Shifting parental beliefs about child development to foster parental investments and improve school readiness outcomes

The study examines why socioeconomic disparities in early child development and parental investments persist and whether shifting parental beliefs about the impact of early inputs can reduce these gaps. Building on human capital theory, the authors posit that parents’ beliefs about how their investments affect child skill formation are key determinants of actual investments, especially from birth to age five when returns are high. They introduce an economic model linking parental beliefs to investments and child outcomes and test its implications with two field experiments targeting low-SES families. The purpose is to assess whether beliefs are malleable, whether belief changes translate into improved parent-child interactions, and whether these changes improve children’s early cognitive, language, and social-emotional outcomes. The work is motivated by the importance of early investments for long-term outcomes and growing SES disparities, and seeks policy-relevant mechanisms to foster school readiness.

The paper contributes to several literatures. Developmental psychology shows parental beliefs predict parenting practices, home environments, and child outcomes and help explain SES disparities in language inputs. Recent economics models incorporate parental beliefs about the technology of skill formation as determinants of investments, documenting SES gradients in beliefs that correlate with investments. This study replicates these patterns and extends the literature by experimentally shifting beliefs and tracing effects on investments and outcomes. It also builds on early language intervention research showing feedback and coaching can enhance conversational turns and improve infant language, broadening evidence to low-SES and Spanish-speaking families and assessing impacts across vocabulary, math, and social-emotional domains. References include Hoff (2003), Rowe (2008, 2018), Cunha et al. (2013, 2020), Attanasio et al. (2019, 2020), Boneva & Rauh (2018, 2020), and intervention trials (e.g., Ferjan Ramírez et al., Leech et al., Leung et al., Suskind et al.).

Design: Two randomized controlled trials (RCTs) with low-SES caregivers in Chicagoland tested whether shifting beliefs about early child development changes parental investments and child outcomes. 1) Newborn Program (NCT02812017): - Setting and sample: 10 pediatric clinics serving medically underserved/underinsured populations. Inclusion: female primary caregivers ≥18; English/Spanish-speaking; infant <30 days; co-resident with legal custody; excluded for significant perinatal/neonatal complications, developmental disabilities, medical problems, or <36 weeks gestation; excluded if SES above thresholds (income >200% FPL or education > bachelor’s), foster parents, or unable to commit. Recruitment during first well-child visit (3–5 days). N=475 parent-infant dyads randomized: Treatment 237; Control 238 (half of controls viewed placebo safety videos; half viewed none). Randomization via Research Randomizer; stratified by language for the newborn study. Consent obtained from parents only. - Intervention: Four ~10-minute educational videos at 1, 2, 4, and 6-month well-child immunization visits, viewed on tablets in clinic waiting rooms. Content: (i) information on brain malleability, infant capabilities, and the role of parents; (ii) practical tips to foster responsive, language-rich serve-and-return interactions using the “3Ts” (Tune In, Talk More, Take Turns). Compliance: ~66% watched all 4 modules; ~25% watched 3; ~8% watched ≤2. Control: placebo safety videos at same visits (subset) or no video; pooled in main analyses. - Outcomes and timing: Repeated assessments of parental beliefs about the impact of parental inputs; parent-child interaction quality; child interactions with parent; child language (CDI) and other skills at 6, 7/9, 12, and 18 months, following the schedule in Fig. 2a. Instruments include SPEAK (beliefs), TOPSE, NCAST, CDI Levels I–II, and TOI; see Supplementary Information (SI) for details. Intent-to-treat (ITT) analyses. 2) Home Visiting (HV) Program (NCT03076268): - Setting and sample: Parents of 24–30-month-old children recruited via clinics, grocery stores, daycare facilities, community fairs, and public transit. Inclusion similar to above; Spanish was the preferred home language. Exclusions for significant cognitive/physical impairments and higher SES as above. N=91 dyads randomized: Treatment 46; Control 45. Consent from parents only. - Intervention: Twelve 1-hour home visits across 6 months (two per month). Each visit included an educational video on a developmental topic (e.g., linguistic interactions, encouragement, embedding math in routines) and coached practice implementing 3Ts during daily activities (e.g., cooking). Second component: individualized feedback and goal-setting using LENA-based recordings of daily language environments (caregiver talk, conversational turns, child vocalizations). Adherence: > two-thirds completed all 12 visits; 24% had <6 visits (9% had none). Control: nutrition information packets reviewed in short home visits every 6 weeks throughout the 6-month period. - Outcomes and timing: Beliefs; parent-child interaction quality (LENA metrics); child interactions; child skills including vocabulary (ROWPVT), math (Woodcock–Muñoz), and social-emotional skills (ASQ-SE, BRIEF-P) at 30–36 months (immediately post-intervention) and 36–42 months (~6 months later). ITT analyses. Measurement and analysis: - Beliefs: Survey of Parents’ Expectations and Knowledge (SPEAK) and related measures; multiple belief constructs collected (SI section 4), with multiplicity corrections across families of outcomes. - Parent-child interactions: Audio/video-based measures (e.g., NCAST; LENA-derived conversational turns, adult word counts; coded interaction quality). - Child outcomes: Language (CDI in Newborn; ROWPVT and Preschool Language Scale in companion data), math (Woodcock–Muñoz), social-emotional (ASQ-SE, BRIEF-P), and child interactions with parent. - Statistical approach: ITT estimates comparing treatment vs. control at each time point; multiplicity-adjusted p-values using List, Shaikh, and Xu (2019) procedure; two-sided t-tests. Balance checks showed minor imbalances (e.g., employment status), with robustness checks reported in SI. Attrition tracked and presented (Fig. 3); additional regressions adjust for imbalances (SI section 7). Power calculations targeted 80% power with 5% alpha and 25% attrition for both studies. Data collection via REDCap; analysis in Stata 15. Additional correlational analysis: - A companion longitudinal study (NCT02216032) of low-SES families followed children from 13–16 to 37–40 months, repeatedly measuring beliefs and language (PLS, PPVT) and social-emotional (ASQ-SE) skills. Simple linear regressions of standardized child outcomes on standardized parental beliefs at multiple ages estimated predictive associations and R-squared values.

- SES gradients in beliefs: At birth, parents from higher SES and with higher maternal education report stronger beliefs that parental inputs affect child development. Low-SES parents’ belief scores average roughly 0.5 SD lower than higher SES parents, with greater variance. Distributions differ at the 1% level (Wilcoxon, p<0.001) between low-SES and both mid- and high-SES groups; similarly significant gradients by maternal education (p<0.001). - Beliefs are malleable: Both interventions increased parental beliefs relative to controls at all measured time points, with effects persisting beyond program end. Effect sizes (standardized) from Table 1: • Newborn program: Beliefs +0.82 at 6m (p<0.01), +0.61 at 7/9m (p<0.01), +0.61 at 12m (p<0.01), +0.38 at 18m (p<0.01). • HV program: Beliefs +1.46 at 30–36m (p<0.01) and +1.25 at 36–42m (p<0.01). Magnitudes are roughly twice those of the Newborn program, with some decline over time in both. - Parental investments (interaction quality): • Newborn program: Small, transient gains in interactions at 7/9m (+0.27, p<0.05), but not sustained at 6m, 12m, or 18m (effects 0.13, 0.10, 0.10; ns after adjustment). • HV program: Interactions improved at 30–36m (+0.62, p<0.10), with smaller and nonsignificant effect by 36–42m (+0.44, p=0.25). - Child outcomes: • Newborn program: Child interactions with parent improved at 7/9m (+0.23, p<0.05) but not consistently at other waves. No consistent gains in vocabulary (CDI) at 6, 7/9, or 12 months (effects near zero and nonsignificant). • HV program: Positive impacts at or after program end: child interactions with parent (+0.50 at 30–36m, p<0.10), vocabulary (ROWPVT) at 36–42m (+0.31, p<0.10), math (Woodcock–Muñoz) at 30–36m (+0.48, p<0.05), and social-emotional skills (ASQ-SE/BRIEF-P) at 30–36m (+0.44, p<0.05) and sustained at 36–42m (+0.45, p<0.10). - Belief-investment linkage: Using random information shocks, the study presents evidence that belief revisions causally increased investments (e.g., improved parent-child interaction quality, especially in the HV program). - Predictive power of beliefs (companion longitudinal data): Standardized parental beliefs are positively associated with standardized child skills across ages and measures. For language (PLS, PPVT), coefficients typically range ~0.8–1.4 SD increase in language per 1 SD increase in beliefs depending on age pairing, with R-squared up to 18.7% (beliefs at 19–22m predicting PLS ~18 months later). PPVT R-squared are ~12–16%. Social-emotional associations are positive and modest (R-squared ~1–8%). Overall, beliefs alone explain a substantial share of variance in language outcomes (often 13–18%).

The findings support the proposed mechanism that parental beliefs about the impact of early investments are a key driver of parental behavior and child outcomes. Both RCTs demonstrate beliefs are malleable; however, only the more intensive, feedback- and coaching-based home visiting program consistently translated belief changes into improved parental investments and broader child outcomes (vocabulary, math, social-emotional skills). This suggests that information alone (brief clinic videos) can shift beliefs but may be insufficient to produce durable behavioral change and downstream child benefits without ongoing practice, tailored feedback, and goal-setting. The results imply that interventions aiming to reduce early SES disparities should address informational frictions and support parents in applying concepts during everyday routines. At the same time, the authors acknowledge that belief-focused strategies cannot by themselves overcome deeper structural inequalities that contribute to persistent gaps; rather, they represent a complementary pathway to enhance school readiness among low-SES families.

This study advances an economic framework linking parental beliefs, investments, and child outcomes and provides experimental evidence that shifting beliefs is feasible and policy-relevant. A brief, scalable clinic-based video program reliably increased beliefs but did not yield sustained changes in investments or child outcomes, whereas a more intensive, coached home visiting program improved beliefs, parental interactions, and multiple school readiness skills months after program completion. These results highlight the importance of intervention intensity and practice-based feedback for translating belief changes into behavior and outcomes. Future research should identify the minimal effective “dose” and components (e.g., feedback, goal-setting) needed to sustain behavioral change; test durability and scalability across diverse populations and settings; unpack mechanisms linking belief changes to different domains (language vs. math vs. socio-emotional); and explore integration with broader policies that address structural determinants of inequality.

- Potential experimenter demand/Hawthorne effects: Although mitigated by persistent impacts, control conditions (placebo videos/nutrition packets), and assessment protocols, residual concerns cannot be fully ruled out. - Generalizability: Samples are low-SES families in Chicagoland; the HV sample is exclusively Spanish-speaking households, which may limit external validity across regions and populations. - Attrition and minor imbalances: Differential attrition over time and some baseline imbalances (e.g., employment status) could bias estimates despite ITT analyses and robustness checks. - Compliance variability: Not all treated participants fully adhered (e.g., some Newborn participants watched <4 videos; ~24% of HV participants had <6 visits), potentially attenuating ITT effects and complicating dose-response inference. - Measurement differences: Different instruments across studies/time points (e.g., CDI vs. ROWPVT/PLS; varying interaction measures) may affect comparability of effects. - Mechanism identification: While belief changes co-move with investments, the exact contribution of beliefs versus other active ingredients (coaching, feedback, goal-setting) remains to be disentangled.