Composing Group-Level Constructs From Individual-Level Survey Data
提出一个方法论框架，用于区分和评估由不同组合方法（直接共识与参照转移共识）构建的群体构念的心理测量质量，并以群体工作设计为例说明，对多层次理论构建和量表开发有参考价值。
Group-level constructs are often derived from individual-level data. This procedure requires a composition model, specifying how the lower level data can be combined to compose the higher level construct. Two common composition methods are direct consensus composition, where items refer to the individual, and referent-shift consensus composition, where items refer to the group. The use and selection of composition methods is subject to a number of problems, calling for more systematic work on the empirical properties of and distinction between constructs composed by different methods. To facilitate and encourage such work, the authors present a methodological framework for addressing the distinction between and the baseline psychometric quality of composed group constructs, illustrated by an empirical example in the group job-design domain. The framework primarily represents a developmental tool with applications in multilevel theory building and scale construction, but also in meta-analysis or secondary analysis, and more general, the validation of group constructs.