Replication Package for The Unequal Effects of Weekday Television on Children’s Academic and Behavioral Skills

This study examines whether children’s weekday television exposure causally affects cognitive and behavioral development, and whether these effects vary across the outcome distribution and socioeconomic status. The central hypothesis is that increased weekday television viewing crowds out cognitively productive activities and parental interaction, leading to weaker academic performance and elevated behavioral problems, particularly at vulnerable points of the outcome distribution. To address endogeneity in television exposure, the analysis uses instrumental variables and IV quantile regression to identify distributional effects rather than average effects alone.

The data are drawn from a nationally representative longitudinal child level dataset that links children’s cognitive test scores, behavioral assessments, family characteristics, and time use measures. Outcomes include standardized reading and mathematics assessments and a validated behavioral problems index. Television exposure is measured as weekday viewing time reported by caregivers, and instruments exploit plausibly exogenous variation in family structure and parental constraints. The dataset preserves the panel structure, identifier consistency, and missing value patterns necessary for causal estimation.

The empirical results show that increased weekday television exposure has heterogeneous effects across the outcome distribution. Negative effects are generally larger at lower quantiles of cognitive outcomes and for children from lower socioeconomic backgrounds, while behavioral outcomes exhibit stronger adverse responses at higher quantiles of problem behavior. These patterns suggest that television exposure exacerbates existing developmental disparities rather than affecting all children uniformly. Robustness checks using alternative instruments, subgroup analyses, and first stage diagnostics support the stability of the findings.

The data and replication code are intended to be used for causal inference on child development outcomes, with careful attention to sample construction, instrument validity, and distributional heterogeneity. Users should interpret coefficients as local quantile treatment effects identified for compliers, rather than average treatment effects, and should account for outcome specific sample restrictions implied by the estimation strategy.

本研究旨在探究儿童平日电视暴露时长是否会对其认知与行为发展产生因果性影响，以及此类影响是否会随结果变量分布和社会经济地位的不同而呈现异质性变化。核心假说为：平日看电视时长增加会挤出认知有益活动与亲子互动时间，进而导致学业表现下滑、行为问题增多，尤其在结果分布的脆弱区间该效应更为显著。为解决电视暴露程度的内生性问题，本研究采用工具变量（instrumental variables）与工具变量分位数回归（IV quantile regression）方法，以识别分布异质性效应而非仅关注平均效应。

本研究数据来自一项具有全国代表性的儿童层面纵向数据集，该数据集整合了儿童的认知测验得分、行为评估结果、家庭特征与时间使用情况。结果变量包括标准化阅读与数学测验成绩，以及经过验证的行为问题指数。电视暴露程度以看护人报告的平日看电视时长衡量，工具变量则利用家庭结构与家长约束中似外生的变异进行构建。该数据集保留了因果推断所需的面板结构、标识符一致性与缺失值分布特征。

实证结果表明，平日看电视时长增加对结果变量分布存在异质性影响。认知结果的低分位数区间以及社会经济地位较低的儿童所受负面影响更为显著；而行为问题指数的高分位数区间则表现出更强的不良响应。上述结果提示，电视暴露会加剧既有的发展差距，而非对所有儿童产生同质化影响。采用替代工具变量、亚组分析与第一阶段诊断检验进行的稳健性检验，证实了研究发现的稳定性。

本数据集与复现代码仅用于儿童发展结果的因果推断研究，使用时需严格关注样本构建、工具变量有效性与分布异质性问题。使用者应将回归系数解读为针对依从者识别的局部分位数处理效应，而非平均处理效应，同时需考虑估计策略所隐含的针对特定结果变量的样本限制条件。