The Diffusion of Innovations: A Methodological Reappraisal
批评传统扩散研究依赖特定分布（如逻辑分布）导致可比性和解释问题，提出用基尼期望平均差衡量扩散速度，并给出非参数估计方法，以CT扫描仪扩散为例说明。
Studies of diffusion have traditionally relied on specific distributions, primarily the logistic, to characterize and estimate those processes. We argue that such an approach gives rise to serious problems of comparability and interpretation and may result in large biases in the estimates of the parameters of interest. We propose instead the Gini's expected mean difference as a measure of diffusion speed. We discuss its advantages over the traditional approach, present a nonparametric estimation procedure, and tackle with it the problems of truncated processes and of intergroup comparisons. We also elaborate on the use of the hazard rate and suggest various extensions. The diffusion of computed tomography scanners is presented as an illustration.