摘要
管理学研究实证化趋势日益明显,常规的普通调查存在着诸多弊端。管理学研究中应大力发展基于准实验设计的现场研究(Quasi-Experimentation Design based Field Research,简称QEDFR),这将有效提高研究的整体可靠性、内在效度和外部效度。QEDFR的概念源于实验室研究,它的创新价值、设计思路、研究案例、实施因素、统计功效都与一般的实证研究有较大差别。严谨的实施过程和完整的设计体系是QEDFR广泛应用于管理学研究的生命力所在。
The article introduces the Quasi-Experimentation Design based Field Research （QEDFR）from the concept, innovation, designing case, implementation and statistic power for the sake of promoting local research practices and development in the future. Management research in china still rests on the large scale survey, which are confronted by many limitations, including randomized sampling challenge, accurate measurement challenge and causality challenge. In the condition of more and more complex management phenomenon, QEDFR has to be strengthened urgently. Manipulation and control are the two fundamental requirements for conducting experimentation. QEDFR not only embodies those advantages, but also characterizes the integration of low subjects randomization and field observation.
The innovation of QEDFR comprises of 8 aspects, including independent variable manipulation, irrelevant variable control, large range of research interests, long-term sustainable effects, moderate total costs, definite causal relationship, high probability of represented reality,and low subjects randomization. These features reflect the relative advantages to the survey, lab, etc.
QEDFR has two kinds of research designs. The most common design of QEDFR is nonequivalent control group design. The main principle is the simultaneous application of experimented group and comparative group. Through the manipulation of independent variable by researchers, the variance can be observed accordingly. This approach must conduct pre-test and post-test within both groups, so as to confirm that the variance is caused only by the independent variables. Another approach is the interpreted time series design, which requires the dependent variable to be tested in multiple times during the observations. The demand of frequent tests could exactly demonstrate the effect and its time-lasting degree influenced by the independent variable which often relies on the monthly performance appraisal. A case about the relationship of payment styles and individual job performance is illustrated to specify the application of the above designs.
The essay offers two points for the QEDFR implementation as well as the statistical power. Firstly, situational factors including four kinds of problems are the most important basis while conducting QEDFR. They are underestimation of controlling exogenous variables, the change or destroy of real situation, the limited generalization of the specific situation, the researcher＇s authority imposed on the subjects. Secondly, subjects and related factors are the activated variance which cannot be neglected by the researchers. These factors have four dimensions, including preventing the subjects＇ treatment prescience, restricting subjects＇ interaction, focusing on the individual difference and the randomization difficulty. The preferred sample size of QEDFR ranges between 30-150. If the minimum number is lower than 30, the probability of category Ⅱ, statistical error will increase. Meanwhile, if the maximum number is higher than 150, the probability of category Ⅰ, statistical error will be enhanced.
出处
《浙江大学学报（人文社会科学版）》
CSSCI
2007年第6期73-80,共8页
Journal of Zhejiang University:Humanities and Social Sciences
基金
教育部重大课题资助项目(06JZD0014)
国家自然科学基金青年资助项目(70702020)
关键词
普通调查
准实验设计
现场研究
内在效度
外部效度
统计功效
survey
quasi-experimentation design
field research
internal validity
external validity
statistical power
