Scaling Promising Parenting Programs in Under-Resourced Communities: Evidence from a Field Experiment in Rural China

Tips-by-Text is an age-appropriate text-messaging program for parents of young children designed to strengthen parenting knowledge, daily practices, and early child development. The program delivers brief, developmentally tailored messages aligned to each child’s age in weeks, beginning at birth and progressing to more complex skills over time. The curriculum emphasizes whole-child development through simple, practical activities embedded in everyday routines, such as talking, playing, and reading.

We adapted an established U.S. text-messaging program because comparable interventions were scarce in rural China and the original model had demonstrated strong impacts on parenting behaviors. The adaptation followed a four-step process: translating messages into Mandarin, revising content with provincial maternal-and-child-health experts to ensure cultural and contextual appropriateness, piloting the messages with rural caregivers of children ages 0–3, and refining content based on caregiver feedback before full implementation.

Caregivers received three messages per week via WeChat. Monday “FACT” messages provided brief information to build understanding and motivation, Wednesday “TIP” messages offered easy, low-cost strategies integrated into daily routines, and Friday “GROWTH” messages reinforced engagement and encouragement. Each week focused on one of five domains: talking with the child, playing with the child, sleep, emotional development, or caregiver self-care.

Mothers began receiving messages shortly after childbirth and continued for an average of 84 weeks. While other household members were encouraged to participate, enrollment was overwhelmingly limited to mothers, reflecting constraints related to literacy, smartphone access, and caregiving roles. Control-group mothers received only holiday greetings and no parenting messages.

2017-10-15

2019-10-01

Maternal Level Outcomes: Parenting Knowledge, Positive and Stimulating Parenting Practices and Mental Health Status

Parenting Knowledge

Maternal parenting knowledge was measured at baseline and endline using a 17-item scale aligned with the Tips-by-Text curriculum. The scale covered five domains: talking with the child, playing with the child, child sleep, emotional development, and caregiver self-care. Each item assessed caregivers’ understanding of developmentally appropriate practices and was coded as correct or incorrect, with higher scores indicating greater parenting knowledge.

Positive and Stimulating Parenting Practices

Parental investments in child development were measured at endline using a structured caregiver questionnaire. We constructed two indices: material investments, capturing the availability and variety of play and reading materials in the home, and time investments, capturing whether mothers engaged in key play and learning activities with their child over the previous three days (e.g., reading, storytelling, singing, playing with toys).

We also measured parenting behaviors using 16 items capturing positive interactions (e.g., expressions of affection), disciplinary practices, and child media exposure. These items were combined into a composite parenting behavior score using principal component analysis.

Maternal Mental Health

Maternal mental health was assessed at baseline and endline using a standardized self-report scale measuring symptoms of depression, anxiety, and stress. Mothers rated how frequently they experienced each symptom during the past week on a four-point scale. Separate domain scores were constructed for depression, anxiety, and stress, with higher scores indicating worse mental health.

Child Development Outcomes
Caregiver Reported Early Childhood Development Instruments (CREDI)
Social-emotional section of Ages & Stages Questionnaires (ASQ:SE)

Caregiver-Reported Early Childhood Development

Child development was measured at endline using an age-appropriate caregiver-reported instrument covering five domains: overall development, cognition, language, motor skills, and social-emotional development. Caregivers responded “yes,” “no,” or “not sure” to items tailored to the child’s age group, and domain-specific scores were constructed based on age-adjusted responses.

Social-Emotional Development

Children’s social-emotional outcomes were also assessed using a caregiver questionnaire targeting social-emotional competence and problem behaviors. Caregivers reported the frequency of specific behaviors and indicated whether any were a concern. Item responses were aggregated into a total score, with higher values indicating greater social-emotional difficulties.

We collected data through baseline and endline surveys. Baseline surveys were conducted with two cohorts in October 2017 and April 2018 at provincial-level WCH facilities, where trained enumerators administered tablet-based questionnaires to eligible pregnant women. In total, 1,096 women from 210 townships completed the baseline survey across the two cohorts.

Participants were randomly assigned to treatment (n = 549) or control (n = 547), with randomization stratified by county to ensure balance. Baseline characteristics were well balanced across groups.

After completion of the baseline survey, participants were randomly assigned at the individual level to either the treatment or control group in a 1:1 ratio. Randomization was stratified by county and cohort to ensure balance across geographic areas and enrollment waves and to improve statistical power. Assignment was implemented by the research team using a computerized random number generator. Baseline balance checks indicate that treatment and control groups were comparable across observed child, household, and maternal characteristics.

individual

No

N/A

1,096 women from 210 townships

Participants were randomly assigned to treatment (n = 549) or control (n = 547)

Power calculations indicate that a sample of approximately 1,050 individuals (1:1 treatment–control) is sufficient to detect a 0.2 SD effect with 90% power at the 5% significance level.