Do type of childcare and age of entry predict behavior problems during early childhood? Results from a large Norwegian longitudinal study

Abstract

Associations between type and age of entry into Norwegian universally-accessible childcare and children’s behavior problems at age 3 years were examined in this study. Data from 73,068 children in the large population-based, prospective Norwegian Mother and Child Cohort Study (MoBa) were used, and included information about childcare arrangements, behavior problems, and a variety of covariates. The results provided little support for childcare being related to children’s behavior problems at age 3 years when controlling for covariates. In fact, previous research has indicated that children may benefit from Norway’s childcare in other areas, such as language development. Results are discussed in relation to differences in countries' early childhood policies as a possible factor explaining discrepancies across studies.

Keywords

behavior problems kindergarten MoBa population-based universal childcare

The influence of non-parental childcare on children’s behavior problems has been a contentious debate: What types of care are best for children, and at what age should children enter childcare (Bradley & Vandell, 2007)? To date, the debate has relied heavily on studies from the United States (US), and two reviews mainly based on US studies concluded that experience of childcare is related to increased levels of behavior problems (Bradley & Vandell, 2007; Jacob, 2009). However, this conclusion found only limited support in a review that also included studies from other countries (Swedish National Institute of Public Health, 2009). Since regulations governing the delivery of childcare vary considerably among countries (UNICEF Innocenti Research Center, 2008), the effect of early childcare may vary across contexts (Love et al., 2003). To understand the impact of childcare, the sociopolitical context must be considered (Lamb & Ahnert, 2006). There is a lack of information about the effects of regimes that offer heavily publicly-subsidized and regulated childcare (Currie, 2001).

The early childcare context in Norway provides unique possibilities to obtain knowledge about this issue. This study examines how children’s attendance at universally-accessible childcare in Norway at ages 1, 1.5, and 3 years is related to behavior problems at age 3 years in a large, epidemiological longitudinal study.

Evidence from previous childcare studies

The United States (US)-based National Institute of Child Health and Human Development Study of Early Child Care and Youth Development (NICHD SECCYD) began in 1991 and has dominated research on the influence of childcare on child development for the last decade. In general, the study shows that children in center care show higher levels of externalizing problems than children in other types of childcare up to the age of 5 (NICHD Early Child Care Research Network, 2006). The study also shows that early entry (before six months of age) into a childcare center is related to higher levels of externalizing problems at the age of 4–5 than later entry (NICHD Early Child Care Research Network, 2003). Several studies, mainly from the US, supported the findings from SECCYD. For instance, a study following approximately 14,000 American children up to the age of 4 found the level of externalizing problems to be related to the age of entry into childcare (Loeb, Bridges, Bassok, Fuller, & Rumberger, 2007). Another American study following 171 children to the age of 8 reported higher levels of externalizing problems in children who spent part of their first year of life in childcare (Youngblade, 2003), while a Canadian study of approximately 1,300 children (Cote, Borge, Geoffroy, & Rutter, 2008) reported higher levels of both externalizing and internalizing problems for children who were in childcare before 1 year of age.

However, several studies outside the United States reported findings that conflict with the SECCYD results. A Danish study of 6,000 children (Gupta & Simonsen, 2009) found no overall difference in behavior problems between children attending childcare and those looked after at home at the age of 3.5. Likewise, in an Australian study of 150 infants (Harrison & Ungerer, 2002; Love et al., 2003) and in a study of approximately 1,000 children from the United Kingdom (Barnes, Leach, Malmberg, Stein, & Sylva, 2010) no association between the use of early childcare and later behavior problems was found. Finally, two Canadian studies reported positive associations with children’s behavior. First, a study of 1,250 children reported lower levels of hyperactivity among children in childcare than among children looked after at home (Nomaguchi, 2006). Second, a study of 4,521 children found that children enrolled in government-regulated childcare showed less physical aggression than children in unregulated care (Romano, Kohen, & Findlay, 2010).

Positive relationships have been found particularly for children from disadvantaged families. A study of 3,290 children in a pooled dataset from the United States and Canada found that poor children with center-care experience were rated as having less externalizing behavior problems than children without center-care experience (Crosby, Dowsett, Gennetian, & Huston, 2010). Furthermore, a study of 3,400 Canadian children found that children from poor families who were looked after at home had a higher risk of developing externalizing problems than children from poor families attending childcare centers (Borge, Rutter, Cote, & Tremblay, 2004). Finally, experimental studies of children from disadvantaged families have showed that children in high-quality childcare have lower levels of behavior problems prior to entering school (Love et al., 2003).

Further, boys and girls may react to different kinds of care differently (Crockenberg, 2003). For instance, findings by Howes and Olenick (1986) suggested that boys may respond more negatively than girls to low-quality childcare. However, gender has often been ignored in the childcare literature (Bornstein, Hahn, Gist, & Haynes, 2006). Moreover, there is a need for more information about the effect of childcare on internalizing problems, since the childcare literature has mainly focused on externalizing problems and, to a lesser degree, on internalizing problems.

A general explanation for the divergent findings in the literature may be the large variation in focus, methodology, and sampling strategies across studies (Bornstein et al., 2006). Some studies were based on cross-sectional data, whereas others were longitudinal; some studies examined the effect of non-maternal care in general, whereas others focused on center care or high- versus low-quality care. Moreover, most studies used samples that are not representative of the entire population, thereby limiting the external validity of these studies.

Furthermore, there are major differences in the contexts in which the studies were conducted, and the regulations governing the delivery of childcare vary considerably between countries. For example, the United States, where the most childcare research has been conducted, met only 3 of the 10 UNICEF benchmarks for early childcare policy; these benchmarks represent minimum standards for number of staff, minimum level of training for all staff, availability, and price (UNICEF Innocenti Research Center, 2008). By comparison, Norway met 8 out of the 10 benchmarks. These variations will have an impact on how childcare relates to children’s development and make it difficult to generalize across contexts (Love et al., 2003; UNICEF Innocenti Research Center, 2008).

In recent years there has been increased research focus on the use of prospective longitudinal studies of large samples and greater attention to potential selection effects. Up to now, most childcare researchers have relied on a set of child and family characteristics (such as parental age, education, income, native language, and maternal mental health) as covariates to cope with possible selection bias. However, the use of multiple analytical approaches to handling possible selection bias has been required in the childcare literature (McCartney et al., 2010).

Childcare in Norway

In Norway, one of the parents receives fully-paid leave for 10 months or 80% paid leave for one year (Ministry of Children, Equality, and Social Inclusion, 2011). From one year of age, all children have the right to center care or family day care. The centers are heavily subsidized, with fees on a sliding scale depending on family income. Quality standards are regulated by law (Ministry of Education, 2010), and following the national curriculum is compulsory (Ministry of Education, 2006a). Center care has a maximum adult-to-child ratio of 3:10 for those younger than age 3, and one adult is required to be a trained childcare teacher. Family day care available in private homes has a maximum group size of 10 children, adult-to-child ratios are lower than 1:5 for children younger than 3 years of age, and the staff receives weekly supervision from a teacher responsible for pedagogical planning (Ministry of Education, 2006b). These elements are widely accepted as indicators of process quality (Organisation for Economic and Co-operative Development's [OECD] Network on Early Childhood Education and Care, 2010).

Consequently, childcare in Norway has a relatively high, homogeneous quality across centers and is universally available (Winsvold & Guldbrandsen, 2009), meaning that children from families with all socioeconomic backgrounds are placed at the same centers. In contrast, socioeconomic status tends to be confounded with quality of care in the United States (Love et al., 2003). The early childcare policy in Norway makes it an ideal society for examining links between type and age of entry into childcare and children’s behavior problems, with a lower probability for selection bias. In fact, previous research into Norway’s childcare system has indicated positive relations in other developmental areas. A Norwegian study by Lekhal, Zachrisson, Wang, Schjølberg, and von Soest (2011) showed a positive relation between children attending childcare and their language development.

This study

In this study, a large-scale longitudinal population study is used to examine how universally-available childcare is related to behavior problems in Norwegian children. More specifically, the study investigates whether being in center care or family day care during the first, second, or third year of life is related to externalizing and internalizing problems at age 3 years. The second aim is to investigate whether the association between type of and age of entry into childcare and externalizing and internalizing problems is moderated by the child’s gender, mother’s and father’s level of education, and family income. The robustness of the results will be examined across two different methodological approaches to control for possible selection bias.

Methods

Participants

Data from the Norwegian Mother and Child Cohort Study (MoBa; see Magnus et al., 2006, and Norwegian Mother and Child Cohort Study, n.d.) was used in this study. MoBa recruited participants from all over Norway from 1999–2008, and 38.5% of the women invited consented to participate. The cohort now includes 108,000 children. Information on the health and lifestyles of the women and the development of the children was collected by questionnaire at the 17th, 22nd, and 30th weeks of gestation (T1 to T3) and when the child was 6, 18, and 36 months of age (T4 to T6). Additionally, the data were linked to register data from the Medical Birth Registries of Norway.

In total, 75,271 children have so far been eligible for the 36-month assessment (T6), whereas the remaining children of the study did not have the opportunity to participate at T6 because they had not yet turned 3 years of age when the data for this article was analyzed. Of the eligible children at T6, 44,636 completed the questionnaire at this time, thereby providing a respond rate of 59.3% of all families who consented to participate at T1. Furthermore, 2,203 of the 75,271 children who were eligible for the 36-month assessment were excluded because of child deafness, severe hearing loss, cerebral palsy, neurodevelopment syndromes, Down’s Syndrome, cleft lip and palate, or serious birth defects. Thus the final sample comprised 73,068 eligible children of whom 42,669 had completed the questionnaire at T6.

Potential self-selection bias in the MoBa was examined by Nilsen and colleagues (2009), showing that young mothers (under the age of 25), mothers living alone, mothers with more than two previous births, and women who smoked were under-represented in the sample compared to all women giving birth in Norway. In addition, stillbirths, neonatal deaths, and pregestational diabetes were less frequent, whereas mothers using multivitamins and folic acid were over-represented. Further, infants in the MoBa had a higher birth weight, longer crown-to-heel length, larger mean head circumference, higher Apgar score at five minutes, and were born after longer gestation than infants of the total Norwegian population. Importantly, Nilsen and colleagues (2009) demonstrated that, despite prevalent differences between the sample and the population, associations between risk exposure and child development outcomes were not significantly different when MoBa participants were compared with the population of Norwegian mothers.

Measures

Childcare arrangements

Mothers reported the current type of childcare arrangements at T5 and T6. At T5 mothers additionally reported the type of care that the child had experienced at age 1 year. The alternatives were: home care by mother or father; family day care/nanny care; and center care (for more detailed information about the nature of the different childcare arrangements, see the earlier paragraph about childcare in Norway). The questionnaire did not distinguish between family day care and nanny care (from now on referred to as family day care). Two dummy variables were constructed at 1, 1.5, and 3 years, where family day care and center care, respectively, were contrasted to the rest of the sample.

Behavior problems at age 3 years

Externalizing and internalizing problems at 3 years of age were assessed at T6 with 16 items from the Child Behavior Checklist (CBCL; Achenbach, 1992). Externalizing problems were assessed by 11 items from two subscales from the CBCL externalizing domain; that is, attention problems (four items) and aggression problems (seven items). The items used were as follows: “Doesn’t seem to feel guilty after misbehaving,” “Punishment doesn’t change his/her behavior,” “Defiant,” “Hits others,” “Doesn’t get along with other children,” “Can’t sit still, restless or hyperactive,” “Can’t concentrate, can’t pay attention for long,” “Can’t stand waiting; wants everything now,” “Demands must be met immediately,” “Poorly coordinated or clumsy,” and “Quickly shifts from one activity to another.” Internalizing problems were assessed by five items from two subscales from the CBCL internalizing domain; that is, emotionally reactive (two items) and anxiety (three items). The items used were as follows: “Disturbed by any change in routine,” “Sudden changes in moods or feelings,” “Clings to adults or too dependent,” “Gets too upset when separated from parents,” and “Too fearful or anxious.” The use of selected items was necessary due to the questionnaire space restrictions that are typical for large interdisciplinary epidemiological studies. The items were consensually selected by clinical and developmental psychologists. Responses range from 1 = “not true” to 2 = “somewhat or sometimes true” to 3 = “very true or often true.” A sum score was calculated, ranging from 11 to 33 for the externalizing scale and five to 15 for the internalizing scale. The internal consistency of the internalizing scale was Cronbach’s α = .52 and Cronbach’s α = .74 for the externalizing scale.

Confounding variables

At T1 mothers reported their own and the father’s age, education, and immigrant background. Family income was assessed as well. At T4 the mothers rated psychological distress (symptoms of anxiety and depression) by using a five-item short version of the Hopkins Symptom Checklist (SCL-5; Hesbacher, Rickels, Morris, Newman, & Rosenfeld, 1980). This short version has good construct validity (Strand, Dalgard, Tambs, & Rognerud, 2003). In this study, the internal consistency was good, with Cronbach’s α = .80. Marital status (single mothers) was assessed at T4, T5, and T6. Information concerning smoking during pregnancy was assessed from the Medical Birth Registries.

Information on the child’s physical health (Apgar scores 1 and 5 minutes after birth), gender, gestational age, and birth weight was retrieved from the Medical Birth Registries. Both gestational age and birth weight were dichotomized, with cut-offs at week 38 for gestational age and 2,500 grams for birth weight.

When the child was 6 months old (T4), mothers rated the children’s behavior on seven items derived from the fussy/difficult subscale of the Infant Characteristic Questionnaire (ICQ; Bates, Freeland, & Lounsbury, 1979). Responses ranged from 1 = “very easy” to 7 = “very difficult.” A sum score was calculated ranging from 7 to 49. Internal consistency was Cronbach’s α = .67.

Children’s behavior at 18 months (T5) was measured by 14 items from the CBCL (Achenbach, 1992). Concerning the externalizing scale, 9 of the 11 selected items that were included at 36 months (T6) were also included at 18 months. The two items that were not included in the 18 months questionnaire because of space limitations were “Poorly coordinated or clumsy” and “Demands must be met immediately.” Concerning the internalizing scale, the item “Sudden changes in moods or feelings” at T6 was replaced with the item “Afraid to try new things” at T5, whereas the four other items were used in both data collection waves. Internal consistency of the internalizing and externalizing scale was Cronbach’s α = .46 and Cronbach’s α = .62, respectively.

Analyses

Following recent recommendations in the childcare literature (McCartney et al., 2010), two different methodological approaches were used to answer the research questions and to assess the robustness of findings across these methods. In the first approach, the relation between type and age of entry into childcare and behavior problems was examined using covariate-adjusted linear regression analyses, including children’s behavior at 6 and 18 months as covariates before entering childcare.

As the second and more conservative methodological approach, propensity score matching was used (Rosenbaum & Rubin, 1984). Propensity score matching is well suited when making causal inferences from observational data in which a subgroup of the observations participated in or experienced some kind of “treatment” (e.g., childcare) without random assignment (Hill & Reiter, 2006). These techniques have been shown to be able to eliminate up to 90% of the covariate bias in non-experimental data (Leon & Hedeker, 2007). The propensity score analysis proceeded in three steps. In the first step, the propensity score or the probability of being in family day care or center care was estimated for each child. This was done using logistic regression with childcare arrangement at ages 1, 1.5, and 3 years of age as the dependent variables and the covariates as predictors. Next, all children were stratified into five equally-large groups based on the distribution of their estimated propensity scores. Group 1 consisted of children least likely to receive center care or family day care, and group 5 included those children most likely to receive center care or family day care. Finally, in the third step the effects of center care and family day care were estimated for these groups of participants. After testing main effects, potential moderators (child gender, mother’s and father’s level of education, and family income) of the association between type and age of entry into childcare and behavior problems were tested. All interaction effects were examined separately by entering interaction terms into separate multiple regression analyses.

Missing data due to attrition was common, with 40.7% missing at age 3 years. The percentage of missing items due to item non-response was less common, with the highest missing percentage of 6.5% for externalizing behavior at age 18 months. To deal with missing data (both due to attrition and item non-response), all analyses were conducted using multiple imputation, which is an adequate method for handling moderate to large amounts of missing data (Schafer & Graham, 2002; Widaman, 2006). Specifically, 20 datasets were estimated based on all covariates in Table 1. All datasets combined observed and imputed values. Moreover, all analyses were repeated using only data from the 42,669 children responding at T6 (i.e., multiple imputation was only used to handle item non-response). Results were compared to analyses using multiple imputation on all data. These comparisons were conducted to ensure that results obtained by using the whole sample were not influenced by any form of methodological artifact related to high percentage of missing replacement by using multiple imputation.

Table 1.

Descriptive statistics (N = 73,068)

Variable	M (SD)/%
Child care at age 1 year
At-home care	86.3%
Family day care	5.8%
Center care	7.9%
Child care at age 1.5 years
At-home care	32.9%
Family day care	20.0%
Center care	47.1%
Child care at age 3 years
At-home care	6.9%
Family day care	3.9%
Center care	89.2%
CBCL 36 months
Externalizing behavior 11 items	16.49 (3.20)
Internalizing behavior 5 items	6.37 (1.44)
Family factors
Mother’s age	30.08 (4.59)
Father’s age	32.70 (5.41)
Mother’s education	14.57 (2.49)
Father’s education	14.01 (2.74)
Mother single at 6 months	2.2%
Mother single at 18 months	2.5%
Mother single at 36 months	3.1%
Non-Norwegian family background	10.3%
Maternal mental health	1.23 (0.38)
Smoking during first trimester	6.7%
Smoking during last trimester	13.0%
Family income, in Norwegian kroner (NOK)
0–275,000	6.9%
275,001–450,000	18.3%
450,001–900,000	54.6%
> 900,000	20.1%
Child factors
Gestational age < 38 weeks	10.8%
Birth weight < 2,500 g	3.9%
Boys	50.9%
Apgar 1 minute after birth	8.67 (1.14)
Apgar 5 minutes after birth	9.41 (0.77)
ICQ 6 months 7 items	16.27 (5.60)
CBCL 18 months
Externalizing behavior 8 items	11.74 (2.43)
Internalizing behavior 5 items	6.10 (1.34)

Results

Descriptive analyses

Table 1 present means and standard deviations. The percentages of children in different childcare arrangements show that a clear majority of children were cared for at home at age 1 year. However, this pattern changed at 1.5 years of age, when about 60% of all children attended family day care or center care. At age 3 years, the predominant form of care was center care. There were an approximately equal number of boys (50.9%) and girls (49.1%) in the sample. Boys were rated as having more externalizing problems than girls (t = 9.70, p < .001); internalizing problems were more prevalent among girls compared to boys (t = −3.55, p < .001). About every tenth child had non-Norwegian family background, and there were few single mothers (ranging from 2.2% to 3.1% from T4–T6). Only 6.9% of the sample was in the lowest income group, and mothers' and fathers' average education level was relatively high, with an average of two years of higher education.

Covariate-adjusted OLS regression

Multiple regression analyses were conducted to examine how type and age of entry into childcare at age 1, 1.5, and 3 years was related to behavior problems when the children were 3 years old. All analyses were done in two steps. In the first step, the analyses were performed without incorporating relevant confounders. In the second step, relevant confounders were incorporated, including children’s behavior at 6 and 18 months, before entering childcare.

Before adjusting for any covariates (Table 2, unadjusted model) three significant results were obtained. Children in family day care at age 1 year showed higher levels of behavior problems both in the externalizing and internalizing domain compared to those cared for at home. Moreover, children in center care at age 1.5 years were reported to have less internalizing problems at age 3 than children who were in home care. All other relationships between childcare and behavior problems were not significant before incorporating covariates. For children in family day care at age 1 year, there was an increase of 3% of a standard deviation in externalizing problems and an increase of 7% of a standard deviation in internalizing problems at age 3. For children in center care at age 1.5 years the results indicated a decrease of 5% of a standard deviation in internalizing problems compared to those cared for at home.

Table 2.

Prediction of behavior problems at age 3 years (CBCL)

	Externalizing				Internalizing
	Unadjusted model		Adjusted model		Unadjusted model		Adjusted model
Predictor	B(SE B)	STDY	B(SE B)	STDY	B(SE B)	STDY	B(SE B)	STDY
Age 1 year
At-home care (reference group)
Family day care	.124 (.052)	.039*	.080 (.053)	.025	.095 (.033)	.070**	.070 (.032)	.052**
Center care	.056 (.050)	.018	.047 (.050)	.015	−.001(.029)	−.002	−.013(.030)	−.011
Age 1.5 years
At-home care (reference group)
Family day care	.039 (.037)	.012	.063 (.038)	.020	−.017(.022)	−.009	−.015(.021)	−.008
Center care	−.028 (.031)	−.009	.036 (.033)	.011	−.067(.017)	−.047***	−.032(.018)	−.023
Age 3 years
At-home care (reference group)
Family day care	−.043 (.052)	−.013	−.021 (.063)	−.007	.006(.035)	−.001	.002(.034)	−.005
Center care	−.035 (.072)	−.011	−.011 (.056)	−.004	−.039(.028)	−.029	.006(.030)	.002

Note: B = Unstandardized coefficient; STDY= Estimate with standardized Y variable (change in SD on behavior problems (CBCL) from at-home care to center care and from at-home care to family day care). Adjusted model: Control for mother’s and father’s age and education, maternal mental health, immigrant background, family income, Apgar score at 1 and 5 minutes after birth, gestational age, birth weight, gender, smoking during pregnancy, marital status at 6, 18, and 36 months, and behavior (ICQ) at 6 months and CBCL at 18 months.

*p < .05

**p < .01

***p < .001.

In the second regression analysis, 19 covariates were included in addition to childcare arrangements as independent variables. The results in Table 2 (adjusted model) show that the significant association between family day care at age 1 year and later externalizing problems was no longer significant after covariates were included in the model. The relation between center care at age 1.5 and lower internalizing problems diminished as well into non-significance when covariates were included in the regression analyses. However, the relationship between family day care at age 1 and internalizing problems at age 3 years remained significant, even though the size of the effect was somewhat reduced. Moreover, all other relations between childcare arrangements and children’s behavior remained non-significant.

Controlling for selection bias using propensity scores matching

As a second analytical approach to handling potential selection effects, the total effects of childcare on children’s behavior problems were estimated by using propensity scores to group children in five homogeneous subclasses. In other words, it was examined whether type and age of entry into childcare predicted level of behavior problems at age 3 years when comparing children with a similar probability of being in family day care or center care. When using this more conservative analytical approach, no significant relations were found between childcare arrangements and children’s behavior problems. This was true independent of whether the children started at age 1, 1.5, or 3 years and across all five propensity categories (all p > .05).

Potential moderators and alternative analyses

After testing main effects, potential moderators were tested by entering interaction terms into separate multiple regression analyses. More specifically, interactions between type of and age of entry into childcare and child’s gender, mother’s and father’s level of education, and family income were examined. The results showed that none of the tested interaction effects between the potential moderators, type of and age of entry into childcare, and behavior problems were significant (all p > .05). Finally, all analyses presented so far were repeated by using only data from children who responded at age 3 (N = 42,669). In these analyses, none of the three significant relationships that were obtained when using data from all children reached significance, probably because of the somewhat lower power of these analyses. Moreover, all other results remained non-significant and were similar compared to results where data from all children who participated at T1 were used, with only small differences in the sizes of the estimated coefficients.

Discussion

This study is among the first to examine the impact of attending universal, high-quality childcare on children’s behavior problems. The results showed that attending family day care at age 1 was related to a greater frequency of behavior problems, both in the externalizing and internalizing domain at age 3. Moreover, attending center care at age 1.5 was related to a lower frequency of internalizing problems compared to children cared for at home. However, the sizes of the associations were small, and diminished into insignificance when adjusting the analyses conservatively for potential selection bias (e.g., covariates or propensity scores matching). Moreover, this study did not find support for the notion that the effect of childcare on behavior problems varies dependent on the child’s gender and indicators for parents' socioeconomic status.

Our results are not in line with research finding more negative behavior outcomes for children attending childcare (Loeb et al., 2007; NICHD Early Child Care Research Network, 2003, 2006; Youngblade, 2003). Differences in countries' early childhood policies may be an important factor explaining this discrepancy. For instance, in the US and NICHD SECCYD where most previous childcare research has been done, positive care giving was rated as ‘not characteristic’ in 60% of the childcare settings, and only 10% of the settings were rated as excellent. Thirty four percent of the parents used multiple childcare arrangements and 72% of children in non-maternal care entered prior to four months of age (Bradley & Vandell, 2007).

In Norway most children do not enter childcare before one year of age. In addition, Norway provides universal access to center care and regulates its quality (in terms of both teacher education and staff-to-child ratios, etc.) via national standards, with the result that the centers are of high quality generally (UNICEF Innocenti Research Center, 2008). These differences will have a large impact on how childcare relates to children’s development. Interestingly, our conclusions are in line with a recent report from Denmark that failed to find association between type of childcare and negative behavior outcomes in a sample where standards for quality in childcare are regulated through government regulation, similar to the Norwegian system (Gupta & Simonsen, 2009). These findings are in accordance with Love and colleagues' (2003) suggestions that, when standards for quality in childcare are regulated through government legislation, behavior problems may not be explained through childcare variables.

This study has several strengths, including being population based, using multiple analytic approaches, and examining effects of childcare in a country where high-quality and relatively homogeneous universal childcare is provided. Moreover, a particular strength is the sample size: With more than 70,000 mothers participating, all analyses conducted have high statistical power, such that non-significant relationships with a high probability reflect non-existent or negligible associations in the population.

However, the study also has limitations. First, a major weakness of this study was that children’s behavior problems were assessed solely by mothers' reports, and no observational data were available for this. Even though research supported the correspondence between the CBCL and direct assessment (Bilenberg, 1999; Novik, 1999) and several other childcare studies which reported associations between childcare and behavior problems based on maternal reports (e.g., Crosby et al., 2010; Romano et al., 2010), the assessment of behavior problems solely through mothers' reports is a significant limitation of this study.

Second, with the relatively low recruitment rate of 38.5% and further attrition during the course of the study, selection bias is probable. In particular, disadvantaged families may have been underrepresented in the study. However, best practice recommendations for handling longitudinal studies with sample attrition were followed (Widaman, 2006). In addition, in Norway the impact of bias of that kind may be reduced by the existence of high-quality social services provided by the government, which minimizes variation among families.

Third, quality in children’s care arrangements was not measured directly. The study would be stronger had it been able to examine quality as a moderator in the relation between type of and age of entry into childcare and children’s behavior problems. However, the fact that universal childcare in Norway is of relatively homogeneous good quality (UNICEF Innocenti Research Center, 2008) still provides the possibility of examining the relationship between high-quality childcare in general and behavior problems, even though no information about the quality at each center is available.

Fourth, the internal consistency of the internalizing scale at 18 and 36 months was relatively low and may have affected the results in this study. Finally, with the current data it was not possible to separate children in family day care from those in the care of a nanny. However, relatively few children in Norway are in nanny care, so it is not likely that separate analysis of the nanny care and family day care children would affect our results dramatically.

Despite these limitations, the study provides important new insights concerning the impact of childcare on children’s behavior problems. In view of the continuing debate on the topic, this study supports the importance of a global perspective for understanding the effects of childcare. The results from this study support the assumption that there are no adverse behavioral consequences for children using childcare in the first three years of life—in particular, not in the Norwegian context of universally-available childcare of generally high quality. Moreover, research indicates that children may benefit from Norway’s childcare in other developmental areas. A study using the same dataset has shown a positive relation between Norway’s childcare system and children’s language development (Lekhal at al., 2011). However, several studies (Barnes et al., 2010; Belsky et al., 2007) noted that the findings may change as the children mature. An important future research aim would be to follow these children as they grow and enter school. Such research will provide the opportunity to examine both the possible long-term consequences of childcare and the effect of childcare on changes in child outcomes.

Footnotes

The Norwegian Mother and Child Cohort Study is supported by the Norwegian Ministry of Health and the Ministry of Education and Research, NIH/NIEHS (contract no NO-ES-75558), NIH/NINDS (grant no.1 UO1 NS 047537 01), and the Norwegian Research Council/FUGE (grant no. 151918/S10).

References

Achenbach

T. M.

(1992). Manual for the child behavior checklist/2–3 and 1992 profile. Burlington, VT: University of Vermont Department of Psychiatry.

Barnes

Leach

Malmberg

L.-E.

Stein

Sylva

(2010). Experiences of childcare in England and socio-emotional development at 36 months. Early Child Development and Care, 180, 1215–1229.

Bates

J. E.

Freeland

C. A.

Lounsbury

M. L.

(1979). Measurement of infant difficultness.Child Development 50, 794–803.

Belsky

Vandell

Burchinal

Clarke-Stewart

McCartney

Tresch Owen

(2007). Are there long-term effects of early childcare? Child Development, 78, 681–701.

Bilenberg

(1999). The child behavior checklist (CBCL) and related material: Standardization and validation in Danish population based and clinically based samples. Acta Psychiatrica Scandinavica, 100, 2–52.

Borge

A. I. H.

Rutter

Cote

Tremblay

R. E.

(2004). Early childcare and physical aggression: Differentiating social selection and social causation. Journal of Child Psychology and Psychiatry, 45, 367–376.

Bornstein

M. H.

Hahn

Gist

N. F.

Haynes

O. M.

(2006). Long-term cumulative effects of childcare on children’s mental development and socioemotional adjustment in a non-risk sample: The moderating effects of gender. Early Child Development and Care, 176, 129–156.

Bradley

R. H.

Vandell

D. L.

(2007). Child care and the well-being of children. Archives of Pediatrics & Adolescent Medicine, 161, 669–676.

Cote

S. M.

Borge

A. I.

Geoffroy

M. C.

Rutter

(2008). Nonmaternal care in infancy and emotional/behavioral difficulties at 4 years old: Moderation by family risk characteristics. Developmental Psychology, 44, 155–168.

10.

Crockenberg

S. C.

(2003). Rescuing the baby from the bathwater: How gender and temperament (may) influence how child care affects child development. Child Development, 74, 1034–1038.

11.

Crosby

D. A.

Dowsett

C. J.

Gennetian

L. A.

Huston

A. C.

(2010). A tale of two methods: Comparing regression and instrumental variables estimates of the effects of preschool child care type on the subsequent externalizing behavior of children in low-income families. Developmental Psychology, 46, 1030–1048.

12.

Currie

(2001). Early childhood education programs. Journal of Economic Perspectives, 15, 213–238.

13.

Gupta

N. D.

Simonsen

(2009). Non-cognitive child outcomes and universal high quality child care. Journal of Public Economics, 94, 30–43.

14.

Harrison

L. J.

Ungerer

J. A.

(2002). Maternal employment and infant–mother attachment security at 12 months postpartum. Developmental Psychology, 38, 758–773.

15.

Hesbacher

P. T.

Rickels

Morris

R. J.

Newman

Rosenfeld

(1980). Psychiatric illness in family practice. Journal of Clinical Psychiatry, 41, 6–10.

16.

Hill

Reiter

J. P.

(2006), Interval estimation for treatment effects using propensity score matching. Statistics in Medicine, 25, 2230–2256.

17.

Howes

Olenick

(1986). Family and child care influences on toddler’s compliance. Child Development, 57, 202–216.

18.

Jacob

J. I.

(2009). The socio-emotional effects of non-maternal childcare on children in the USA: A critical review of recent studies. Early Child Development and Care, 179, 559–570.

19.

Lamb

M. E.

Ahnert

(2006). Nonparental child care: Context, concepts, correlates, and consequences. In Damon

Lerner

R. M.

Renninger

K. A.

Sigek

I. E.

(Eds.), Handbook of child psychology: Vol. 4. Child psychology in practice (5th ed., pp. 950–1016). Hoboken, NJ: Wiley.

20.

Lekhal

Zachrisson

H. D.

Wang

M. V.

Schjølberg

von Soest

(2011). Does universally accessible child care protect children from late talking? Results from a Norwegian population-based prospective study. Early Child Development and Care, 181, 1007–1019.

21.

Leon

A. C.

Hedeker

(2007). Quintile stratification based on a misspecified propensity score in longitudinal treatment effectiveness analyses of ordinal doses. Computational Statistics & Data Analysis, 51, 6114–6122.

22.

Loeb

Bridges

Bassok

Fuller

Rumberger

R. W.

(2007). How much is too much? The influence of preschool centers on children’s social and cognitive development. Economics of Education Review, 26, 52–66.

23.

Love

J. M.

Harrison

Sagi-Schwartz

van

Ijzendoorn M. H.

Ross

Ungerer

J. A. …

Cohen

(2003). Child care quality matters: How conclusions may vary with context. Child Development, 74, 1021–1033.

24.

Magnus

Irgens

L. M.

Haug

Nystad

Skjaerven

Stoltenberg

(2006). Cohort profile: The Norwegian Mother and Child Cohort Study (MoBa). International Journal of Epidemiology, 35, 1146–1150.

25.

McCartney

Burchinal

Clarke-Stewart

Bub

K. L.

Owen

M. T.

Belsky

(2010). Testing a series of causal propositions relating time in child care to children’s externalizing behavior. Developmental Psychology, 46, 1–17.

26.

Ministry of Children, Equality and Social Inclusion. (2011). Parental leave policies. 4 April. Retrieved 4 April 2011, from http://www.regjeringen.no/en/dep/bld.html?id=298.

27.

Ministry of Education. (2006a). Rammeplan for barnehagens innhold og oppgaver [Framework plan for content and duties in kindergartens]. Retrieved 4 April 2011, from http://www.regjeringen.no/nb/dep/kd/dok/lover_regler/reglement/2006/rammeplan-for-barnehagens-innhold-og-opp.html?id=278626.

28.

Ministry of Education. (2006b). Direktiv for familiebarnehager [Directive for family day care]. Retrieved 4 April 2011, from http://www.lovdata.no/cgi-wift/ldles?doc=/sf/sf/sf-20051216-1555.html.

29.

Ministry of Education. (2010). Barnehageloven [The kindergarten act]. Retrieved 4 April 2011, from http://www.lovdata.no/all/nl-20050617-064.html.

30.

National Institute of Child Health and Human Development (NICHD) Early Child Care Research Network. (2003). Does quality of child care affect child outcomes at age 4½? Developmental Psychology, 39, 451–469.

31.

National Institute of Child Health and Human Development (NICHD) Early Child Care Research Network. (2006). Child-care effect sizes for the NICHD Study of Early Child Care and Youth Development. American Psychologist, 61, 99–116.

32.

Nilsen

R. M.

Vollset

S. E.

Gjessing

H. K.

Skjaerven

Melve

K. K.

Schreuder

P. …

Magnus

(2009). Self-selection and bias in a large prospective pregnancy cohort in Norway. Paediatric and Perinatal Epidemiology, 23, 597–608.

33.

Nomaguchi

K. M.

(2006). Maternal employment, nonparental care, mother–child interactions, and child outcomes during preschool years. Journal of Marriage and the Family, 68, 1341–1369.

34.

Norwegian Mother and Child Cohort Study (n.d.) Retrieved 10 June 2011, from http://www.fhi.no/moba-en.

35.

Novik

T. S.

(1999). Validity of the Child Behavior Checklist in a Norwegian sample. European Child and Adolescent Psychiatry, 8, 247–254.

36.

Organisation for Economic Co-operation and Development's (OECD) Network on Early Childhood Education and Care. (2010). Network on early childhood education and care. Revised literature overview for the 7th meeting of the network on Early Childhood Education and Care, 17 November. Retrieved 15 June 2011, from http://www.oecd.org/officialdocuments/displaydocumentpdf/?cote=EDU/EDPC/ECE%20(2010)3/REV1%20&doclanguage=en.

37.

Romano

Kohen

Findlay

C. L.

(2010). Associations among child care, family, and behavior outcomes in a nation-wide sample of preschool-aged children. International Journal of Behavioral Development, 34, 427–440.

38.

Rosenbaum

P. R.

Rubin

D. B.

(1984). Reducing bias in observational studies using subclassification on the propensity score. Journal of the American Statistical Association, 79, 516–524.

39.

Schafer

J. L.

Graham

J. W.

(2002). Missing data: Our view of the state of the art. Psychological Methods, 7, 147–177.

40.

Strand

B. H.

Dalgard

O. S.

Tambs

Rognerud

(2003). Measuring the mental health status of the Norwegian population: A comparison of the instruments SCL-25, SCL-10, SCL-5 and MHI-5 (SF-36). Nordic Journal of Psychiatry, 57, 113–118.

41.

Swedish National Institute of Public Health. (2009). Child day care center or home care for children 12–40 months of age—what is best for the child? Rapport 09/2009. Ostersund, Sweden: The Swedish National Institute of Public Health.

42.

UNICEF Innocenti Research Center. (2008). Report Card 8: The child care transition. Florence, Italy: UNICEF Innocenti Research Center.

43.

Widaman

K. F.

(2006). Missing data: What to do with or without them. Monographs of the Society for Research in Child Development, 71, 42–64.

44.

Winsvold

Guldbrandsen

(2009). Kvalitet og kvantitet: kvalitet i en barnehagesektor i sterk vekst [Quality and quantity: Quality in a child-care sector in rapid growth]. Research report 2/2009. Oslo, Norway: NOVA.

45.

Youngblade

L. M.

(2003). Peer and teacher ratings of third- and fourth-grade children’s social behavior as a function of early maternal employment. Journal of Child Psychology and Psychiatry and Allied Disciplines, 44, 477–488.