Stimulation beliefs, parental reading involvement, and social inequalities in children’s language development

Objective: This study investigates whether greater identification of mothers and fathers in different-sex couples with the stimulation dimension of intensive parenting promoted both parents’ involvement in reading and benefited children’s language development between ages 3 and 5 in the UK. In addition, we explore social class variations. Background: Our study tests the frequently assumed relationship of parenting beliefs about stimulation with language development, and contributes to our understanding of parenting practices and their implications for child development. Method: We draw on a large representative sample of young children from the UK Millennium Cohort Study (N = 8,071) and apply path analyses in the framework of structural equation models. Results: For mothers and fathers, stimulation beliefs partially mediated the relation between parental education and reading frequency. Mothers’ and fathers’ stimulation beliefs had positive effects on children’s language abilities, but their impact was small compared to the direct associations with parental education. Conclusion: While parental education emerges as a key determinant of children’s language development, our study reveals the nuanced role of beliefs about stimulation within intensive parenting, prompting further investigation into the multifaceted nature of parental involvement.


Introduction
Over the past decades, parenting behaviors have intensified and perceived demands on parents, especially mothers, have increased (Faircloth, 2020;Gauthier et al., 2021;Schiffrin et al., 2015;Smyth & Craig, 2015).Gauthier et al. (2021) note that it is no longer sufficient to fulfill the child's basic needs; instead, it has become the parents' responsibility to promote their children's physical, cognitive, and socio-emotional development (Craig et al., 2014;Faircloth, 2020), often by investing significant amounts of time and financial resources.
While most of the scientific discussion is centered on the expansion of motherhood (Craig et al., 2014), pressures on fathers to get more involved in their child's upbringing and education have increased as well (Gauthier et al., 2021;Wall & Arnold, 2007).Expanding fathers' involvement in child care has also been an explicit policy objective, and many European countries have implemented individual parental leave entitlements for fathers.Although these policies have primarily been aimed at facilitating the reconciliation of work and family care for mothers, they might also pass on the intensive parenting logic to fathers, increasing the child care time for both parents (Doepke & Zilibotti, 2019;Faircloth, 2014a).Building on the broader perspective of reproduction of social inequalities by transmission of cultural capital from parents to children (Bourdieu, 1977), Lareau (2003) has argued that parents of higher socio-economic status (SES) are more likely to purposefully organize their children's play time with structured activities as part of a childrearing logic that she termed 'concerted cultivation'.Through means such as increased cognitive stimulation, these practices are meant to foster children's development and to help them persist in competitive environments.Although concerted cultivation and intensive parenting practices have been previously found to be more prevalent among high SES parents (Attanasio et al., 2022;Bonke & Esping-Andersen, 2011;Fischer, 2022;Fox, 2006;Jaeger & Breen, 2016;Schiffrin et al., 2015), some recent evidence suggests that this relation may be growing weaker (Craig et al., 2014;Weisleder & Fernald, 2013).
Increased perceived parenting 'obligations' do not come without negative consequences for parents: Mothers and fathers alike have been found to compromise their mental health by trying to reach the unattainable standards raised by intensive parenting (Faircloth, 2020;Forbes et al., 2020;Rizzo et al., 2013).Thus, it seems worthwhile to investigate in more detail whether the cognitive stimulation dimension of these parenting logics is indeed as beneficial for children as has been assumed.
In the study at hand, we focus on parenting beliefs about stimulation, which are integral to intensive parenting and concerted cultivation.It can be assumed that these beliefs are strongly associated with parental activities aimed at furthering their child's development.Additionally, we concentrate on one activity, reading to children, which has frequently been found to be among the most effective for promoting children's language skills during early childhood and elementary school (Boonk et al., 2018).We analyze how the frequencies with which mothers and fathers in different-sex two-parent families read to children is related to changes in children's language development in the UK for children between the ages of three and five years.We also examine whether beliefs about stimulation and the frequency of reading with children as one intensive parenting practice are more widespread among parents with higher educational qualification and hence potentially contribute to reinforcing social inequalities in child development.Drawing on a large representative sample of young children from the UK Millennium Cohort Study (MCS; University of London, Institute of Education, Centre for Longitudinal Studies, 2020Studies, , 2021aStudies, , 2021bStudies, , 2021c)), we extend previous research on parenting and home learning environments by empirically examining indirect effects of both parents' level of education on their reading frequency and on their child's language development by considering the trend of increased stimulation efforts and their potentially mediating effects.Our study tests the frequently assumed relationship of these beliefs with language development, and contributes to our understanding of parenting practices and their implications for child development

Intensive parenting: A widespread ideal?
In her seminal work, Hays (1996) coined the term 'intensive mothering' to describe the prevailing discourse of contemporary parenting.She conducted interviews with mothers to explore their beliefs about child-rearing and found that it was predominantly considered a maternal obligation, rather than a shared parental responsibility.Furthermore, she outlined five key features of intensive motherhood: being a 'good mother' required child-centered, expert-guided, emotionally absorbing, labor-intensive, and financially costly practices (p.8).Hays was one of the first scholars to critically examine this ideology and its implications for mothers' well-being and gender inequality.Contemporary parenting discourse addresses intensive parenting messages to both mothers and fathers, acknowledging the increasing involvement of fathers in child care (Faircloth, 2020;Shirani et al., 2012).While cultural emphasis remains largely on mothers, recent research notes a shift in men's perception of 'good fathering', (at least to some extent) aligning with the 'intensive' mothering model (Dermott, 2014).As existing economic studies on beliefs on the importance of parental time investments in children also do not find significant variations between mothers and fathers in England (Attanasio et al., 2022), we focus on the stimulation dimension of intensive parenting beliefs, extending beyond the realm of intensive mothering.Existing evidence based on surveys in which higher educated respondents were oversampled and on qualitative studies suggest that intensive parenting is a widespread ideal to which parents react when they develop their parenting style (Faircloth, 2014b).In their qualitative study of Australian parents of preschool-aged children, Smyth and Craig (2015) discovered that only a very small minority mentioned any doubts or unwillingness regarding conscious efforts to further children's learning.They conclude that most parents simply accepted it as part of their parental duties.
To date, representative information as to how widespread the endorsement of intensive parenting beliefs is across different social groups and countries is still scarce.One exception is the study by Gauthier et al. (2021), which is based on representatively sampled survey data from the CROss-National Online Survey panel (CRONOS).They found that the defining characteristics of intensive parenting seem to be comparable in Estonia, Great Britain, and Slovenia.Yet, the authors admitted that the generalizability of their results might be limited due to a very low overall response rate.Two recent economic studies based on three surveys of parents in England (Attanasio et al., 2022;Boneva & Rauh, 2018) showed mixed results regarding SES differences in the economic returns of time investments in school children, such as helping with school work, which may be interpreted as being related to parental stimulation beliefs.While Attanasio et al. (2022) found no significant differences by education or income of the parents for primary or secondary school children in a representatively sampled survey, in two non-representative parent surveys by Boneva and Rauh (2018), parents with low SES perceived the economic returns to early investment in primary school children to be significantly lower compared to parents with high SES.Furthermore, their findings indicate that parents from lower SES backgrounds exhibit a reduced tendency to endorse the malleability of children's skills or their capacity for skill acquisition.

Intensive parenting and child development
At the core of intensive parenting is the fundamental belief that increased parental involvement is instrumental in fostering child development, a notion substantiated by prior research (Conger & Donnellan, 2007;Sandler et al., 2011).Building upon this foundation, previous studies on the relationship between parental involvement and children's cognitive development point to significant variations across different activities (Cano et al., 2019;Fiorini & Keane, 2014;Hsin, 2009;Hsin & Felfe, 2014).Notably, these studies underscore that parent-child time invested in educational activities exerts more positive effects on cognitive development than time spent in other activities.This discernment aligns seamlessly with the intensive parenting philosophy, emphasizing the value of intellectual stimulation and active engagement in a child's developmental process.Therefore, the mediating role of parental beliefs can be seen as a theoretical bridge connecting the overarching belief in heightened parental involvement to specific educational activities, such as reading, as a means to enhance children's language development.The influence of parental investments on reading frequency may be channeled through the intermediary of parental beliefs, illuminating the cognitive mechanisms through which intensive parenting practices impact children's language outcomes.
We add to some previous economic research on specific parenting intervention programs, which also analyzed the impact of parental beliefs on time investments and child development (e.g., Carneiro et al., 2019).Specifically, we concentrate on the identification of mothers and fathers in different-sex couples and their alignment with the stimulation dimension of intensive parenting.Despite the existing economic evaluation studies on specific intervention programs and target populations, there exists a notable gap within the sociological literature and more representative populations as to how these intensive parenting beliefs translate into the frequency of educational activities and their subsequent influence on children's development.We aim to contribute unique insights to the broader discourse on early childhood development, illuminating the intricate dynamics between parental beliefs, engagement in educational activities, and children's language development during the critical ages of three to five in the UK.Our approach complements existing research, which emphasizes the importance of activities like reading in shaping language skills and contributing to brain development (Sulzby, 1985;Takeuchi et al., 2015).
In one of their meta-analyses using a subset of intervention studies from the Database of Raising Intelligence (DORI), Protzko et al. (2013) found that regularly reading to a child in an interactive fashion raised his or her IQ by over six points.However, this only occurred if the parents started this habit before the child was four years old, and benefits were larger for early starters.Although Schiffrin et al. (2015) point out that there might be an 'optimal range' (p.2322) of involvement and that further stimulation could even impede the child's development once a certain level of stimulation has been reached, this concern is more likely to apply in cases of too many structured activities.Reading 'too much' to a child is unlikely to negatively affect the child's cognitive development around the age of three to five years.

Intensive parenting: Reproducing social inequalities?
According to sociological perspectives (Bourdieu, 1977;Lareau, 2003), parents of higher SES transmit educational advantages to their children, inter alia by actively transmitting cultural capital and fostering children's social and cognitive skills as part of their parenting practices.In this sense, Lareau's concerted cultivation conceptualization may be considered as closely intertwined with the intensive parenting perspective.Stimulation -the focus of this study -is a central tenet of both.One could argue that concerted cultivation is a type of intensive parenting -as are 'helicopter parents' and 'tiger moms'.According to Adrian (2019), each of these parenting styles 'requires significant parental investment, and there is overlap between the [..] styles, but the popularity, methods of engagement, and outcomes are different ' (p.33).
Several scholars have pointed out that intensive parenting is especially widespread in families of the middle and upper class, partly because the associated practices are time-intensive and require substantial financial resources (Fox, 2006;Nelson, 2010), and partly because the pressure to 'form' and foster children in this way to enable them to lead successful lives has (allegedly) been focused on these social classes (Schiffrin et al., 2015).Indeed, recent empirical studies have found that highly educated mothers and fathers in particular spend a considerable amount of time in activities that stimulate the development of their children (Boonk et al., 2018;Fischer, 2022;Jaeger & Breen, 2016).
Similarly, when examining variations in parental beliefs concerning the efficacy of investments over different time periods, Boneva and Rauh (2018) reported that parents of lower SES perceived the returns on helping children with school work in school years 3 to 6 to be markedly lower compared to those of higher SES.The authors also found that perceived returns on time investments correlated with actual time spent on activities with children.
However, the strength of the association between parental education and their involvement in child care tasks appears to vary by the type of task (Gauthier et al., 2021) and might be declining based on recent results from an Australian time use study (Craig et al., 2014).Attanasio et al. (2022), who investigated parental perceptions of the returns to parental time investments, material investments, and school quality among a representative sample of parents residing in England, did not detect a significant positive association between parental education and time invested in children.Moreover, for parents of elementary school-aged children in the US, Ishizuka (2019) found general support of concerted cultivation beliefs regardless of social class.It has to be noted though that he did not examine specific child care tasks.Instead, he used vignettes which covered parallel interactions in time-bound situations in three domains (organization of leisure, language use, and interactions with institutions).
For reading with children specifically, most evidence across Western countries still points to parents with lower SES engaging less frequently in shared book reading with their children (Hemmerechts et al., 2017).As a consequence, the former children tend to have smaller vocabularies and process language more slowly when they are young (Fernald et al., 2013), leading to a disadvantage in language and cognitive development that might translate to lower academic achievements in the future (National Institute of Child Health and Human Development Early Child Care Research Network, 2005;van Zwieten et al., 2021).To date, studies exploring the extent to which intensive parenting beliefs may contribute to this phenomenon are still scarce.However, recent research by Carneiro et al. (2019) constitutes a pivotal stride towards addressing this gap in literature by elucidating the impact of interventions designed to modify parental self-efficacy beliefs and practices on child development outcomes.Their study empirically evaluated a large-scale parenting program implemented in Chile, specifically targeting economically disadvantaged families.Utilizing a semi-structured curriculum facilitated by trained professionals, the intervention aims at reshaping parental beliefs regarding their pivotal role in fostering positive practices conducive to enriched parent-child interactions.The findings of Carneiro et al. (2019) unveil sustained effects on parental selfefficacy beliefs, practices, and subsequent child outcomes, persisting even three years post-intervention cessation.This contribution offers invaluable insights into the intricate relationship between parental investment beliefs and the observed disadvantages in language and cognitive development among children within marginalized communities.
Extending these findings, we focus on the stimulation dimension of intensive parenting to further investigate whether parents with different educational qualifications vary in the weight they put on stimulation and whether this is subsequently linked to more frequent reading and improved language development for children.

The present study
Despite the sustained interest that parental involvement and child development have elicited in the research community, to date we lack representative evidence as to how intensive parenting beliefs relate to parental SES on the one hand and child development on the other to better understand how they contribute to reproducing social inequalities.
We extend previous studies in various ways: First, we draw on representative data of young children in the UK, which allow us to capture the stimulation dimension of intensive parenting for mothers and fathers, and investigate how beliefs in the importance of cognitive stimulation vary between mothers and fathers of different educational backgrounds.Following Koops (2021), we use parental education as an indicator of parental SES, as it is rather stable over the life course, unlike profession or income which are prone to change, particularly when children are still young.
Second, we explore how beliefs about stimulation relate to both parents' involvement in reading to children during a critical phase for literacy development between the ages of three and five, when children in the UK usually start primary school.We focus specifically on reading to children as one activity that has received widespread support for benefiting children's language and literacy development during early childhood and the transition to elementary school (Carpentieri et al., 2011;van Steensel et al., 2011).
Shared book reading, a practice highly compatible with the tenets of intensive parenting outlined by Hays (1996), exhibits characteristics that can be both labor-and time-intensive.
Contemporary parenting advice underscores the significance of dedicating 20-30 minutes daily to reading with one's child (Renaissance Learning, Inc, 2015).This recommendation, while well-intentioned, places considerable pressure on parents, often grappling with time constraints.Research by Sidebotham (2001) reveals that parents may experience feelings of guilt and a sense of inadequacy when unable to meet these expectations, compounded by the societal belief that they should enrich their children's lives with various beneficial activities.
It is crucial to clarify that the stress associated with shared book reading is not necessarily intrinsic to the activity itself but emerges from the cumulative demands across multiple areas and societal expectations.Our acknowledgment of these challenges aims to elucidate the broader context within which shared book reading operates as one among several expected parental activities.In contrast to most previous studies, which were either unable to distinguish between activities performed by mothers and fathers or used time diaries that were usually completed by the mother, we draw on reports of reading involvement by both mothers and fathers.
Third, we examine any direct and indirect effects through reading involvement of the stimulation beliefs for both parents on children's language development.We focus on just one aspect of cognitive development, that is, children's vocabulary, for which close causal links with reading with parents have been suggested (Carpentieri et al., 2011;McWayne et al., 2013).Across all analyses, we focus on progress in children's language development between age three and five years by controlling for children's baseline language skills.
Fourth, by conducting a path analysis in the framework of a structural equation model (SEM), we are able to consider direct and indirect relations of mothers' and fathers' educational backgrounds, beliefs about stimulation, and reading involvement with children's language development in the same model and examine whether the beliefs of either parent represent an important mediator of these relations.
We formulated the following hypotheses: Hypothesis 1: More highly educated mothers (H1a) and fathers (H1b) are more likely to hold beliefs that emphasize the importance of stimulation for a child's development.
Hypothesis 2: Mothers' (H2a) and fathers' (H2b) beliefs about stimulation mediate the relation between their own educational attainment and their own involvement in reading to the child.
Based on widespread evidence that frequent reading with a parent is beneficial for a child's language development, we expected the following mediation relations: Hypothesis 3: Mothers' (H3a) and fathers' (H3b) beliefs about stimulation mediate the relation between their own educational attainment and children's language development.
In addition, parents who believe that stimulation is important for a child's development are likely to promote this development in various other ways, such as talking to children and performing other stimulating activities, leading us to expect an additional direct effect on children's language development: Hypothesis 4: Even after accounting for the relations with parental reading to children, mothers' (H4a) and fathers' (H4b) beliefs about stimulation are expected to be associated with children's language development.

Sample
To examine the relations between beliefs about stimulation, the frequencies with which mothers and fathers read to children, and children's early language development, we used the UK Millennium Cohort Study (University of London, Institute of Education, Centre for Longitudinal Studies, 2020Studies, , 2021aStudies, , 2021bStudies, , 2021c)).The multidisciplinary, nationally representative longitudinal study is funded by the Economic and Social Research Council and a consortium of government departments.At the time of recruitment, the sample comprised 19,519 children born in the UK between September 2000 and January 2002.The data set is well suited for this investigation, as the data was collected several years after Hays (1996) influential book on intensive mothering was published and contain self-reported parenting beliefs of mothers and fathers.
The study employed a clustered stratified design to ensure an accurate representation of the smaller countries in the UK, impoverished areas, and ethnic minorities.A total of 18,552 families agreed to participate in face-to-face interviews in their homes when they were contacted at 9 months, thus resulting in information on 18,818 children (including twins and triplets).We used the second and third follow-ups at child ages three (MCS 2) and five (MCS 3) because we required comparable measurements of children's language development over time, which were available at only these two time points.The phase between ages three and five is particularly interesting because reading to children should be especially effective for children's literacy development as they transition to elementary school, whereas older children are more likely to read more on their own.The implemented multi-actor approach with repeated interviews of mothers and resident fathers or partners of the mothers provide us with self-reports of stimulation beliefs and reading activities by mothers and fathers, respectively.
For families that participated in Waves 1 to 3 (N = 13,235), we included only two-parent families consisting of the same partners until wave 3 (N = 9,393).Families in which the respondents were not able to participate in the interviews (themselves) were excluded as well (n = 286), as were those in which respondents did not give any information on their current legal marital status (n = 22).Due to the specific circumstances and small sample sizes of these subgroups, we excluded families where one parent or the child lived in the household only part-time (n = 895), and twin births (n = 119).These criteria resulted in a sample of 8,071 families.

Key variables
We measured children's language competencies with the ability scores from the naming vocabulary subscale of the British Ability Scales (BAS), based on tests that were administered when the child was roughly three and five years old.The tests assess the children's knowledge of the English language by showing them colored pictures of various objects and asking them to identify these (Hansen, 2014).The ability score reflects the total number of correct responses (accounting for the difficulty of the item sets) and was obtained by transforming the raw score using item response theory and the Rasch model (Rasch, 1960(Rasch, , 1961) ) in particular.Readers interested in the specifics of the transformation from raw scores to ability scores are referred to Elliot et al. (1996).The item sets depended on the child's age, and the successful completion of item blocks.The child was tested until (s)he was not able to correctly name five consecutive items, or until (s)he incorrectly identified three or more items after item 16 (ear), or after item 30 (hourglass), or until (s)he reached item 36 (easel) of the naming vocabulary subscale.The pool of items was the same in Waves 2 and 3, a complete list can be found in the appendix.
Mothers' and fathers' beliefs about stimulation were considered by creating an additive index of two items for each parent (αMother = .68;αFather = .61),entailing their beliefs on whether stimulation and talking to the child is important for his or her development.The items used were '...babies need to be stimulated if they are to develop well' and '...talking, even to a young baby, is important.'Originally, the response options were measured on a 5-point Likert scale, with 1 indicating 'Strongly agree', 3 expressing 'Neither agree nor disagree', and 5 referring to 'Strongly disagree' but were reverse coded to ensure higher values indicated stronger agreement with intensive parenting beliefs.Both items were adapted from the Avon Longitudinal Study of Parents and Children (ALSPAC) 1 , a renowned birth cohort study.They are similar to one item of the Intensive Parenting Attitudes Questionnaire (IPAQ) (Liss et al., 2013).'It is important to interact regularly with children on their level (e.g., getting down on the floor and playing with them).'Although the measures did not specify who should be providing the stimulation and who should talk to the child, it can be assumed that the respondents answered these questions with the parental role in mind, as this information was collected when the child was nine months old -an age where the children spent most of the time with their parents.For both items, the large majority of mothers and fathers strongly agreed that stimulation and talking, respectively, are important for a child's development, followed by between one fifth and one quarter, who agreed.Only a small minority attached lower importance to stimulation.The mean for the combined index was about 4.8 for mothers and fathers with variation from 1 to 5.
Parental involvement in reading was measured when the child was three years old, by a question to mothers and fathers, respectively, asking them how often they read to the target child.They were given six answer options, starting with 'Every day' (1) and ending with 'Not at all' (6).As very few parents read to their children less than once a week, we combined two categories.To facilitate the interpretation of effects, we further reversed their order, resulting in four categories: 'Rarely' (1), '1-2 times a week' (2), '2+ times a week' (3), and 'Every day' (4).
Parents' educational attainment at child age three was measured by their respective levels of National Vocational Qualification (NVQ), which are British work-based awards that can be achieved through assessment and training.They were first proposed by the Review of Vocational Qualifications (Young, 2011) and range from 'Level 1' (lowest) to 'Level 5' (highest).More specifically, Level 1 corresponds to the ability to perform basic work routines, whereas Level 5 refers to senior management activities.The NVQ also takes academic qualifications into account: Level 1 comprises basic school-leaving qualifications, and Level 5 covers postgraduate qualifications or higher degrees.For more detailed information on how qualifications have been assigned to NVQ levels during that time, interested readers are referred to Jessup (2003).NVQs have been widely used as measure for educational attainment and have been deemed useful for empirical research (Dearden et al., 2002): Contrarily to using measures such as years of education as a proxy for educational attainment, NVQs take into account that British qualifications frequently require a similar amount of time in education, although they do not necessarily provide graduates with the same level of competences (Connelly et al., 2016).

Covariates
We considered a number of possible confounding influences on stimulation beliefs, parental reading, and children's language development as control variables throughout the analysis.Most control variables were measured when children were about three years old.
Recent research has shown that traditional gender role beliefs were related to adherence to the intensive parenting ideology (Lamprianidou & Gibb, 2022), so we included a measure of mothers' and fathers' gender ideologies.An approach similar to the one for intensive parenting beliefs was employed.At child age nine months both parents indicated their level of agreement regarding three statements: 'Mother and family are happier if she goes to work', 'A child suffers if the mother works before s(he) starts school', and 'Family suffers when a woman has a full-time job'.Again, answer options were measured on a 5-point Likert scale, with 1 indicating 'Strongly disagree' and 5 referring to 'Strongly agree'.After recoding the variables so that higher values reflected more modern opinions, we generated additive indices for both parents (αMother = .60;αFather = .63).
It has been shown that parental psychological well-being is essential for positive parent-child interactions and child well-being (Newland, 2015), and that depressive symptoms (often used as an indicator for psychological well-being) limit a parent's ability to engage in favorable parenting practices (McLearn et al., 2006).Furthermore, as Larson et al. (2015) found that maternal depression partially mediated SES gradients in US children's cognitive ability at kindergarten entry, we included a dummy that captured whether the mother was experiencing depressive symptoms.This dummy was based on eight symptoms (being e.g.easily upset, constantly jittery, or often miserable), and if a mother reported four or more of these, she was considered to suffer from depressive symptoms.Unfortunately, we could not include the same variable for fathers due to a higher proportion of item non-response and low prevalence of depressive symptoms, which led to little variation on this variable.
Whereas almost all mothers in the UK are very regularly involved in various child care activities, fathers' involvement varies and has been shown to relate positively to children's cognitive development (Cano et al., 2019).We therefore control for fathers' involvement in playing and putting the child to bed.To control for other possible sources of educational stimulation, we included a dummy variable indicating whether the child's grandparents live in the family home as well.
Several dummy variables indicating whether the mother and the father speak only English or also another language at home were included because this might affect the child's exposure to English vocabulary, which in turn should influence his or her vocabulary test score.Additionally, we considered whether the child belonged to an ethnic minority based on the mother's reports.
Whether the household income was below the poverty line was used as a proxy for the family's financial resources, which have been found to be associated with children's achievement (see e.g.Hansen, 2014;Sylva, 2014).The child's exact age (years plus months) and his or her birth weight were included to control for variations in cognitive development (Hansen, 2014;Larson et al., 2015).As previous research has shown that girls hold a small advantage over boys regarding verbal and non-verbal abilities in early childhood (Toivainen et al., 2017), the child's sex was also taken into account.The number of siblings living in the household was deemed an indicator of the time parents were able to allot to the cohort child.Table 1 presents descriptive statistics for the variables used in the analyses.

Analytic strategy
To investigate the joint association of mothers' and fathers' educational qualifications with each parent's beliefs about stimulation and parental reading involvement and the direct and indirect associations with children's language development, we conducted path analyses in the framework of standard linear SEM using Stata 16.1.Wherever applicable, we specified a covariance between the error terms of mother's and father's reading frequency, as well as for those of their respective stimulation beliefs.Covariances between all exogenous variables are automatically assumed by the software.For all models, we specified that we wanted to obtain standardized coefficients to facilitate interpretation of the results.We specified the maximum likelihood with missing values (MLMV) option, which is synonymous with full information maximum likelihood (FIML).Stata's default approach in structural equation modeling is to use listwise deletion, but the goal of MLMV is to utilize the maximum amount of information that can be gained from observations with missing values (StataCorp, 2013).Therefore, all observations are used.However, there are two underlying assumptions for this method that should be taken into account: joint normality of all variables and that values are missing at random.Furthermore, the analyses took the survey design of the MCS into account: samples from two strata per UK country (advantaged and disadvantaged).We also considered the clustering at ward level (primary sampling units), and used a finite population correction factor, which determines the extra precision that is achieved when the sample size approaches the population size (Ketende & Jones, 2011).Additionally, we employed the design and non-response weight for wave 3, which was provided with the data set for analyses based on the whole UK data.All of these adjustments -for clustering, stratification, and weighting -were implemented by using the Stata tool set for complex survey data (svy).
We also re-ran the analyses without adjustment for survey characteristics and show the results of both approaches below.Due to the MLMV option, the standard errors are linearized in the model with adjustment for survey data.In the model without this specification, they are obtained from the observed information matrix.
It should be noted that we were unable to model change over time in all variables, despite using data from several waves.However, this should not have biased our results significantly: First, we assume that levels of education of parents are fairly stable across early childhood.Second, we do not believe that short term in-or decreases in parental reading frequencies translate directly to changes in vocabulary scores.To reduce potential bias as much as possible, we included several covariates that have been shown to impact the outcomes under study, thus decreasing the risk of omitted variable bias.Although path analyses generally do not allow for a causal interpretation of their results, we lagged our explanatory variables to support our assumptions about directionality.

Descriptive results
We started our analyses by examining the bivariate correlations of the main variables (see Table 2).Most of the correlations were significant at the 0.1%-level.As could be expected, the strongest correlation was between the two naming vocabulary test scores at age three and five (0.51, p < .001).A father's level of education correlated with his beliefs about stimulation (0.15, p < .001).This was very similar for mothers, whose level of education also correlated with their stimulation beliefs (0.16, p < .001).
For the correlations between beliefs about stimulation and reading frequency, we observed positive correlations for fathers (0.15, p < .001)and for mothers (0.14, p < .001).Level of education was also positively correlated with reading frequency for both parents (both 0.23, p < .001).Both mothers' and fathers' agreement with the belief that stimulation is important for children's development was positively correlated with the child's language development when (s)he was about five years old (both 0.14, p < .001).The parents' level of education was also associated with their child's language development at this age: for mothers, the correlation coefficient amounted to 0.20 (p < .001),and for fathers it was 0.18 (p < .001).

Multivariate results from path analyses
To test our hypotheses, we conducted several path analyses in the framework of linear standardized (STDYX) SEM.Model 1 was adjusted for survey characteristics.To calculate goodness of fit measures, however, we had to re-run the analyses without the survey estimation command (Model 2), which resulted in coefficients and significance levels similar to those obtained in Model 1.Following commonly used cutoff criteria (Bentler & Bonett, 1980;Hu & Bentler, 1999), the model fit was acceptable.For Model 2, the following values resulted from the goodness of fit test: RMSEA = 0.06, CFI = 0.98, and TLI = 0.60.As we had missing values, Stata was unable to report the SRMR.The overall coefficient of determination (CD) amounted to 0.52.The χ2 test statistic of 190.11 with 6 degrees of freedom and a p-value smaller than 0.001 required the rejection of the model at the 5% level of significance.Subsequently, only the results of the models adjusted for survey characteristics are described.Table 3 as well as Figure 1 depict the results of the path analysis in a standardized SEM (in a reduced form), showing the direct effects of the main variables.
The results indicated that, on average, more highly educated parents agreed more strongly with the stimulation beliefs (0.14 for mothers and 0.13 for fathers, both p < .001).These findings are in line with Hypothesis 1, which posited that more highly educated parents were more likely to believe that stimulation is important for a child's development.
Hypothesis 2 assumed that mothers' and fathers' beliefs about stimulation would mediate the relation between their own educational attainment and their own involvement in reading to the child.In line with this hypothesis, a parent's stimulation beliefs also impacted the respective parent's reading frequency: for mothers, the coefficient amounted to 0.06 (p < .001),and for fathers, it was 0.08 (p < .001).Furthermore, our results showed that parental education remained significantly, positively associated with parental reading frequency (0.16 for mothers and 0.17 for fathers, both p < .001)when simultaneously considering agreement with the stimulation beliefs, indicating that the latter partially mediated the relation.To determine the strength of the mediations, we calculated total, direct, and indirect effects using the estat teffects command (Sobel, 1987).The total effect of mother's level of education was 0.17 (p < .001).This is the effect we would find if there was no mediator.If rounded to two decimal places, the direct effect amounted to 0.16.It was still significant (p < .001),and only marginally smaller than the total effect.The indirect effect of mother's level of education that affected reading frequency through mother's stimulation beliefs was 0.01 and also statistically significant (p < .001).Put another way, the proportion of the total effect that was mediated (i.e. the relative indirect effect) was about 0.05 or 5 %.The ratio indirect/direct effect for mother's level of education was about 0.06.Lastly, the ratio of the total effect to the direct effect was about 1.06.For father's level of education, a similar picture emerged: the total effect was 0.18, the direct effect amounted to 0.17, and the indirect effect via father's stimulation beliefs was 0.01.All of the effects were statistically significant on the 0.1 %-level.In terms of proportions, the proportion of total effect mediated was 0.06, the ratio of indirect to direct effect was 0.06 as well, and the total effect was about 1.06 times the direct effect.
Taking all of these findings together, it can be concluded that there was indeed a small, but statistically significant, effect of a parent's agreement with the stimulation beliefs on reading frequency, thereby supporting Hypothesis 2.
Hypothesis 3 suggested that mothers' (H3a) and fathers' (H3b) stimulation beliefs would mediate the relation between their own educational attainment and children's language development.We found that mothers' and fathers' agreement with the stimulation beliefs was associated with their child's test score (mothers: 0.03, p = .011;fathers: 0.03, p = .004).In line with Hypothesis 3, there was evidence for a partial mediation, as parental education remained significantly, positively associated with the child's naming vocabulary ability score (0.08 for mothers and 0.07 for fathers, with p < .001for both), even when taking into account the parents' agreement with stimulation beliefs.The total effect of mother's level of education was 0.09, the direct effect was 0.08, and the indirect effect of mother's level of education that affected the child's naming vocabulary ability score through mother's stimulation beliefs was 0.01.All of the effects were statistically significant (p < .001).The proportion of the total effect that was mediated was about 12 %.The ratio indirect/direct effect was about 0.13, and the ratio of the total effect to the direct effect was about 1.13.
For fathers, the total effect of level of education was 0.09, the direct effect was 0.07, and the indirect effect of father's level of education that affected the child's naming vocabulary ability score through father's stimulation beliefs was 0.01.Again, all of the effects were statistically significant at the 0.1 %-level.Comparing the proportions, we found that the proportion of total effect mediated was 0.16, the ratio of indirect to direct effect was 0.20, and the total effect was about 1.20 times the direct effect.
Hypothesis 4 assumed that both parents' stimulation beliefs were associated with the child's language development, even after controlling for the effects of parental reading to children.The significant associations of mothers' and fathers' stimulation beliefs with the child's vocabulary test score even after accounting for effects of both parents' reading involvement and educational qualifications lent support to this hypothesis.
Considering the covariates (see appendix, Table A1) for the equations with stimulation beliefs as dependent variable, mothers' and fathers' equations showed many similarities.Parents with more egalitarian gender beliefs tended to agree less strongly with stimulation beliefs (mothers: -0.03, p = .035;fathers: -0.05, p < .001).For mothers, the prevalence of their own depressive symptoms was negatively associated with agreement with stimulation beliefs (-0.04, p = .035).Lastly, fathers' agreement with stimulation beliefs was negatively associated with living in poverty (-0.06, p < .001)and positively associated with playing with the child every day (0.06, p < .001).
The results for the covariates for the reading frequency and the child development equations were in line with previous studies.

Sensitivity analyses
To test the robustness of our results, we re-estimated the model without the adjustment for survey characteristics which led to substantively unchanged results: the estimates of the parameters and the corresponding significance levels were almost identical.We also re-estimated the model using a measure of parental reading frequency by combining mothers' and fathers' reports of reading frequency (see Table A1 in the appendix), which showed similar results regarding parameter estimates and their significance levels.Furthermore, we tested whether the results remained unchanged when basing the analyses on listwise deletion, which they did indeed.Again, parameter estimates and their corresponding significance levels were very close to the ones previously obtained.
To examine whether an increase in children's vocabulary test scores could also be partially attributed to child care and preschool enrollment, we included a variable that captured the weekly number of hours a child spent in child care.Moreover, we accounted for the parents' labor market participation by adding the weekly work hours of mothers and fathers, respectively.However, these analyses resulted in similar parameter estimates and significance levels.

Discussion
Guided by conceptualizations and family research regarding parenting beliefs and practices, we explored how a parent's educational background and his or her agreement with stimulation beliefs influenced the frequency with which they read to children, and how this reading frequency was related to the language development of their children.For our analyses, we focused on mothers and fathers in different-sex twoparent families in the UK with a child between the ages of three and five years.In contrast to many previous studies, we were able to draw on fathers' and mothers' direct reports regarding home learning activities and focused on reading to children as one activity that has received widespread support for benefiting children's language and literacy development during early childhood.Furthermore, we extended current literature by singling out one dimension of intensive parenting, the belief that stimulation is critical to a child's successful development.Using data from the first three waves of the UK Millennium Cohort Study, we addressed the question whether this belief is still more prevalent among high-SES families as suggested by Lareau (2003), or whether it has spread to families with a lower SES as well.
Our results differ from some economic studies, such as Attanasio et al. (2022), who could not detect any significant association between parental education and the amount of time invested in children.However, our analyses support previous results in other fields, in that, on average and compared to their counterparts with a lower level of education, parents with a higher level of education agreed more strongly with stimulation beliefs (Craig et al., 2014;Nelson, 2010;Schiffrin et al., 2015).It is also in line with Boneva and Rauh (2018), who discovered that parents from lower SES backgrounds perceived significantly diminished returns on early investments in helping children with school work compared to their higher SES counterparts.We could not replicate Gauthier et al.'s (2021) finding for the UK, Estonia and Slovenia that a higher level of education was associated with lower scores on the stimulation and child centeredness dimension among parents.However, it has to be noted that they included parents of children of all ages and their stimulation measure mainly related to extracurricular activities.Additionally, Gauthier et al. (2021) also acknowledged that their results can only be generalized with caution due to the overall low response rate in the original sample.Considering all this, it becomes apparent that more work is needed for a better understanding of the relations between SES and different dimensions of intensive parenting across different countries and whether this has changed over time.
Closely related to this, our analyses also provided evidence for the positive relation between parental educational background and reading frequency, which is in line with the perspectives of cultural capital transmission (Bourdieu, 1977), concerted cultivation (Lareau, 2003) and with previous empirical studies (Ehmig & Reuter, 2013;Guryan et al., 2008).We extended previous studies by jointly estimating the effects of mothers' and fathers' education on stimulation beliefs and parental involvement and consequently, on children's language development.While the concept of intensive parenting has received increasing attention by family scholars over the past years, its direct and indirect effects on child development have rarely been investigated empirically in Sociology.We found that agreement with stimulation beliefs partially mediated the relation between a parent's level of education and his or her reading frequency.
On top of that, we detected a partial mediation effect of agreement with stimulation beliefs on the association between a parent's level of education and his or her child's language development.The proportion of the total effect that was mediated was larger than for the mediation on parental reading frequency: 12 % for mothers, and 16 % for fathers, respectively.Even after accounting for the positive relation between parental reading frequency and their child's naming vocabulary test score, mothers' and fathers' levels of agreement with the stimulation beliefs remained positively associated with children's language development.In combination with the significant educational gradient of reading to children for mothers and fathers, this suggests that child-rearing practices in this UK birth cohort impacted children's vocabulary development across the transition to elementary school.The modest (mediation) associations of the stimulation beliefs with parental reading frequency and children's language development may be seen as somewhat surprising given the strong links between reading with children and child development reported by other studies (see e.g.Boonk et al. (2018)).Especially compared to the findings of Carneiro et al. (2019), which underscore enduring shifts in parenting practices and home environments, coupled with notable enhancements in children's receptive language skills over the medium term, our findings point to more modest associations.However, it is crucial to note that the intervention they studied targeted disadvantaged social groups, for which effects tend to be larger, and focused on enhancing parental selfefficacy and altering perceptions regarding the impact of parental behaviors on child development.They used a number of psychological scales to measure these perceptions, which were better suited to capture variations in beliefs and practices, whereas variability in the two stimulation belief items in our study was limited, with most parents agreeing or strongly agreeing.At least for our sample, the modest strength of the associations in our study should not be due to a small variance in reading frequencies, as the data was collected in the early 2000s, when daily reading with children was not an imperative practice in the UK.
Another plausible explanation for the observed patterns in our results could be attributed to potential shifts in parental beliefs about stimulation over time.As children undergo significant linguistic advancements from infancy to preschool age, parents may dynamically adjust their strategies to support language development (see e.g.Cunha et al.'s (2010) discussion of endogeneity of inputs).This dynamic nature of parental beliefs suggests that more longitudinal assessments in future research would be beneficial.Such investigations could illuminate the intricate interplay between the evolving nature of parental beliefs and their subsequent impact on children's language outcomes.Furthermore, it is plausible that other parenting beliefs related to children's cognitive development, not explicitly encompassed within the intensive parenting concept, may play a role and potentially obfuscate the influence of stimulation beliefs.For instance, beliefs about the malleability of children's development (Boneva & Rauh, 2018) or about children being 'active constructivists' in their development (Vygotsky, 1978) might influence how parents perceive the necessary frequency of shared activities for their child's successful development.
It is important to note that our analyses can only serve as a preliminary indication of the effects of intensive parenting beliefs, as we only focused on the stimulation beliefs and reading involvement of parents.Although we cannot rule out the possibility that our measure of stimulation beliefs is proxying for unmeasured beliefs about other parenting dimensions, we are confident that these two questions in particular are closely related to reading with children.
However, it is crucial to acknowledge the limitations and challenges associated with measuring stimulation beliefs.There are currently no established instruments specifically designed for this purpose.Historically, most studies have concentrated on parenting activities rather than beliefs (see e.g.Yerkes et al., 2021).In recent years, individual items have been included in large-scale surveys to measure some aspects of intensive parenting beliefs, but these mostly focus on other dimensions, such as child-centeredness, exhaustion and essentialism (Brüderl et al., 2024;Ruckdeschel, 2018).Liss et al. (2013) developed the IPAQ using a snowball sampling technique to recruit participants through Facebook and parenting blogs.This method raises concerns about generalizability.Moreover, the authors did not discuss the distribution of individual items or subscales for different dimensions of intensive parenting and hence, the validity of specific items and subscales remains unclear.
Additionally, the stimulation items in the IPAQ and subsequent adaptations for international studies (e.g.Gauthier et al., 2021) are predominantly concerned with extracurricular activities.These instruments are not as appropriate for young children as the age group under investigation.One of the items included in IPAQ that can be applied for parents of young children, 'Parents should begin providing intellectual stimulation for their children prenatally, such as reading to them or playing classical music,' might offer greater variation in responses among parents with high stimulation ideals and therefore might be promising to include in future studies.Future research should also include measures of stimulation beliefs that pertain to more specific types of stimulation (e.g., cognitive, artistic) and employ validated instruments, as exemplified in the work by Liss et al. (2013).This would undoubtedly contribute to a more nuanced understanding of intensive parenting beliefs and their impact on child development.
We used the BAS naming vocabulary test as an indicator of early language development, but poor performance on this test might also be related to a child's reluctance to speak.However, our models included the previous test score at age three, which should at least partially control for relatively stable personality influences (e.g., shyness and introversion).It also has to be kept in mind that the BAS naming vocabulary test is a test of expressive language skills and cannot reliably capture receptive language abilities.Unfortunately, vocabulary was assessed at only two time points for the whole sample, which prevented us from estimating panel models over longer periods.
Furthermore, we were unable to include any measures of the total time parents spent with children or of the quality of the interactions, including parents' reading proficiency and reading styles which depend on the sensitivity and instructional expertise of the adult (Pollard-Durodola et al., 2015).It was also not possible to control for children's interest in reading, which has been found to mediate the association between parental reading frequency and children's language skills, at least in low-income minority families (Malin et al., 2014).
As is often the case when using self-reported measures, there is a risk of biases due to social desirability.Increasing intensive parenting ideals could lead parents to report such ideals and to overestimate the frequency of reading.Thus, these relationships are difficult to examine using survey data.Yet, this bias is unlikely to drive the associations as using other measures of involvement (such as playing indoor or physically active games, telling stories or engaging in musical activities with the child) did not yield similar results.
Lastly, we have confined attention to a sample of children where both parents are present and remain partnered to each other over the first five years of the child's life.Although this is unavoidable considering our research questions, it has to be acknowledged that the focus on stable, two-parent families has reduced the share of families with lower educational attainment.Testing whether the associations we found are different for single-parent families would be an interesting avenue for future research.
These limitations notwithstanding, our study provides a rigorous test of how one dimension of intensive parenting, beliefs about stimulation, is linked to child development based on representative UK data.
To conclude, mothers' and fathers' convictions that stimulation is indispensable for ideal child development appear to have a modest, positive effect on their child's language abilities, but their impact is small compared to the direct associations with parental education.As our study can only make a small contribution to the broader field of research on intensive parenting by focusing specifically on beliefs related to stimulation, future research should continue to examine in more detail the relation between different dimensions of intensive parenting, e.g.child-centeredness, reliance on experts, and parental responsibility, and SES, as well as their impact on parental involvement, and, consequently, on child development.It is important to gain empirical evidence on which dimensions of intensive parenting and/or parental involvement actually further children's development.This approach is crucial for providing parents with evidence-based information so they do not feel pressured to comply with every aspect of intensive parenting ideals.Moreover, exploring how negative effects of trying to attain goals set by intensive parenting ideals -such as stress and feeling over-whelmed -influence parental well-being and the quality of parental involvement, and indirectly also the child's development, seems another promising avenue for future research.

Figure 1 :
Figure 1: Path diagram and parameter estimates for separate parental reading frequencies.Coeffi-cients are standardized (STDYX).Adjusted for survey characteristics.Control variables were included in the analyses but are not shown due to readability

Table 1 :
Descriptive statistics Note: UK Millennium Cohort Study (MCS), variables from Waves 1 to 3. Based on listwise deletion; weighted.Data on household level.N = 8,071.Percentages may not add up to 100% due to rounding

Table 3 :
Path Analyses for Child's Naming Vocabulary Ability Score at Age 5, FIML Observed Information Matrix.Control variables included: birth weight in kg, naming vocabulary ability score at age three, age in years and months, gender, number of siblings, OECD poverty indicator, whether child belongs to an ethnic minority, whether mother/father speaks other language than English, whether mother showed depressive symptoms, whether a grandparent lives in the household, mother's/father's gender ideology score, frequency with which father gets child ready for bed, whether father plays with child every day.