社会与工程心理学研究室知识库

1.目的

以大学生为被试，以大学生的综合素质测评和管理者评定为效标，并通过相关分析、多元回归分析和分层逐步回归分析的方法考察五因素人格问卷、结构化面试、无领导小组讨论和罗夏墨迹技术的效标关联效度和增益效度，并考察这些常用的人事测评方法组合后的效度，探讨人事选拔的最佳模式。

2.方法

以自愿报名的形式征集了某大学教育学院2005级的68名本科学生。问卷测试采用的是五因素人格问卷NEO-FFI, 测量神经质N、外向性E、开放性0、宜人性A、责任心C共S个维度。结构化面试设计了9个测评维度，分别是礼仪风度，言语表达能力，逻辑分析能力，计划组织能力，协调能力，建立伙伴关系，自我管理能力，责任心，应变能力。无领导小组讨论中设计了言语表达、逻辑分析、计划组织、协调、建立伙伴关系、说服6个测评维度，前S个维度与辅导员评定和结构化面试完全相同。

上级管理者评定采用的是被试所在班级的专职辅导员和学生干部对所有被试在9个面试维度上的评定。评定时依据的是操行、在班级中的表现、与同学关系、对集体活动的参与和组织等。综合测评则包括智育(占65“%)、德育(占25%)和体育成绩(占10% )三部分组成。

3.结果

(1)五因素人格问卷中的责任感、神经质和宜人性维度与管理者评定和综合测评都出现了有意义的相关，证明这三个维度都在一定程度上测量到了与适应性行为有关的心理品质。外向性和开放性两个维度的预测效度则较差。说明人事测评中可以采用问卷法。但问卷法只有在单独使用时才有一定的预测效度，如果与结构化面试和罗夏同时使用，五因素人格问卷的增益效度是很低的。因此在人事测评中如果使用了结构化面试或罗夏技术，则没有必要使用五因素人格问卷。

(2)无领导小组讨论的各个维度及总分与管理者评定的各个维度和总分间多出现了正相关，与德育和综合测评的总分也出现了显著正相关。说明无领导小组讨论也有一定的预测效度，但效度不是太高。虽然单独以无领导小组讨论为预测变量，对管理者评定和综合测评有一定预测力，但在与结构化面试和罗夏一起作为预测源时，无领导小组讨论只能提供很小的增益效度。

(3)以管理者评定和综合测评为效标，情景性面试与行为性面试的效标关联效度都较好，相比之下行为性面试的效度略高些，但不明显。两类面试的构想效度都基本可以接受，但行为性面试的构想效度好于情景性面试。情景性面试受社会期望性即SDR的影响更大。行为性面试也受评委效应的影响，但情景性面试中评委效应对面试结果的影响更大。这些结果说明情景性和行为性面试都是有效的面试形式。只要程序设计严谨，过程控制严密，都可用作有效的人事测评工具。

以面试维度为预测变量，以管理者评定和综合测评为因变量时，回归方程都是显著的，而且有较高的解释率。在分层逐步回归分析中，结构化面试对方程的解释率较大，而且能提供较好的增益效度。说明结构化面试在人事测评中具备其他测评技术所不具备的性能，在人事测评中是不可替代的。而问卷和无领导小组讨论则不能比结构化面试提供更多信息，因此结构化面试可以代替问卷和无领导小组讨论。而罗夏技术与结构化面试则是相互补充而不是互相替代。

(4)罗夏记分与综合测评、管理者评定、结构化面试、无领导小组讨论和问卷测评中的两种或两种以上方法间存在有意义的相关，证实了这些罗夏记分在人员素质测评中的效度。说明罗夏技术确实能测量出被试信息处理的质量与效率，信息处理的动机强度，以及不同的信息处理风格;对情绪刺激的趋近与回避，及情绪的管理能力;自我知觉和自我经验的成熟性，以及人际知觉和人际交往的兴趣与能力。

以管理者评定和综合测评为因变量的多元回归分析中，罗夏都是预测能力很高的预测源，预测能力明显大于问卷和无领导小组讨论。在以上级管理者评定和综合测评为因变量的分层逐步回归中，只有罗夏和结构化面试能够互相在对方的基础上提供增益效度。说明罗夏和结构化面试在人事测评中不能相互替代。而当罗夏和结构化面试中的一个在回归方程中时，再加入问卷和无领导小组讨论都不能增加方程的解释力。说明问卷和无领导小组讨论都不能在罗夏和结构化面试的基础上提供更高的增益效度。说明在人事测评中应采用罗夏和结构化面试方法。罗夏和结构化面试可代替问卷和无领导小组讨论。

4．结论

罗夏墨迹技术、结构化面试、无领导小组讨论和问卷测评在的事测评中都有一定效度，罗夏和结构化面试的效度较高，无领导小组讨论和问卷的效度较低。罗夏和结构化面试能代替无领导小组讨论和问卷，但不能被无领导小组讨论和问卷代替，罗夏和结构化面试也不能互相代替。因此人事测评中应使用罗夏和结构化面试。

Aim

In a group of volunteered university students, Neo-Five-Factor-Inventory (NEO-FFI), Leadless Group Discussion (LGD), Structural Interview (S1) and Rorschach Ink-Blot Technique (RIM) are used as measures, Administrator Rating (AR) and Comprehensive Measures of School Performance (CMSP)are used as criterions. Through bivariate correlation, Multiple Regression and Hierarchical Stepwise Regression analysis the relationship between measures and criterions are explored. We hope to find whether the measures are valid in personnel selection.

Method

Six-eight university students in grade 3 are recruited as subjects, who were administrated all the measures mentioned above. One measure is NEO-FFI. It contains 5 personality dimensions which are Nervous Extroversion, Openness, Agreeableness, Conscientiousness. In Structural Interview 9 dimensions are designed, they are Etiquette, Verbal, Reasoning, Organization, Coordinating, Alliance, Self-Management, Responsibility, Flexibility. While In Leaderless Group Discussion 6 dimensions are designed, they are Verbal, Reasoning, Organization, Coordinating, Alliance, And Persuasion.

Administer Rating (AR) also contains 9 dimensions which are the same as the 9 dimensions in structural interview in name and contents. Ratings are based on past behaviors of the students, such as interpersonal relationship with teachers and peer students, participation in team work, efficacy of learning and daily affairs, and so on.

Comprehensive Measures of School Performance (CMSP) mainly contain two contents: one is the average academic scores of all university courses (the weight is 65%), another is the moral records of every students (the weight is 25%).

Result

1. Conscientiousness, Nervous, Agreeableness in NEO-F1FI correlate significantly with Administer Rating (AR) and Comprehensive Measures of School Performance (CMSP), indicate that Chinese edition of NEO-FFI can be used as valid predicator of overt behavior. However in Hierarchical Regression equation NEO-FFI shows little increment validity over the Rorschach and Structural Interview.

2. Most dimensions of Leaderless Group Discussion (LGD) also correlate significantly with Administer Rating (AR) and Comprehensive Measures of School Performance (CMSP)9 indicate that LGD can be used as a valid measure. But in Hierarchical Regression Equation LGD also shows little increment validity over the Rorschach and structural Interview.

3. Most dimensions of Structural Interview (SI) show modest to high correlations with Administer Rating (AR) and Comprehensive Measures of School Performance (CMSP). In Hierarchical Regression equation SI provide substantial increment validity over NEO-FFI and LCD. Shows that SI is more valid than NEO-FFI and LGD, and that NEO-FFI and LGD provide little extra utility if SI is used in personnel selection.

Two kinds of Situational Interview are also compared in our study. Situational Interview (SI) and Behavior Description Interview (BDI) all have fairly high criterion-related validity. Confirmatory factor analysis for the MTMM also attain acceptable model fit for SI and BDI, these results indicate good validity of SI and BDI in Chinese culture. SI shows more complex factor structure than BDI, in SI interviewer effect and SDR have more influence on interview scores, this suggest that in SI stricter control in program design and administration are needed.

4. Some scores of the Rorschach Ink-Blot Technique (RIM) correlated significantly with Administer Dating (AR) and Comprehensive Measures of School Performance (CMSP), these scores also show significant correlation with SI, LGD, or NE0-FFI. Since these RIM scores correlated significantly with at least two kinds of other measures, their validity in personnel selection was verified. RIM can be used as valid measure of efficacy and motivation of information process, emotion regulation, ego and interpersonal perception.

In Hierarchical Regression equation RIM provide substantial increment validity over NEO-FFI and LGD. Shows that RIM is more valid than NEO-FFI and LGD, while NEO-FFI and LGD provide little increment validity if RIM is in the regression equation. RIM and SI are more valid than NEO-FFI and LGD, if RIM and SI are used in personnel selection, there is no necessity to use NEO-FFI and LGD.

Conclusion

Rorschach Ink-Blot Technique, Structural Interview, Leadless Group Discussion, NEO-FFI are proved to be valid measures in personnel selection. The validity of RIM and SI are higher than LGD and NEO-FFI, if RIM and SI are used in personnel selection, there is no necessity to use NEO-FFI and LGD. So RIM and SI are highly recommended.