A Poisson-Gamma model for speed tests
A Poisson-Gamma model for speed tests
复制标题
用于速度测试的泊松伽玛模型
DOI:
--
复制
发表时间:
2009
期刊:
影响因子:
--
通讯作者:
F. Kamphuis
中科院分区:
文献类型:
--
作者:
N. Verhelst;F. Kamphuis
The present report discusses two applications of the Poisson measurement model for counts as originally developed by Rasch. To account for the distribution of the latent variable, a gamma distribution is hypothesized. Parameter estimation for the measurement model and for the distribution model are discussed in detail. In the rst application a collective test for technical reading the model ts the observed data very well. In the second application an individual test for technical readingpoor t was found. Extending the distribution model to a mixture of two gamma variables leads to an excellent t. Special attention is given to the estimation of the reliability. Key words: Poisson model, gamma distribution, latent class models, reliability 1 Introduction The main purpose of this report is to present an investigation on a psychometric model for pure speed tests. Speed tests can be categorized in two main types. In the rst type, the testee is performing a given task, and the basic observation is the time needed to nish the task. The second type consists of performing an, in principle innite, series of subtasks, and the basic outcome is the number of subtasks nished within a given time period. The We are indebted to Niels Veldhuijzen for his careful reading of the manuscript and the 106 suggestions for improving it.