Changes in children's well-being and mental health across the early school years: Links with academic and social competence.

Are Child Well-Being and Mental Health Distinct Constructs?

According to the “traditional model,” well-being and poor mental health are two ends of the same continuum ( Greenspoon & Saklofske, 2001 ). In other words, the absence of symptoms of mental health conditions is viewed as an indicator of well-being. By contrast, the “dual-factor model” proposes that subjective well-being and mental health are separable and make distinct contributions to children’s developmental outcomes ( Greenspoon & Saklofske, 2001 ; Petersen et al., 2020 ; S. M. Suldo & Shaffer, 2008 ). In line with this view, the World Health Organisation (2022) described mental health as comprising two dimensions representing (a) well-being (i.e., the degree of positive affect and satisfaction) and (b) symptoms of mental health conditions. So, while individuals experiencing symptoms of mental health conditions are more likely to report lower levels of well-being, it is possible to experience high levels of well-being in the context of mental health conditions ( World Health Organisation, 2022 ).

Several studies have investigated the links between mental health and well-being in children and adolescents. For example, data from the U.K. Millennium Cohort Study revealed modest negative correlations between caregiver-reported psychopathology and children’s self-reported happiness, both at age 11 years ( Patalay & Fitzsimons, 2016 ) and at age 14 years ( Patalay & Fitzsimons, 2018 ). Results from age 14 years showed that the correlation between subjective well-being and psychopathology was stronger when both constructs were rated by the same informant, but still not entirely overlapping ( Patalay & Fitzsimons, 2018 ). Data from the Avon Longitudinal Study of Parents and Children in the United Kingdom revealed similarly modest associations between caregiver-rated externalizing and internalizing and self-rated school enjoyment (an indicator of well-being) at ages 10–11 years and 13–14 years ( Cadman et al., 2021 ). The weak-to-moderate correlations between mental health and well-being suggest these two dimensions are distinct, supporting the dual-factor model in middle childhood and early adolescence. However, existing work on the correlations between mental health and well-being has relied on using observed summed scores, which reflect both the underlying true score and measurement error, such that correlations between constructs may be underestimated ( Brown, 2015 ). Latent variable modeling provides a method for testing the overlap or distinctiveness of error-free true scores and comparing the fit of competing measurement models to a given data set (i.e., a “traditional” one-factor model vs. a dual-factor model).

In children, mental health is typically measured using indicators of internalizing (e.g., depressive symptoms, anxiety symptoms, emotional difficulties) and externalizing problems (e.g., conduct problems, impulsivity, oppositional behavior) with standardized scales completed by caregivers (e.g., parents, teachers; Stone et al., 2010 ). Reflecting the long-standing emphasis on a deficit approach to mental health, there is less consensus on how best to measure well-being in children ( S. Suldo et al., 2011 ). Moreover, existing studies of children’s well-being have largely focused on middle childhood and adolescence, using self-reported or informant-rated indicators of happiness (e.g., life satisfaction, positive affect) that focus on either global or specific domains (e.g., school; Morris et al., 2021 ; Patalay & Fitzsimons, 2018 ; Rees et al., 2020 ; S. Suldo et al., 2011 ). As a result, the links between well-being and mental health in younger children are poorly understood.

A key barrier to extending the developmental scope of research on well-being and mental health was the lack of age-appropriate measures of well-being for younger children. To address this challenge, Ford et al. (2012 ) developed a child-friendly seven-item self-report measure of well-being called “How I Feel About My School” (HIFAMS). The HIFAMS has shown acceptable levels of test–retest and internal consistency reliability, as well as modest cross-sectional correlations between observed summed scores for subjective well-being and caregiver- and teacher-rated mental health, echoing findings from older children ( Allen et al., 2018 ). The first aim of our study was to examine the relations between mental health and well-being in early primary education by using latent variable analysis to compare the fit of competing models (i.e., one- vs. two-factor models) and the role of informant effects (i.e., caregiver and child ratings).

Stability and Continuity in Child Well-Being and Mental Health in the Early School Years

Numerous studies have investigated stability (i.e., rank-order consistency) and continuity (i.e., group mean consistency) in mental health across childhood ( Bornstein et al., 2017 ; Patalay & Fitzsimons, 2018 ), but comparatively little is known about stability and continuity in young children’s subjective well-being. In early adolescence, longitudinal data suggest that individual differences in subjective well-being exhibit weaker rank-order stability than measures of psychopathology, suggesting that ratings of subjective well-being capture a stable sense of happiness but may be more influenced by transient feelings than mental health condition symptoms ( Cadman et al., 2021 ; Patalay & Fitzsimons, 2018 ). Patterns of stability in well-being observed in adolescence cannot be extrapolated to earlier developmental periods. Young children show a propensity to live in the moment, both because autobiographical memory emerges slowly across the life span ( Fivush, 2011 ) and because children’s limited working memory capacity constrains their ability to engage in episodic future thinking ( Ferretti et al., 2018 ). On this basis, young children’s ratings of well-being may be less stable over time than adolescents’ and less stable than caregiver-rated child well-being. Challenging this view, 4-year-old children’s self-reported enjoyment at school showed moderate levels of rank-order stability over 6 months ( Jirout et al., 2023 ).

Alongside assessing the stability of individual differences, we were also interested in tracking continuity in mean levels of children’s well-being and mental health. Large-scale cross-sectional work on subjective well-being in middle childhood indicates that child-reported well-being declines between the ages of 10 and 12 years ( Rees et al., 2020 ). Mirroring these findings, longitudinal data from early adolescence indicate within-person decreases in average levels of subjective well-being between 11 and 14 years of age ( Patalay & Fitzsimons, 2018 ). While it is not possible to generalize patterns of discontinuity from early adolescence to early childhood, environmental factors might influence average levels of subjective well-being in early childhood. The move from “Reception” to “Year 1” of primary school (i.e., from the first to the second year of formal schooling in England, equivalent to the transition from pre-Kindergarten to Kindergarten in the United States) is marked by a move from the “Early Years Foundation Stage,” which is characterized by an active play-based approach to learning, to “Key Stage 1,” which is characterized by a more structured, formal curriculum and may bring challenges that are sometimes overlooked ( Fisher, 2010 ). However, it would be wrong to assume that these new demands necessarily lead to reduced well-being. Indeed, greater levels of autonomy could lead to gains in subjective well-being ( Ryan & Deci, 2001 ).

There are likely to be marked individual differences in the degree of change in well-being over time. Alongside the paucity of research examining rank-order stability and mean-level continuity in subjective well-being, existing studies have yet to examine either variation in the degree of change in children’s well-being or whether changes in well-being are related to changes in children’s mental health. Our second aim was therefore to examine stability and continuity in subjective well-being and mental health across the early primary years and to investigate the extent to which changes in well-being relate to changes in children’s mental health. To this end, we used latent change score models to investigate baseline correlations between mental health and well-being, cross-domain associations between initial levels of well-being and changes in mental health, between initial levels of mental health and changes in well-being, and cross-domain coupling in changes in both constructs ( Kievit et al., 2018 ).

Well-Being, Mental Health, and Success at School

According to the dual-factor model, mental health and well-being make distinct contributions to children’s development ( S. M. Suldo & Shaffer, 2008 ). Research on the correlates of mental health and subjective well-being lends support to the dual-factor model. First, longitudinal data indicate that these constructs have distinct predictors. Among 11-year-old children, cognitive ability and parental mental health each uniquely predict child mental health more strongly than they predict subjective well-being, while social relationships are more strongly related to subjective well-being than to child mental health ( Patalay & Fitzsimons, 2016 ). Moreover, subjective well-being and mental health make unique contributions to children’s academic achievement and social competence. Specifically, cross-sectional studies of adolescents show that those with poor mental health and high subjective well-being had better social functioning than those with poor mental health and low subjective well-being ( S. M. Suldo & Shaffer, 2008 ). Over and above potential confounds, such as early cognitive ability and parental socioeconomic status (SES), mental health (i.e., internalizing and externalizing) and well-being (i.e., enjoyment of school) made unique contributions to academic outcomes at age 16 ( Cadman et al., 2021 ; Morris et al., 2021 ). Mental health and subjective well-being were as strong as SES as predictors of academic success ( Morris et al., 2021 ). Subjective well-being and mental health may therefore each contribute uniquely to children’s social and academic “success” at school.

Extending the focus of this work to the first years of primary school is important because problems in early academic performance and social competence each cast a long shadow ( Duncan et al., 2007 ). To isolate unique associations between subjective well-being, mental health, and social and academic outcomes, it is important to include potential confounds such as cognitive ability, parental SES, and gender (e.g., Morris et al., 2021 ). Our third aim was to extend existing work by examining the relations between initial levels and changes in children’s subjective well-being and mental health and early indicators of social and academic skills. We used structural equation modeling to examine the unique associations between well-being and mental health and academic performance and social competence, adjusting our estimates for potential confounds such as earlier cognitive ability, parental SES, and child sex.

Summary of Aims

In summary, our study had three aims. The first aim was to use latent variable modeling to compare the fit of competing models of mental health and well-being (i.e., one- vs. two-factor models) and the role of informant effects (i.e., caregiver and child ratings). Our second aim was to examine stability and continuity in subjective well-being and mental health across the first year of formal education and to investigate how changes in well-being relate to changes in children’s mental health. Our third aim was to investigate the relations between initial levels and changes in children’s well-being and mental health and early indicators of social and academic skills in the first years of formal education.

Participants

Children and their primary caregivers were recruited from across England in the Spring/Summer 2021 via targeted mailings to primary schools and paid targeted social media advertising. To participate in the study, children were required to be enrolled in the first year of primary school in England (“Reception”) and have no history of developmental delay. In England, children typically start the reception year of primary school in the September after their fourth birthday. The primary caregiver and participating child had to be able to communicate in English. We sought to recruit 250 children into the longitudinal study (see the Supplemental Materials for sample size justification). Just under 500 caregivers expressed an interest in learning more about the study ( N = 494), and 260 of these families agreed to participate (52.6%). Of these 260 families, five families did not provide sufficient information to establish eligibility, one child was not attending Reception, and two families planned to leave England before follow-up. At Time 1, 252 children (131 girls) aged 5.40 years ( SD = 0.31) and their caregivers (89.8% mothers, M age = 38.63 years, SD = 4.66) participated in the study. Children were predominantly from two-parent heterosexual households (92.1%). Caregivers were highly educated (83.7% had degree-level education). On the subjective ladder of social status, 74.3% of caregivers rated themselves as 6/10 or above on a 10-point scale where 1 indicated the lowest levels of education, income, and status and 10 indicated the highest levels of education, income, and status. Following the U.K. Census ethnic group categories, 80% of children were identified as “White,” 13% as “Mixed or multiple ethnicities,” 6% as “Asian,” and 1% as “Black.”

Procedure

Children and their primary caregivers were seen on two occasions approximately 12 months apart (mean interval = 12.36 months, SD = 1.08 months) using a remote assessment protocol to mitigate the spread of COVID-19. Data collection was timed to take place when children had completed at least one term in Reception year and then again after completing at least one term in Year 1. At both time points, following written consent, caregivers participated in a remote interview with a graduate researcher lasting approximately 15–20 min and then completed an online questionnaire pack. During this initial meeting, researchers checked for connectivity issues and ensured families had all necessary research materials (i.e., access to a desktop computer or device with a working keyboard, safe receipt of a package that contained supporting materials including stickers and a sticker chart). Children then participated in a remote testing session using videoconference software led by two graduate researchers with a caregiver present. The remote testing session followed a standardized protocol and included a battery of tasks to measure cognitive skills (i.e., executive function, verbal ability, theory of mind), self-concept, and subjective well-being. These sessions lasted approximately 45–60 min. Children were given rest breaks and received rewards (e.g., stickers) for completing each task regardless of task outcome. No feedback on task performance was given during the testing session. Families received a voucher (£10) as a gift for participating in each wave of the study. Teachers were invited to complete a short questionnaire about each study child. Ethical approval for the study was obtained from the Research Ethics Committees at the University of Birmingham and University of Cambridge.

Children’s Mental Health

Primary caregivers completed the Strengths and Difficulties Questionnaire (SDQ; Goodman et al., 2000 ) at Time 1 and Time 2. The SDQ is used in research on children’s mental health and shows evidence of internal consistency and test–retest reliability as well as concurrent validity with more lengthy measures of psychopathology and clinician diagnoses ( Stone et al., 2010 ). The SDQ consisted of 25 items divided into five subscales: emotional problems (e.g., “Often unhappy, downhearted”), conduct problems (e.g., “Often fights with other children”), hyperactivity (e.g., “Easily distracted, concentration wanders”), peer problems (e.g., “Rather solitary, tends to play alone”), and prosocial skills (e.g., “Shares readily with other children”). Each item was scored on a 3-point rating scale (i.e., “Not True,” “Somewhat True,” “Certainly True”). Summed scores for each subscale had a possible range of 0–10, with high scores on each scale indicating greater levels of difficulty (note that in the Prosocial scale, higher scores indicated fewer difficulties). See Supplemental Table S3 for reliability. Based on U.K. normative data (see https://www.sdqinfo.org ), the proportion of children scoring in the “average” range was 83.6% for emotional problems, 72% for conduct problems, 78.9% for hyperactivity, and 79.7% for peer problems.

Academic Skills

Academic skills were measured using a multimethod, multi-informant approach. At Time 2, children completed two direct assessments to capture number skills and reading ability. In the Numeracy Screener ( Nosworthy et al., 2013 ), children decided which of two numbers (ranging from 1 to 9) was greater in a series of 56 symbolic and 56 nonsymbolic pairs. The total number of correct items was summed together and age residualized. In the Sight Word Efficiency subtest of the Test of Word Reading Efficiency ( Torgesen et al., 1999 ), children were given 45 s to read aloud a list of words. The total number of correct words was summed together and age residualized.

Caregivers and teachers reported on children’s academic performance. Caregivers completed the Competence and Coping Questionnaire ( Mischel et al., 1988 ), in which they rated how their child was performing academically relative to their peers along a 7-point scale (e.g., “Not as strong as most peers” to “Much stronger than most peers”). Teachers rated children’s performance on six skills from the National Curriculum (i.e., word reading, comprehension, writing, number knowledge, addition/subtraction, geometry) relative to their classmates using a 7-point scale (i.e., “Not as strong academically as most peers” to “Much stronger academically than most peers”). Ratings were averaged to give a possible score ranging from 0 to 6 (see Supplemental Table S3 for reliability).