Marginal maximum likelihood estimation of item parameters: Application of an EM algorithm

Maximum likelihood estimation of item parameters in the marginal distribution, integrating over the distribution of ability, becomes practical when computing procedures based on an EM algorithm are used. By characterizing the ability distribution empirically, arbitrary assumptions about its form are avoided. The Em procedure is shown to apply to general item-response models lacking simple sufficient statistics for ability. This includes models with more than one latent dimension.Supported in part by NSF grant BNS 7912417 to the University of Chicago and by SSRC (UK) grant HR6132 to the University of Lancaster.We are indebted to Mark Reiser and Robert Gibbons for computer programming. David Thissen clarified a number of points in an earlier draft.     

汉语句子格角色分配的一种神经网络方法

提出了一个基于分布式表征的计算模型,通过并行分布加工方式完成六类汉语句子的格角色分配任务。模型是一个四层的前传网络,包括输入层（词的分布式表征层）,两个隐层,输出层（格角色层）;其中第一隐层的一部分反馈到输入层。模型采用误差反传算法,通过提供学习样本和目标输出,不断调整三个权值矩阵,使得网络稳定时能得到正确的结果。经过训练后的网络具有一定的稳定性和鲁棒性。还对这种方法与传统的符号处理方法作了比较和分析。     

THE RELIGIOUS DIMENSIONS OF THE BIOLOGICAL NARRATIVE

Abstract. A cell/molecular biologist challenges the thesis that science and religion are two ways of experiencing and interpreting the world and explores instead the possible ways that the modern biological worldview might serve as a resource for religious perspectives. Three concepts—meaning, valuation, and purpose—are argued to be central to the entire biological enterprise, and the continuation of this enterprise is regarded as a sacred religious trust.     

Management Selection in Australia: A Comparison with British and French Findings

The present study provided data on the cross-national use of management selection methods by collecting information from the Australian context. The data is compared with the British and French findings obtained by Shackleton and Newell (1991). A total of 254 companies (53.4%) responded to the survey. The overall pattern of results indicated that Australian management selection methods were closer to the British than to the French findings. Compared to previous Australian studies, the present data indicated that the main selection tools being used are unchanged. Some evidence was found that selection methods may differ within organizations comprising the private sector and between the private and public sector. The business sector made significantly greater use of cognitive testing and assessment centres compared to manufacturing, retail and government sectors. The government sector reported a significant difference in using two or three interviewers. It is proposed that, although surveys of this type have been useful as general reviews of the area, future research would benefit by collecting: (1) data on the specific measures used, (2) distinguishing between internal and external recruitment, (3) distinguishing between private and public sector recruitment, and (4) ensuring respondents have the same conceptual understanding of the selection methods covered by the survey.     

European Management Selection Methods: A Comparison of Five Countries

There is some disagreement in the recent literature on how similar or different are the methods used to select employees in different European countries. The confusion comes about in part because different samples and questions have been used in different countries to investigate this issue, making comparison between countries very difficult. This study investigates managerial selection methods in three members of the European Community, Belgium, Germany and Italy, using the same questionnaire and sample characteristics previously used in Britain and France. This allows a direct comparison between the five countries. The samples comprised 250 companies randomly drawn from the top 1,000 in each country. Questionnaires sent to the companies asked a range of questions concerning the frequency of use of selection methods and attitudes towards their use. Results show major differences in frequency of use of different methods. Some of these differences are: British and German companies tend to use assessment centres much more often than other countries, while Germany and Italy are relatively infrequent users of psychological tests. Companies in the Flemish (Dutch speaking) part of Belgium are the most likely to use biodata, while their French speaking compatriots are similar to the French in their liking for graphology. Both Belgium and France make much less use of references than do Britain, Germany and Italy. Results suggest that harmonization of selection practice in Europe is a long way off. Habit, tradition and culture determine the choice of selection method much more than do the relative predictive validities of the techniques.     

Equipercentile equating via data-imputation techniques

In the design of common-item equating, two groups of examinees are administered separate test forms, and each test form contains a common subset of items. We consider test equating under this situation as an incomplete data problem—that is, examinees have observed scores on one test form and missing scores on the other. Through the use of statistical data-imputation techniques, the missing scores can be replaced by reasonable estimates, and consequently the forms may be directly equated as if both forms were administered to both groups. In this paper we discuss different data-imputation techniques that are useful for equipercentile equating; we also use empirical data to evaluate the accuracy of these techniques as compared with chained equipercentile equating.A paper presented at the European Meeting of the Psychometric Society, Barcelona, Spain, July, 1993.     

Validity and Application of Computer-simulated Scenarios in Personnel Assessment

In Germany computer-simulated scenarios have been applied in personnel management for several years. Their goals are either to assess how potential managers cope with complex and connected problem states (job aptitude testing aspect) or to provide feedback on manager strengths and weaknesses in dealing with complexity, and how to modify these when necessary (training aspect). This paper presents and evaluates those scenarios that have been subjected to validity studies.     

Burhoe's Second-Hand Influence

Many of us not part of the "old Burhoe gang" are nonetheless deeply influenced by the ideas of Ralph Wendell Burhoe, albeit in indirect ways. This remembrance summarizes six such ways: Three are "procedural" influences, namely (1) that dialogue is most valuable, especially in the science/religion interface, when carried on among those who may not agree; (2) that scholarship is necessary to refine and improve preliminary opinions; and (3) that organizations are crucial to accomplishing the first two tasks. The three "substantive" influences are (4) Burhoe's focus on human values; (5) his work in defining God; and (6) his contribution to defining what it means to be human. As is well known, his emphasis in all three substantive cases was on the power and nuances of biological and social evolution, especially on the dynamics of natural selection.     

A Bayesian approach to nonlinear latent variable models using the Gibbs sampler and the metropolis-hastings algorithm

Nonlinear latent variable models are specified that include quadratic forms and interactions of latent regressor variables as special cases. To estimate the parameters, the models are put in a Bayesian framework with conjugate priors for the parameters. The posterior distributions of the parameters and the latent variables are estimated using Markov chain Monte Carlo methods such as the Gibbs sampler and the Metropolis-Hastings algorithm. The proposed estimation methods are illustrated by two simulation studies and by the estimation of a non-linear model for the dependence of performance on task complexity and goal specificity using empirical data.