Methods for Quantifying Patterns of Dynamic Interactions in Dyads
Methods for Quantifying Patterns of Dynamic Interactions in Dyads
复制标题
DOI：
10.1177/1073191116641508
复制
发表时间：
2016-08-01
期刊：
影响因子：
3.8
通讯作者：
Liu, Siwei
中科院分区：
文献类型：
--
作者：
Gates, Kathleen M.;Liu, Siwei
临床和研究环境中的个人越来越多地提供多个时间点的数据。例如，从可穿戴技术(如加速计)收集的测量、心理生理测量、编码观察、社交媒体行为和日常日记数据。当每个人都有许多观测数据时，这些数据属于时间序列数据类别，可以从动态系统的角度进行检查。我们提供了一个广泛的概述目前的分析方法，以量化之间的关系使用这种数据类型。这些技术包括来自线性建模框架内的技术，包括测量模型的方法，以及检查周期关系的方法。我们还讨论了一些特殊的主题，例如允许模型随时间变化和适应不同个体的异质性的方法。最后，描述了在时间上处理相似形状的非线性曲线的方法。从这些选项中，我们希望帮助指导从业者、临床医生和研究人员为他们的数据和问题行选择最佳方法。为了进一步帮助这一选择，我们指出了每种技术可用的程序。这里提供的例子二元组的范围从母婴、病人-照顾者和丈夫-妻子；然而，分析方法可以应用于任何类型的二元组。
Individuals in both clinical and research settings increasingly provide data across numerous time points. Examples include measurements collected from wearable technology (e.g., accelerometers), psychophysiological measures, coded observations, social media behaviors, and daily diary data. When numerous observations are available for each individual, the data fall under the class of time series data and can be examined from within a dynamic systems perspective. We provide a broad overview of current analytic methods for quantifying relations among dyads using this data type. The techniques include those from within a linear modeling framework, approaches that include a measurement model, and methods for examining cyclical relations. We also discuss some special topics, such as methods that allow for models to shift across time and that accommodate heterogeneity across individuals. Finally, methods that account for similar shapes of nonlinear curves across time are described. From this breadth of options, we hope to help guide practitioners, clinicians, and researchers in choosing the optimal method for their data and line of questions. To further aid in this choice, we indicate programs available for each technique. Example dyads presented here range from mother-infant, patient-caretaker, and husband-wife; however, the analytic methods can be applied to any type of dyad.