When designing intervention research that has a longterm goal, fade-out is an important consideration. Bailey, Duncan, Cunha, Foorman, and Yeager (2020; this issue) offer several important takeaways for such interventions, beginning from the initial plan to later longitudinal analyses of treatment effects. For example, researchers would be well advised to consider the contextual influences, such as whether the treatment is in a low-income urban school district undergoing pending gentrification plans at the onset of the intervention, which could change the demographic characteristics of the targeted student population. Gentrification of a neighborhood may have profound implications for the initial sample selection, instrumentation, and measurement. The authors also suggest that intervention designers with long-term goals need to request additional support for subsequent data-collection efforts. We assume this would include such factors as obtaining overpowered initial treatment and control samples, identifying stable contextual conditions (e.g., neighborhood, student and teacher mobility), and a clear temporal vision of subsequent treatment outcomes, all of which are likely to affect the sample balance necessary for evaluating the impact of the intervention over time. Although Bailey and colleagues are comprehensive in their focus on fade-out and possible remediation of its effect, we argue that the dominance of the psychological perspective on education interventions and their purposes tend to overlook other research designs in which problems of fade-out can more easily be adjusted (e.g., quasiexperiments with generalizable longitudinal samples that include nested interventions) or other naturally occurring treatment effects (e.g., use of online instruction during a pandemic). The authors focus on interventions designed to enhance psychological traits or skill-based tools and bring in other research in economics and sociology that they perceive as complementary to their perspective. Our review highlights some additional problems of designing interventions involving randomized controlled trials (RCTs) that specifically focus on avoiding fade-out and recognize the complexity of measures required to understand persisting effects of an intervention on either psychological traits or skill-based tools. In addition, we put forward several measurement issues that arise when considering postintervention analyses for RCTs or quasiexperiments.

中文翻译：

---

我们正在了解干预效果的淡出：评论

在设计具有长期目标的干预研究时，淡出是一个重要的考虑因素。Bailey、Duncan、Cunha、Foorman 和 Yeager（2020 年；本期）为此类干预措施提供了几个重要内容，从最初的计划到后来的治疗效果纵向分析。例如，建议研究人员考虑环境影响，例如治疗是否在低收入城市学区进行，在干预开始时正在进行中产阶级化计划，这可能会改变目标学生群体的人口特征. 邻里的高档化可能对最初的样本选择、仪器和测量产生深远的影响。作者还建议，具有长期目标的干预设计者需要为后续的数据收集工作请求额外的支持。我们假设这将包括以下因素，例如获得过强的初始治疗和控制样本、确定稳定的环境条件（例如，邻里、学生和教师的流动性）以及对后续治疗结果的清晰时间愿景，所有这些都可能影响样本评估干预随着时间推移的影响所必需的平衡。尽管 Bailey 及其同事在关注淡出及其影响的可能补救方面是全面的，但我们认为，心理学观点对教育干预的主导地位及其目的往往会忽视其他研究设计，在这些研究设计中，淡出问题可能更容易调整（例如，包括嵌套干预的可推广纵向样本的准实验）或其他自然发生的治疗效果（例如，在大流行期间使用在线指导）。作者专注于旨在增强心理特征或基于技能的工具的干预措施，并引入他们认为与他们的观点相辅相成的其他经济学和社会学研究。我们的综述强调了一些涉及随机对照试验 (RCT) 的干预措施设计的额外问题，这些试验特别关注避免淡出，并认识到理解干预措施对心理特征或基于技能的工具的持续影响所需的措施的复杂性。此外，