GROWTH CURVE ANALYSIS IN ACCELERATED LONGITUDINAL DESIGNS
GROWTH CURVE ANALYSIS IN ACCELERATED LONGITUDINAL DESIGNS
复制标题
DOI：
10.1177/0022427892029004001
复制
发表时间：
1992-11-01
影响因子：
3.2
通讯作者：
CHAN, WS
中科院分区：
文献类型：
--
作者：
RAUDENBUSH, SW;CHAN, WS
加速纵向设计使研究人员能够通过在相对较短的时间间隔内收集数据来研究生命过程中较长时间间隔内的个体发展。这样的设计也创造了标准面板设计无法提供的可能性，将发展效应与群组和时期效应分开。然而，这些设计给研究人员带来了一系列特殊的推理挑战，并给统计分析带来了复杂性。在这篇文章中，作者采用分层线性模型来说明应用生长曲线分析的数据从加速纵向设计。其目标是通过将全国青年调查的两个队列的数据联系起来，构建一幅11至18岁对越轨行为的态度发展的图景，每个队列只观察了5年。该示例说明了分析师如何控制时变和时不变协变量，并检验队列效应和队列与年龄的相互作用。有趣的成长特征包括一个拐点（在这个年龄，越轨态度的增长速度开始放缓）和一个高峰年龄（最大越轨态度的年龄）。
Accelerated longitudinal designs enable researchers to study individual development over a long interval of the life course by gathering data during a comparatively short interval of time. Such designs also create possibilities not available in standard panel designs for separating developmental effects from cohort and period effects. However, these designs confront the investigator with a special set of inferential challenges and introduce complexity into statistical analysis. In this article the authors employ a hierarchical linear model to illustrate the application of growth curve analysis to data from an accelerated longitudinal design. The goal is to construct a picture of the development of attitudes toward deviance from ages 11 to 18 by linking data from two cohorts of the National Youth Survey, each observed for only 5 years. The example illustrates how the analyst may control for time-varying and time-invariant covariates and test for cohort effects and cohort-by-age interactions. Interesting features of growth include an inflection point (age at which the rate of increase in prodeviant attitude begins to slow down) and a peak age (age of maximally prodeviant attitude).