argue that Japanese parenting is “child centered” (i.e., parents adjust their parenting in accordance with the perceived desires and needs of the children) and Western parenting is “parent centered” (i.e., parents influence their children in accordance with their own beliefs and attitudes).

Such descriptions lead us to hypothesize that parenting in Japan will differ from child to child even within the same family, whereas parenting in Western cultures will show the greatest differences from family to family. The use of twin siblings reared in the same family may help clarify these differences. There are two different types of twins: identical (monozygotic [MZ]) twins and fraternal (dizygotic [DZ]) twins. MZ twin siblings share identical genes, whereas DZ twin siblings share approximately half of their segregating genes. A sizable body of behavioral genetic research has found evidence for genetic influences on almost every human behavior examined (Turkheimer, 2000), by demonstrating greater behavior similarities in MZ twin siblings than in DZ twin siblings. Of course, these findings do not mean that human behavior is determined by genetics because genetic effects cannot manifest themselves without environmental intervention. Nevertheless, many individual differences in behavior are due to differences in genes.

Based on behavioral genetics and cultural psychology theory, the following predictions with respect to parenting can be made: in Japan, parenting of MZ twin siblings will be more similar than that for DZ twin siblings because Japanese parenting is substantially affected by the child’s genetic makeup. In contrast, in Western cultures, parenting of twin siblings will not differ between MZ and DZ twins because Western parenting is essentially determined by parents’ attributes. These predictions can be tested using the twin method.

The twin method, a framework of behavioral genetic research, can decompose observed (phenotypic) variance (V _P ) into variances in genetic (V _A ), shared environmental (V _C ), and nonshared environmental (V _E ) components.

V _A reflects the variation in multiple genotypes with small and additive influence. V _C refers to variations in environmental characteristics that make family members alike and differ between families. V _E reflects the variation in environmental characteristics that make family members different even if they live together. V _E also includes measurement errors.

In terms of molecular genetics, parenting is unlikely to be a target phenotype for analysis. However, in terms of behavioral genetics, any outcome variables accompanying individual differences, like any human phenotypic traits, can be exposed to genetic analysis. Thus, according to the aforementioned variance components of ACE, the estimated effects (expressed by the relative size of variance components) of ACE on parenting can be interpreted as follows. First, the presence of genetic effects in children suggests that parenting behavior is elicited by children’s genetic makeup and is responsive to variation in the children’s genotype. In addition, if the measurement relies on the recollections of children, perceptions of parenting received might be filtered through the child’s characteristics such as personality. If personality is influenced by genetic factors, then the recollected parenting could be genetically influenced. In the behavioral genetic framework, any trait that is invoked by an individual’s genetically influenced characteristics or influenced by the environment that is associated with the person’s genetically influenced characteristics can be observed as genetically influenced. Second, shared environmental effects on parenting suggest that parenting of children in the same family is the same regardless of the children’s characteristics and may indicate that parenting is a parent-centered, coherent, disciplined practice. Third, nonshared environmental effects suggest that children brought up in the same family receive different parenting not attributable to the children’s genetics, thereby making parenting a differentiated or random practice. In accordance with our hypothesis, we predict that parenting in Japan would be explained more by the child’s genetic effects, whereas parenting in Western cultures would be explained more by shared environmental effects.

To measure parenting, we employed the Parental Bonding Instrument (PBI; Parker, Tupling, & Brown, 1979). The PBI is a widely used, refined psychological measure in which the children retrospectively rate their parents’ general behavior or attitudes toward them. Parker postulated “care” and “overprotection” as two dimensions that account for various parenting styles. However, a number of studies using the PBI in different cultures have confirmed that a three-factor structure is more robust, suggesting that Parker’s “overprotection” be split into two factors: psychological control and behavioral control (Cubis, Lewin, & Dawes, 1989; Murphy, Brewin, & Slika, 1997; Sato et al., 1999).

Several behavioral genetic studies have examined genetic and environmental etiologies of the three dimensions of the PBI in Western families. A study using American female twins suggested that children’s genetic factors were important for the “warmth” (care) dimension while “protectiveness” (psychological control) and “authoritarianism” (behavioral control) dimensions were largely affected by the family background of the parent (Kendler, 1996). Another study using Swedish female twins demonstrated that genetic effects were significant only for “warmth” and found substantial shared environmental effects across dimensions (Lichtenstein et al., 2003). In Australian female twins, “coldness” (inversely scaled to “warmth”) was genetically influenced, whereas “overprotection” (psychological control) and “autonomy” (inversely scaled to behavioral control) were influenced by both genetic and shared environmental factors (Gillespie, Zhu, Neale, Heath, & Martin, 2003). A Canadian study revealed that all three parenting dimensions were explained by pure environmental models, with substantial shared environmental but no significant genetic influences (Jang, Dick, Wolf, Livesley, & Paris, 2005). Although some differences in degree have been observed, these studies suggest that, in Western families, parental warmth is influenced more by children’s genetics and less by shared environment than the two control dimensions. On the other hand, all these studies found that parental control, psychological and behavioral, is largely affected by shared environment.

The salient shared environmental effects detected in the PBI support our hypothesis from the side of Western parenting. However, to date, no studies of genetic and environmental influences on parenting in Japanese families have been reported. Thus, one of the purposes of the present study is to examine the relative magnitude of genetic and shared environmental effects in the etiology of the three PBI dimensions and compare them between Japanese and Western samples.

Furthermore, even in studies of Western parenting, how the genetic or environmental influences of one dimension overlaps with another, that is, the genetic and environmental factor structure across the maternal and paternal dimensions, has not been well identified. The existing literature examining the phenotypic level of the PBI has clarified that the warmth and control dimensions are negatively correlated (Parker, 1983) and that the same three-factor structure can be produced for both maternal and paternal items for both Japanese and Western samples (e.g., Cubis et al., 1989; Sato et al., 1999). Given that, we predicted that the phenotypic correlations of the PBI dimensions within maternal items and within paternal items were similar between both cultures. However, whether the correlation derives from the underlying common genetic factor or common environmental factor and how the maternal and paternal dimensions are intercorrelated are unknown.

If Japanese parenting is a child-centered practice, maternal and paternal parenting styles should be closely tied through correlated genetic influences of the child. Western parenting, as a unitary practice of the family, should display large overlap in the underlying shared environmental effects across both parents and their parenting styles. Accordingly, we predict that in Japan there will be a genetic influence (from the child’s genes) that explains correlations not only within maternal or paternal parenting dimensions but also across parents. In other words, genetically influenced characteristics of the child will influence generally the covariation among the dimensions between parents. In contrast, in Western cultures, we predict that shared environmental influences will be significant while genetic influences will not. Thus, the second purpose of the present study is to examine the genetic and environmental correlations among the PBI dimensions in Japanese and Western samples.

We conducted a cross-cultural study using a behavioral genetic approach to examine whether cultural differences in parenting appeared as a difference in genetic and environmental etiologies. We chose a Swedish sample as the Western population for comparison. Japan and Sweden are assumed to be equally ethnically homogeneous. Therefore, it is reasonable to presuppose a similar level of genetic variation owing to ethnic diversity in both populations. The relatively close GDP per capita and Gini index between Japan and Sweden imply at least a similar level of economic standards of living and distributions. Parenting in Japan and Sweden was measured using the PBI. Genetic and environmental etiologies of parenting were directly compared between the two populations in terms of the relative contribution of genetic and environmental effects of each maternal and paternal dimension, and intercorrelations across the dimensions of the PBI.

Method

Japanese Participants

The Japanese participants included 1,497 twins among the 2,073 twins registered with the Keio Twin Project (Shikishima, Ando, Ono, Toda, & Yoshimura, 2006) who were recruited from a population-based twin residential list for the Tokyo area. All were native Japanese adolescents or adults ranging from 14 to 32 years old (M = 20.3, SD = 3.9).

The zygosity of each same-sex twin pair was initially diagnosed by a 3-item questionnaire based on physical resemblance (Ooki, Yamada, Asaka, & Hayakawa, 1990). Among them, gene polymorphisms were examined for 285 pairs. It was confirmed that 93.3% of these DNA-based diagnoses were in agreement with initial questionnaire-based diagnoses. As a result, the effective number of twin pairs was broken down as follows: 334 female monozygotic (MZf), 158 male monozygotic (MZm), 96 female dizygotic (DZf), 48 male dizygotic (DZm), and 84 opposite-sex (DZo) pairs. DZo pairs were excluded from genetic analyses. Female pairs held at 68% in genetic analyses.

Swedish Participants

The Swedish participants were 1,733 twins among the 909 twin pairs participating in the Twin and Offspring Study in Sweden project (Neiderhiser & Lichtenstein, 2008) who were recruited through the population-based Swedish Twin Registry (Lichtenstein et al., 2002). They were required to have an adolescent child who was the same sex and within 4 years of age as his or her co-twin’s child, and to be with their partner for 5 years or more. All were consistent with the population of Sweden and were principally Caucasian. Age of the participants ranged from 32 to 60 years (M = 44.9, SD = 4.9).

The zygosity of each same-sex twin pair was initially assessed by a questionnaire asking the extent to which twins viewed themselves as “similar as two berries” (Nichols & Bilbro, 1966). Zygosity assignment was confirmed for most of the sample via genotyping (Neiderhiser & Lichtenstein, 2008). The effective number of twin pairs according to zygosity was 245 MZf, 124 MZm, 267 DZf, and 188 DZm pairs. Female pairs held at 62% in all analyses.

In addition to a higher voluntary participation rate to the project for females than males commonly observed for both Japan and Sweden, the lower DZ twinning rate in Japan (Imaizumi, 2003) produced an unbalanced distribution of the category of twins in the Japanese sample.

Measurement

Parenting was assessed using the Japanese or Swedish version of the PBI (Kitamura & Suzuki, 1993; Parker et al., 1979). Offspring were asked to rate the behavior and attitudes of each parent during their childhood up to the age of 16. Participants were required to choose the most appropriate answer from a 4-point Likert-type scale for 25 items concerning maternal parenting and 25 items concerning paternal parenting. We conducted principal axis factor analyses with promax rotation on each of the 25 Japanese and Swedish items. The score for each subscale was calculated by summing the total score of the items convergent to each factor.

Statistical Analysis

Univariate genetic analysis

We conducted univariate genetic analyses on the three dimensions of maternal and paternal parenting. With the analysis, the phenotypic variance (V _P ) of each dimension can be partitioned into three variance components: additive genetic (V _A ), shared environmental (V _C ), and nonshared environmental (V _E ), which can be formulated as follows.

V_{p} = V_{A} + V_{C} + V_{E} .

The difference in genetic resemblance between MZ and DZ twin siblings (sharing 100% and 50% of segregating genes), and the equivalence in environmental resemblance for the two types of twins can yield the following two equations concerning the observed covariances for MZ (Cov_MZ) and DZ (Cov_DZ) twin siblings.

\begin{aligned} {C o v}_{M Z} = V_{A} + V_{C}, \\ {C o v}_{D Z} = 0. {5 V}_{A} + V_{C} . \end{aligned}

The estimates of the three parameters, V _A , V _C , and V _E , can be computed by solving the above three equations.

In the present study, first, we fit our data to two distinct models: a heterogeneity model (the size of V _A , V _C , and V _E differed between the Japanese and Swedish populations) and a homogeneity model (contributions were equivalent across the two populations; Figure 1). A better model-fitting index for the heterogeneity model would imply that the etiologies of parenting were culturally specific. No deterioration for the homogeneity model would suggest that the etiologies of parenting were cross-culturally invariant.

Second, we conducted model-fitting analyses for each dimension to explore the best-fitting model among the ACE model (full model), the AE model (submodel in which the effect of shared environment is set to zero), and the CE model (submodel in which the effect of genetics is set to zero).

Third, we presented graphic representations of V _A , V _C , and V _E for each dimension for Japan and Sweden. Relative proportions of ACE contributions were also shown.

Multivariate genetic analysis

We conducted multivariate genetic analyses on the three dimensions of maternal and paternal parenting to explore the extent to which genetic and environmental factors mediate the phenotypic covariance between the six dimensions, using Cholesky decompositions (Neale & Maes, 2002). With this technique, we described the extent to which the genetic effects on one dimension overlap the genetic effects on another dimension, yielding a genetic correlation coefficient (r_G; Figure 2). The shared environmental correlation (r_C) and nonshared environmental correlation (r_E) were likewise obtained (for an explanation of r_G, r_C, and r_E, see Neale & Maes, 2002). We conducted the analyses separately for Japan and Sweden.

We highlighted the estimates whose 95% confidence intervals did not include zero. When the confidence intervals included zero, the results would be equivocal because they could mean small size of correlation or large size of standard error as a result of small size of variance components.

Results

Phenotypic Factor Structure for 25 Items

A principal axis factor analysis on the 25 items of parenting showed exactly the same convergence pattern of the three-factor solution for both maternal and paternal items, except that the second and third components appeared in the opposite order between the two countries (Table 1). We labeled the three factors “Warmth,” “Authoritarianism,” and “Protectiveness,” as in Lichtenstein et al. (2003). The total score for “Authoritarianism” was reversed to allow a higher score to indicate higher authoritarianism. High internal consistency of each subscale represented by Cronbach’s α guaranteed high reliabilities of the measurement (.72 to .91 for Japan and .82 to .91 for Sweden; Table 1).

Table 1.

Phenotypic Factor Structure of 25 Items of PBI in Japan and Sweden.

Item	Maternal						Paternal
	Japan			Sweden			Japan			Sweden
	1	2	3	1	3	2	1	2	3	1	3	2
Frequently smiled at me	.81			.83			.83		.23	.84
Enjoyed talking things over with me	.76			.68			.81			.74
Was affectionate to me	.76			.87			.74			.84
Spoke to me with a warm and friendly voice	.74			.74			.80			.74
Seemed emotionally cold to me	−.69			−.74			−.72			−.68		.23
Could make me feel better when I was upset	.68			.46			.67			.54
Did not talk with me very much	−.65			−.77	.23		−.72			−.78	.24
Did not praise me	−.64			−.68			−.72			−.72
Appeared to understand my problems and worries	.61			.74			.62			.73
Made me feel I wasn't wanted	−.50			−.45		.21	−.43		.35	−.35		.35
Did not help me as much as I needed	−.46			−.44			−.58			−.42
Did not seem to understand what I needed or wanted	−.46		.22	−.59		.28	−.49		.27	−.63
Gave me as much freedom as I wanted		.93			.78			.87			.78
Let me go out as often as I wanted		.81			.76			.78			.79
Let me do those things I liked doing		.75		.29	.39			.81		.32	.49
Let me dress in any way I pleased		.43			.55			.34			.54
Let me decide things for myself		.40	−.24		.60			.47			.65
Liked me to make my own decisions	.31	.35		.28	.45		.31	.38		.31	.50
Felt I could not look after myself unless she/he was around		.21	.74			.70			.70			.71
Tried to make me dependent on her/him			.64			.72			.65			.71
Was overprotective of me	.29		.57			.61	.38	−.22	.32	.21		.64
Tended to baby me			.54			.62			.43	−.23		.56
Tried to control everything I did		−.25	.51		−.23	.62		−.29	.48		−.21	.62
Invaded my privacy			.42			.66			.44			.71
Did not want me to grow up			.38			.64			.36			.61
Variance explained by rotated factors	6.64	5.35	4.54	7.59	5.28	5.83	7.08	5.52	4.20	7.61	5.11	5.68
Cronbach’s α	.89	.82	.77	.90	.82	.85	.91	.82	.72	.91	.85	.85

Note. Principal axis factor analysis with promax rotation was conducted. Loadings more than ±.20 are shown.

Basic Statistics Comparison for the Six Dimensions

The means, ranges, and variances for each of the three maternal and paternal parenting dimensions for Japan and Sweden are shown in Table 2. The levels for Warmth and Authoritarianism were similar in Japan and Sweden, whereas large differences were found for Protectiveness levels, with the effect size greater than 0.80. The variance for maternal Warmth was equivalent between the two countries, but the variances for the other dimensions were significantly larger for Japan.

Table 2.

Basic Statistics of the PBI Dimensions for Japan and Sweden.

Dimension		Mean			Range		Variance
Dimension		Japan	Sweden	ES	Japan	Sweden	Japan	Sweden	F Value	p
Maternal	Warmth	39.28	38.06	0.19	13−48	12−48	40.59	40.87	0.92	NS
	Authoritarianism	11.55	12.36	0.23	6−24	6−24	14.43	10.08	84.11	<.001
	Protectiveness	14.06	10.66	0.82	7−27	7−28	16.61	12.24	25.01	<.001
Paternal	Warmth	35.33	36.37	0.15	12−48	13−48	58.58	44.69	27.61	<.001
	Authoritarianism	11.18	12.19	0.29	6−24	6−24	13.83	10.41	82.73	<.001
	Protectiveness	13.24	10.11	0.84	7−27	7−26	13.01	10.31	9.25	<.01

Note. ES indicates the effect size in terms of Cohen’s d.

Univariate Genetic Analysis for the Six Dimensions

As shown in Table 3, the heterogeneity model provided a better fit than the homogeneity model for all dimensions, indicating that the genetic and environmental structures of the parenting dimensions differed between the two cultures. The Japanese and Swedish ACE contributions were therefore estimated separately.

Table 3.

Fitting of Heterogeneity and Homogeneity Models.

Dimension		Model	−2LL	df	AIC	Δχ²	Δdf	p
Maternal	Warmth	Heterogeneity	19142.73	2983	13176.73	−	−	−
	Warmth	Homogeneity	19153.44	2986	13181.44	10.71	3	<.05
	Authoritarianism	Heterogeneity	15525.08	2983	9559.08	−	−	−
	Authoritarianism	Homogeneity	15570.60	2986	9598.65	45.57	3	<.001
	Protectiveness	Heterogeneity	16073.19	2982	10109.19	−	−	−
	Protectiveness	Homogeneity	16103.16	2985	10133.20	29.97	3	<.001
Paternal	Warmth	Heterogeneity	19275.29	2931	13413.29	−	−	−
	Warmth	Homogeneity	19300.51	2934	13432.51	25.22	3	<.001
	Authoritarianism	Heterogeneity	15304.36	2940	9424.36	−	−	−
	Authoritarianism	Homogeneity	15332.01	2943	9446.01	27.65	3	<.001
	Protectiveness	Heterogeneity	15251.10	2940	9371.10	−	−	−
	Protectiveness	Homogeneity	15266.97	2943	9380.97	15.88	3	<.01

Note. −2LL = log likelihood fit statistic; df = degrees of freedom; AIC = Akaike information criterion. A smaller value for AIC indicates a better fit to data.

For the Japanese sample, dropping the genetic effect (from the ACE model to the CE model) significantly worsened the model fit for all dimensions except maternal Protectiveness, while dropping the shared environmental effect (from the ACE model to the AE model) did not (left half of Table 4). For four of the six dimensions, the AE model showed the best fit in terms of both χ² test and Akaike Information Criterion (AIC) value. In contrast, for the Swedish sample, dropping the shared environmental effect significantly worsened the model fit for all dimensions except paternal Protectiveness. The AE model was not the best model in terms of both model-fitting indices for any of the other dimensions for Sweden (right half of Table 4).

Table 4.

Fitting of Full Models and Submodels for Japan and Sweden.

Dimension			Japan						Sweden
Dimension		Model	−2LL	df	AIC	Δχ²	Δdf	p	−2LL	df	AIC	Δχ²	Δdf	p
Maternal	Warmth	ACE	8405.8	1304	5797.8	−	−	−	10736.9	1680	7376.9	−	−	−
		AE	8407.4	1305	5797.4	1.6	1	NS	10753.1	1681	7391.1	16.2	1	<.001
		CE	8412.1	1305	5802.1	6.3	1	<.05	10748.5	1681	7386.5	11.6	1	<.001
	Authoritarianism	ACE	7035.3	1304	4427.3	−	−	−	8855.1	1679	5497.1	−	−	−
		AE	7038.1	1305	4428.1	2.8	1	NS	8861.5	1680	5501.5	6.3	1	<.05
		CE	7039.4	1305	4429.4	4.2	1	<.05	8860.9	1680	5500.9	5.7	1	<.05
	Protectiveness	ACE	7218.1	1304	4610.1	−	−	−	8489.8	1679	5131.8	−	−	−
		AE	7226.2	1305	4616.2	8.1	1	<.01	8508.0	1680	5148.0	18.2	1	<.001
		CE	7219.0	1305	4609.0	0.9	1	NS	8491.2	1680	5131.2	1.4	1	NS
Paternal	Warmth	ACE	8552.8	1277	5998.8	−	−	−	10722.5	1654	7414.5	−	−	−
		AE	8554.1	1278	5998.1	1.3	1	NS	10749.1	1655	7439.1	26.6	1	<.001
		CE	8569.4	1278	6013.4	16.6	1	<.001	10727.4	1655	7417.4	4.9	1	<.05
	Authoritarianism	ACE	6820.4	1278	4264.4	−	−	−	8483.9	1662	5159.9	−	−	−
		AE	6820.9	1279	4262.9	0.5	1	NS	8507.6	1663	5181.6	23.7	1	<.001
		CE	6828.1	1279	4270.1	7.7	1	<.01	8483.9	1663	5157.9	0.0	1	NS
	Protectiveness	ACE	6747.4	1277	4193.4	−	−	−	8503.7	1663	5177.7	−	−	−
		AE	6749.9	1278	4193.9	2.5	1	NS	8506.9	1664	5178.9	3.2	1	NS
		CE	6751.8	1278	4195.8	4.4	1	<.05	8510.3	1664	5182.3	6.6	1	<.05

Note. −2LL = log likelihood fit statistic; df = degrees of freedom; AIC = Akaike information criterion. Δχ² was tested compared to the ACE model. The respective best statistics in terms of χ² test and AIC values are highlighted in boldface.

Variance components of ACE for maternal and paternal Warmth dimensions are shown in Figure 3. The relative proportion for genetics (A) in the Japanese sample (34% for maternal and 48% for paternal) was higher than that in the Swedish sample, especially for paternal Warmth (29% for maternal and 19% for paternal). By contrast, the shared environmental (C) contribution in the Swedish sample (33% for maternal and 42% for paternal) was much higher than that for the Japanese sample (18% for maternal and 17% for paternal).

A similar pattern was observed for Authoritarianism (Figure 3). The relative proportion for A in Japan was substantial (27% for maternal and 41% for paternal), whereas that in Sweden was negligible or even absent for paternal Authoritarianism (12% for maternal and 0% for paternal), and a substantial C component was present instead (36% for maternal and 42% for paternal).

The pattern for Protectiveness differed from patterns for the other two dimensions (Figure 3) by having a higher C contribution for maternal Protectiveness in the Japanese sample (39% for Japan and 22% for Sweden).

Phenotypic Correlations Among the Six Dimensions

Phenotypic correlation (r_P) matrices for the three maternal and three paternal dimensions were very similar between the two countries (Table 5). Negative correlations were obtained between Warmth and Authoritarianism (Japan: –.50 for maternal and –.51 for paternal; Sweden: –.53 and –.51) and between Warmth and Protectiveness (Japan: –.36 and –.28; Sweden: –.46 and –.46). Positive correlations were obtained between Authoritarianism and Protectiveness (Japan: .58 and .57; Sweden: .55 and .51). Those patterns also were observed in the cross-correlations of maternal and paternal dimensions.

Table 5.

Phenotypic, Genetic, Shared Environmental, and Nonshared Environmental Correlations Among Six Dimensions for Japan and Sweden.

Dimension		1		2		3		4		5		6
Phenotypic correlations (r_P)
1.	Maternal Warmth			−.53	[−.56, −.49]	−.46	[−.49, −.42]	.56	[.53, .60]	−.37	[−.41, −.33]	−.38	[−.42, −.33]
2.	Maternal Authoritarianism	−.50	[−.54, −.46]			.55	[.51, .58]	−.32	[−.37, −.28]	.69	[.67, .72]	.39	[.35, .43]
3.	Maternal Protectiveness	−.36	[−.40, −.31]	.58	[.54, .61]			−.24	[−.29, −.20]	.31	[.27, .36]	.62	[.59, .65]
4.	Paternal Warmth	.46	[.42, .50]	−.27	[−.32, −.22]	−.23	[−.28, −.22]			−.51	[−.54, −.47]	−.46	[−.49, −.42]
5.	Paternal Authoritarianism	−.34	[−.38, −.29]	.59	[.55, .62]	.34	[.29, .38]	−.51	[−.55, −.47]			.51	[.48, .55]
6.	Paternal Protectiveness	−.22	[−.27, −.17]	.35	[.30, .39]	.47	[.43, .51]	−.28	[−.32, −.23]	.57	[.53, .60]
Genetic correlations (r_G)
1.	Maternal Warmth			−.34	[−.36, .55]	−.28	[−.63, .18]	.65	[.24, ,90]	−.43	[−1.00, 1.00]	−.34	[−.63, .23]
2.	Maternal Authoritarianism	−.77	[−.96, −43]			.83	[.51, 1.00]	.12	[−.57, .90]	.61	[−.82, 1.00]	.52	[−.48, .97]
3.	Maternal Protectiveness	−.68	[−1.00, .11]	.76	[−.14, 1.00]			−.26	[−.71, .41]	.91	[−.25, 1.00]	.74	[.21, .97]
4.	Paternal Warmth	.96	[.74, 1.00]	−.60	[−.90, −.18]	−.66	[.−1.00, .27]			−.59	[−1.00, .40]	−.43	[−.84, .18]
5.	Paternal Authoritarianism	−.89	[−1.00, −.54]	.92	[.67, 1.00]	.61	[−.31, 1.00]	−.72	[−.98, −.34]			.87	[−.42, 1.00]
6.	Paternal Protectiveness	−.55	[−.90, −.07]	.95	[.62, 1.00]	.77	[−.14, 1.00]	−.36	[−.45, .14]	.77	[.41, .98]
Shared environmental correlations (r_C)
1.	Maternal Warmth			−.82	[−.99, −.61]	−.77	[−1.00, −.40]	.50	[.22, .65]	−.55	[−.82, −.33]	−.58	[−.94, −.22]
2.	Maternal Authoritarianism	−.09	[−.61, .65]			.73	[.45, .94]	−.46	[−.59, −.22]	.81	[.70, .91]	.58	[.30, .97]
3.	Maternal Protectiveness	−.20	[−.68, .60]	.79	[.62, 1.00]			−.19	[−.57, .23]	.36	[−.07, .68]	.49	[−.36, .78]
4.	Paternal Warmth	−.44	[−1.00, .29]	.35	[−.17, .96]	.13	[−.40, .88]			−.69	[−.88, −.51]	−.69	[−1.00, −.35]
5.	Paternal Authoritarianism	.80	[−.14, 1.00]	.19	[−.69, .60]	.32	[−.50, .83]	−.20	[−.92, .65]			.79	[.50, 1.00]
6.	Paternal Protectiveness	.26	[−.40, .99]	−.22	[−.96, .32]	.33	[−.58, .66]	−.21	[−.84, .61]	.66	[−.29, .95]
Nonshared environmental correlations (r_E)
1.	Maternal Warmth			−.41	[−.48, −.32]	−.40	[−.48, −.32]	.58	[.51, .65]	−.26	[−.35, −.17]	−.31	[−.39, −.21]
2.	Maternal Authoritarianism	−.47	[−.53, −.40]			.39	[.31, .45]	−.37	[−.44, −.28]	.64	[.58, .68]	.29	[.20, .37]
3.	Maternal Protectiveness	−.33	[−.40, −.25]	.42	[.35, .48]			−.29	[−.38, −.20]	.20	[.12, .29]	.61	[.54, .67]
4.	Paternal Warmth	.37	[.29, .44]	−.26	[−.34, −.18]	−.23	[−.31, −.15]			−.37	[−.44, −.29]	−.37	[−.45, −.28]
5.	Paternal Authoritarianism	−.27	[−.35, −.20]	.50	[.44, .56]	.26	[.18, .34]	−.45	[−.51, −.38]			.36	[.28, .43]
6.	Paternal Protectiveness	−.19	[−.27, −.10]	.27	[.19, .34]	.49	[.42, .55]	−.24	[−.32, −.15]	.39	[.32, .46]

Note. 95% confidence intervals are shown in brackets. Coefficients whose confidence interval did not include zero are highlighted in boldface. For each type of correlations, the values above the diagonal indicate Swedish coefficients and those below the diagonal indicate Japanese coefficients.

Multivariate Genetic Analysis Among the Six Dimensions

Genetic correlations (r_G) among the six dimensions for Japan and Sweden are shown in Table 5. In Japan, most dimensions were highly correlated with the exception of maternal Protectiveness, whose genetic variance component was not large enough. In particular, large overlaps were observed between maternal and paternal dimensions (.96 for maternal and paternal Warmth; .95 for maternal Authoritarianism and paternal Protectiveness; and .92 for maternal and paternal Authoritarianism), mediated by the same genetic factor in the children. In Sweden, most of the dimensions were not genetically correlated with substantial components.

Shared environmental correlations (r_C) showed a pattern opposite to genetic correlations (r_G; Table 5). In Japan, they were low or components were too small, whereas in Sweden they were higher with substantive components. A significant correlation was seen only for maternal control dimensions in Japan. However, in Sweden, all the correlations except for maternal Protectiveness and paternal dimensions were largely correlated by shared environment, and maternal and paternal dimensions firmly overlapped with the same shared environment. A high shared environmental correlation (r_C) between maternal and paternal Authoritarianism (.81) also was noteworthy. Purportedly, in Sweden, discipline of children is consistent across parents, mediated by the common family environment.

With respect to nonshared environmental correlations (r_E), the pattern was similar between Japan and Sweden and also was close to the phenotypic correlations (Table 5). The moderate correlations (–.47 and –.50 for Japan and –.41 and –.64 for Sweden) imply that the degree of overlap across the dimensions across parents was similar for both Japan and Sweden.

Discussion

In our study, in contrast to the similar phenotypic structure, the genetic and environmental etiologies of Japanese and Swedish parenting were different. With the exception of maternal Protectiveness, each parenting style in Japan was influenced more by the child’s genetics and was largely affected by the same genetic factor. In Sweden, on the other hand, the role of shared environmental factors replaced that of genetic factors. The Swedish results obtained in this study are consistent with most previous findings from the PBI and other instruments using Western samples in indicating that the warmth/care dimension is somewhat genetic and control dimensions are not genetic and are affected by substantial shared environment (Rowe, 1981, 1983). However, contrasting with the existing literature, in Japan, control dimensions were not only driven by genetics but also closely tied to the other dimensions by the same genetic factor.

Inconsistent results in maternal Protectiveness in Japan are puzzling. Given that the level of psychological control of children of nonworking mothers tends to be higher than that of working mothers (Tachibanaki & Urakawa, 2007), it might be attributable to greater variation in mothers’ lives in Japan; the employment rate of mothers with a child younger than 6 years is around 40% in Japan compared with more than 80% in Sweden (Organisation for Economic Cooperation and Development [OECD], 2002). Future research is needed to identify the specific aspects of the shared environment that may explain the findings of the shared environment latent factor.

The salient genetic influence on Japanese parenting can be explained in two ways: (1) parents are responding to a child’s genetics or (2) the child’s genetic makeup is influencing their recollections of parenting. In the first explanation, the parents’ behaviors are attuned to the children’s behavior, resulting in child-centered, adjustment-oriented (Morling, Kitayama, & Miyamoto, 2002), permissive (Azuma, 1986; Yamada, 2004) practices, often described as a typical Japanese parenting style. If the second explanation is correct, then the degree of genetic influences is expected to become higher as the individual gets older. However, in this study, the older Swedish sample did not exhibit larger genetic effects than the younger Japanese sample; therefore, the former interpretation is more plausible.

Shared environmental effects on Swedish parenting suggest that children brought up in the same family should receive the same parenting, irrespective of the children’s characteristics. This is consistent with the description of Western parenting often referred to as a parent-centered, influence-oriented (Morling et al., 2002), uniform, and coherent practice in comparison to the Japanese counterpart. The high intercorrelations among maternal and paternal practices mediated by the same family environment showed that parenting is purposeful family work generated from family background including beliefs, values, and religion.

The cultural differences observed in the relative magnitude of genetic and environmental influences in the present study are in marked contrast with the previous literature on other traits, suggesting that, unlike parenting, the genetic and environmental effects on intrinsic phenotypes might be universal. For example, published literatures using a Japanese or Swedish adult twin sample indicate that estimates for relative genetic and environmental contributions for temperamental traits of harm avoidance (A: .41, C: .00, and E: .59 for Japan vs. A: .42, C: .00, and E: .58 for Sweden; Ando et al., 2004; Yuh et al., 2008), depression (A: .31, C: .00, and E: .69 for Japan vs. A: .24, C: .04, and E: .72 for Sweden; Shikishima et al., 2006; Yuh et al., 2008), and general intelligence (A: .83 and E: .17 for Japan vs. A: .82 and E: .18 for Sweden; Plomin, Pedersen, Lichtenstein, & McClearn, 1994; Shikishima et al., 2009) are consistently and highly similar between Japan and Sweden. Although cultures can moderate manifestations of genetic and environmental effects of human traits, the degree of moderation can vary according to the trait. It may be that the more extended from the individual the phenotype (trait) is, the more likely it is to be moderated by the culture. As the phenotype becomes more specific to the context, it requires more interplay with other people or even society. We found differences in the genetic and environmental etiologies of parenting, presumably because parenting reflects cultural differences in social dynamics and the way people interact with each other as represented by secondary control versus primary control (Weisz et al., 1984), collectivism versus individualism (Triandis et al., 1987), or interdependence versus independence (Markus & Kitayama, 1991).

If this is the case, we would observe similar cultural differences in etiologies of other interpersonal relations such as friendship. In Western societies, individuals may express their friendship irrespective of the recipients’ (friends’) attributes, while in Eastern societies, individuals may adjust their behavior with each friend. The same can be predicted about the behavior or attitudes of teachers, mentors, relatives, and so on. Future studies will clarify the point.

As a limitation of the present study, given that the PBI measures children’s recollections of parenting, there is a possibility that the findings could reflect differences in the way adolescents and adults in each culture remember the way they were parented rather than a reflection on differences in parenting practices per se. Another limitation is that we did not directly compare the size of A, C, and E variance components between the two countries because our interest lies more in examining differences in the overall structure of ACE contributions than the absolute size of each component. In addition, the ages differed for the Japanese and Swedish samples. Although high retest coefficients on PBI scores over decades are reported (Wilhelm, Niven, Parker, & Hadzi-Pavlvic, 2005), the possibility that the age difference was associated with the differences found for genetic and environmental influences in the two samples cannot be completely ruled out.

Despite these limitations, the present study demonstrated that cultural behavior genetics can collaborate well with cultural psychology. Cultural behavior genetic approaches can identify cultural differences that are not otherwise detected. Future cultural behavior genetic research will offer unique opportunities to test hypotheses generated by the literature in cultural psychology, as was done in this study, and find cultural differences theorized by cultural psychologists.

Footnotes

Acknowledgments

We thank Drs. P. Lichtenstein, D. Reiss, N. Pedersen, E. L. Spotts, and J. Ganiban for kindly agreeing to share their data with us. We also thank Miss K. Davey for kindly editing the draft manuscript.

Declaration of Conflicting Interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The author(s) disclosed receipt of the following financial support for the research of this article: The Keio Twin Project in Japan was supported by Grant-in-Aid for Scientific Research (S) from the Japanese Ministry of Education, Culture, Sports, Science and Technology (No. 21223002) to Juko Ando. The Twin and Offspring Study in Sweden was supported by R01MH54610 from the National Institute of Mental Health to Jenae M. Neiderhiser.

References

Ando

Suzuki

Yamagata

Kijima

Maekawa

Ono

Jang

K. L.

(2004). Genetic and environmental structure of Cloninger’s temperament and character dimensions. Journal of Personality Disorders, 18, 379–393.

Azuma

(1986). Why study child development in Japan? In Stevenson

Azuma

Hakuta

(Eds.), Child development and education in Japan (pp. 3–12). New York, NY: Freeman.

Azuma

Kashiwagi

Hess

(1981). The influence of attitude and behavior upon the child’s intellectual development. Tokyo: University of Tokyo Press.

Bornstein

M. H.

Tamis-LeMonda

C. S.

Tal

Ludeman

Toda

Rahn

C. W.

… Vardi

(1992). Maternal responsiveness to infants in three societies: The United States, France, and Japan. Child Development, 63, 808–821.

Conroy

Hess

R. D.

Azuma

Kashiwagi

(1980). Maternal strategies for regulating children’s behavior: Japanese and American families. Journal of Cross-Cultural Psychology, 11, 153–172.

Cubis

Lewin

Dawes

(1989). Australian adolescents’ perceptions of their parents. Australian and New Zealand Journal of Psychiatry, 23, 35–47.

Gillespie

N. A.

Zhu

Neale

M. C.

Heath

A. C.

Martin

N. G.

(2003). Direction of causation modeling between cross-sectional measures of parenting and psychological distress in female twins. Behavior Genetics, 33, 383–396.

Imaizumi

(2003). A comparative study of zygotic twinning and triplet rates in eight countries, 1972–1999. Journal of Biosocial Science, 35, 287–302.

Jang

K. L.

Dick

D. M.

Wolf

Livesley

W. J.

Paris

(2005). Psychosocial adversity and emotional instability: An application of gene-environment interaction models. European Journal of Personality, 19, 359–372.

10.

Kendler

K. S.

(1996). Parenting: A genetic epidemiologic perspective. American Journal of Psychiatry, 153, 11–20.

11.

Kitamura

Suzuki

(1993). A validation study of parental bonding instrument in Japanese population. Japanese Journal of Psychiatry and Neurology, 47, 29–36.

12.

Lichtenstein

De Faire

Floderus

Svartengren

Svedberg

… Pedersen

N. L.

(2002). The Swedish Twin Registry: A unique resource for clinical, epidemiological and genetic studies. Journal of Internal Medicine, 252, 184–205.

13.

Lichtenstein

Ganiban

Neiderhiser

J. M.

Pedersen

N. L.

Hansson

Cederblad

Reiss

(2003). Remembered parental bonding in adult twins: Genetic and environmental influences. Behavior Genetics, 33, 397–408.

14.

Markus

Kitayama

(1991). Culture and the self: Implications for cognition, emotion, and motivation. Psychological Review, 98, 224–253.

15.

Morling

Kitayama

Miyamoto

(2002). Cultural practices emphasize influence in the United States and adjustment in Japan. Personality and Psychology Bulletin, 28, 311–323.

16.

Murphy

Brewin

C. R.

Silka

(1997). The assessment of parenting using the Parental Bonding Instrument: Two or three factors? Psychological Medicine, 27, 333–342.

17.

Neale

M. C.

Maes

H. H. M.

(2002). Methodology for genetic studies of twins and families. Dordrecht, The Netherlands: Kluwer Academic Publishers.

18.

Neiderhiser

J. M.

Lichtenstein

(2008). The twin and offspring study in Sweden: Advancing our understanding of genotype–environment interplay by studying twins and their families. Acta Psychologica Sinica, 40, 1116–1123.

19.

Nichols

R. C.

Bilbro

W. C.

(1966). The diagnosis of twin zygosity. Acta Genetica et Statistica Medica, 16, 265–275.

20.

Ooki

Yamada

Asaka

Hayakawa

(1990). Zygosity diagnosis of twins by questionnaire. Acta Geneticae Medicae et Gemellogiae, 39, 190–115.

21.

Organisation for Economic Cooperation and Development. (2002). OECD Employment Outlook, June 2001, 134–135.

22.

Parker

(1983). Parental overprotection: A risk factor in psychosocial development. New York, NY: Grune &Stratton.

23.

Parker

Tupling

Brown

L. B.

(1979). A parental Bonding Instrument. British Journal of Medical Psychology, 52, 1–10.

24.

Plomin

Pedersen

N. L.

Lichtenstein

McClearn

G. E.

(1994). Variability and stability in cognitive abilities are largely genetic later in life. Behavior Genetics, 24, 207–215.

25.

Rowe

D. C.

(1981). Environmental and genetic influences on dimensions of perceived parenting: A twin study. Developmental Psychology, 17, 203–208.

26.

Rowe

D. C.

(1983). A biometrical analysis of perceptions of family environment: A study of twin and singleton sibling relationships. Child Development, 54, 416–423.

27.

Sato

Narita

Hirano

Kusunoki

Sakado

Uehara

(1999). Confirmatory factor analysis of the Parental Bonding Instrument in a Japanese population. Psychological Medicine, 29, 127–133.

28.

Shikishima

Ando

Ono

Toda

Yoshimura

(2006). Registry of adolescent and young adult twins in the Tokyo area. Twin Research and Human Genetics, 9, 811–816.

29.

Shikishima

Hiraishi

Yamagata

Sugimoto

Takemura

Ozaki

… Ando

(2009). Is g an entity? A Japanese twin study using syllogisms and intelligence tests. Intelligence, 37, 256–267.

30.

Shweder

(2003). Why do men barbecue? Recipes for cultural psychology. Cambridge, MA: Harvard University Press.

31.

Tachibanaki

Urakawa

(2007). The effect of changes in family structures on intergenerational transfer of inequality in Japan. Inequalization Trend and Policy Options in Japan (Chap. 2, pp.4–36). ESRI International Collaboration Projects 2006.

32.

Triandis

H. C.

Bontempo

Villareal

Asai

Lucca

(1988). Individualism and collectivism: Cross-cultural perspectives on self-ingroup relationships. Journal of Personality and Social Psychology, 54, 323–338.

33.

Turkheimer

(2000). Three laws of behavior genetics and what they mean. Current Directions in Psychological Science, 9, 160–164.

34.

Ujiie

(1997). How do Japanese mothers treat children’s negativism? Journal of Applied Developmental Psychology, 18, 467–483.

35.

Weisz

J. R.

Rothbaum

F. M.

Blackburn

T. C.

(1984). Standing out and standing in: The psychology of control in America and Japan. American Psychologist, 39, 955–969.

36.

Wilhelm

Niven

Parker

Hadzi-Pavlovic

(2005). The stability of the parental bonding instrument over a 20-year period. Psychological Medicine, 35, 387–393.

37.

Yamada

(2004). Japanese mothers' views of young children's areas of personal discretion. Child Development, 75, 164–179.

38.

Yuh

Neiderhiser

J. M.

Spotts

E. L.

Pedersen

N. L.

Lichtenstein

Hansson

… Reiss

(2008). The role of temperament and social support in depressive symptoms: A twin study of mid-aged women. Journal of Affective Disorders, 106, 99–105.