MARKERS AND MODIFICATION OF EARLY READING FAILURE

描述（由申请人提供）：本研究的目的是提高对阅读失败的早期识别筛选的准确性，并确定二级预防在多大程度上改变了其概率。在一个高度识字的社会中，阅读失败的情感和经济后遗症是一个公共卫生问题。尽管全国对阅读失败的问题给予了前所未有的关注，但一个准确识别谁会出现早期阅读失败，谁不会出现早期阅读失败的模型并不存在。潜在的标记变量数量众多，但尚未协调成一个有凝聚力的系统，该系统在理论上是稳健的，并经过系统测试。此外，研究性阅读干预的持续时间及其对高危儿童的长期影响也很少被研究。两个研究问题旨在解决过去研究的局限性：(a)是否通过纳入流畅性和增长措施提高了早期识别筛选的准确性；(b)早期识别屏幕定义的阅读失败概率是否与教学时间相互作用，从而立即和长期地改变阅读失败的风险？该设计的一个新颖方面是将连续读取失败的风险定义为概率水平，而不是二分类。研究计划采用多队列设计，其中选择三个连续的一年级儿童队列（N=250-300年）。在一年级，测试早期识别屏幕的措施包括准确性和流畅性措施，收集作为点估计（静态测量）和跨越时间（增长测量）。我们假设缺乏对流利性和成长的关注阻碍了过去早期识别的努力。采用多元回归和优势分析来评估对连续测量读数的影响，并减少变量集。采用逻辑回归的方法，建立最佳拟合算法，预测读取失败的概率。该算法应用于二年级和三年级，以识别失败概率较高的儿童（N=180）。这些儿童被分配到无治疗控制组或干预组，持续10周或16周。持续时间的变化是为了了解阅读干预研究中通常使用的干预时间的收获和损失。所有三个队列都被诊断组跟踪到三年级。

DESCRIPTION (provided by applicant): The purpose of this investigation is to improve the precision of early identification screening for reading failure and to determine the extent to which secondary prevention modifies its probability. The emotional and economic sequelae of reading failure in a highly literate society represent a public health concern. Despite the unprecedented national attention given to the problem of reading failure, a model that accurately identifies who will and who will not exhibit early reading failure does not exist. The number of potential marker variables are legion but have yet to be coordinated into a cohesive system that is theoretically robust and systematically tested. In addition, the duration of research-based reading intervention and its long term effects on at-risk children are rarely studied. Two research questions are designed to address the limitations of past research: (a) Is the precision of early identification screening improved by incorporating fluency and growth measures; and (b) Does the probability of reading failure defined by the early identification screen interact with instructional duration to modify the risk of reading failure both immediately and longitudinally? A novel aspect of the design is defining risk of reading failure continuously as probability level rather than dichotomously. The research plan incorporates a multiple cohort design in which three successive cohorts of first grade children are selected (N=250-300 across years). In Year 1, measures tested for the early identification screen include accuracy and fluency measures collected as point estimates (static measurement) and across time (growth measurement). We hypothesize that the lack of attention to fluency and growth has hampered past early identification efforts. Multiple regression and dominance analysis are used to assess impact on reading measured continuously and to reduce the variable set. Logistic regression is used to develop the best fitting algorithm to predict reading failure and produce probabilities. This algorithm is applied in Years 2 and 3 to identify children with higher probabilities of failure (N=180). These children are assigned to a no treatment control or to intervention that lasts 10 weeks or 16 weeks. Duration is varied to understand what is gained and lost by the intervention lengths typically used in reading intervention research. All three cohorts are followed through third grade with the diagnostic battery.