【作者】 谢小云；

【导师】 王重鸣；

【作者基本信息】 浙江大学 ， 企业管理， 2005， 博士

【摘要】 人力资源测评与选拔经过将近一个世纪的发展,已经在人力资源管理实践中发挥了重要的作用。早期的测评与选拔研究大都聚焦于预测指标的开发和构思验证,并在此基础上作出人事决策;这种过于关注选拔效用的倾向导致测评研究中缺乏良好的效度证据,难以准确的对选拔的质量和有效性进行评估。随着效度理论和效度概化研究的发展,理论界开始重视对测评工具的效度验证;但是,由于对效标内涵的理解尚未取得充分的进展,效度验证的结果呈现出明显的情景特定性;即对测评与选拔的有效性判断缺乏相对一致的结论。因此,研究者关注的焦点正逐步转向效标的开发和构思验证,希望藉此增进对效标的理解,从而提升测评与选拔的效度。 在以往研究的基础上,本研究以提升效度和效度概化水平为目标,运用半结构化访谈、问卷测量和现场研究等方法,借鉴多元分析和结构方程建模技术的进展:研究了效标测量的构思模型、效标测量过程机制、效标的动态属性、效标选择以及效标的多水平构思框架等问题。 在本文的第三部分中,主要内容是效标构思建模和客观效标构思探索。针对一批进行任职资格认证的企业管理人员样本,开展相关的访谈和测量;在绩效理论进展的基础上,提出多个效标测量假设模型,并采用层级验证性因素分析的方法对效标所测量到的构思进行验证。统计分析的结果表明效标并不是以往研究中的绩效维度模型或者单纯的方法模型,而是多方法多维度的综合模型;这个结论为人事测评中的情景特定性现象提供了多种解释的备择方案。此外,我们还借鉴诺莫网络的思想对客观效标的构思进行了初步的探讨;研究结果显示,经历等客观效标与行为效标之间只有较低的相关,说明它们测量的是不同的构思;在本研究中,晋升经历更多的与任务绩效相关,而资历和荣誉奖励经历等效标与周边绩效之间有更高的相关。 第四部分的主要内容是对效标测量的过程机制进行探讨,以此作为对第三部分效标构思模型中的方法因素进行解释。以一批参加真实公开招聘的候选人作为样本,开展现场研究和问卷测量工作。因为在前人研究中对于上级评价效标的过程模型已经得到了较为充分的证据,所以本研究聚焦于对自我评价效标和上级评更多还原

【Abstract】 Studies on personnel selection and assessment have over almost one century of scientific history. Nowadays, they were widely used in human resource management and played a crucial role. In the early period, the most studies focused on the utility of selection, which paid more attentions to development of predictor domains, and made administrative decision based on scores of predictors. In 1970s, In order to justify the outcomes from selection and assessment, the evidence about quality and effectiveness of selection which named validation was advocated intensively. Although this research was extensive in 1980s, it’s far from conclusive. Many predictors were proved to be valid in some extent with differential criterias, but the results from validity generalizability procedure were pessimistic. The most validity coefficients were unstable, which has known as situational specific in personnel selection and assessment. According to the scheme of validation, researchers in industrial and organizational psychology recognized they need more understandings on criteria, and shifted their focus from predictors to criteria domains. In particular, they placed the emphasis on criteria development and construct validation, and hoped to receive an improvement of validity evidence.Based on the outcomes from literature review, the thesis aimed to enhance the validity generalizability of selection. By using semi-structural interview, questionnaire survey and field study, the present study, with five sub-studies, attempted to develop and validate the criteria model, examine the procedural mechanism of criteria measurement, dynamic criteria, analysis the predictors-criteria match, and examine the multilevel framework in criteria nature. More details about result were shown as below.In ChapterIII, we focused on development and validation of criteria model, and exploratory analysis on construct of objective criteria using a managerial sample. Based on the foregoing progress about performance theory, we put forward several hypothesis criteria models and employed a validation by Hierarchical ConfirmatoryFactor Aanalysis method. The result was interesting and differential from the foregoing outcomes, the multimethods-multidimensions combinating model received the promising support when we compared several fit indices, and it gave us more understanding about criteria and supplied several possible interpretations to the situational specific. Further, we developed an exploratory analysis of construct validity of objective criterias by nomological network. The results showed there were low-convergent relationship between behavior based criteria and objective criteria, but the promotion showed more overlap with task performance, the merit related experience and seniority showd more overlap with contextual performance.In Chapter IV, we focused on procedural mechanism of criteria measurement in personnel selection and assessment, and hoped to discover the nature of method variance in the multimethods-multidimensions model. By using a field study and questionnaire survey, we employed a competitive testing between situational specific model and cross-situatinal consisitency model, aimed to classify the procedural mechanism of self-rating criteria, and the recursive solution was compared with supervisory rating criteria. The result showed there was much differential between the two rating criteria, and particularly, the method variance was interpretated as neither bias nor error, but valid performance measurement which named uniqueness in statistical term. The result inspired us to re-evaluate the validity of personnel selection and we’d get a more optimistic opinion about "select validly". Further, the dynamic nature of criteria was examined; we classified three career stages on the time serial which named adaptive period, continuous period, and transformal period, and found that the procedural mechanisms were much differential among three stages. The result confirmed the dynamic nature in critera, and gave us some theoretical clues about criteria sel更多还原