根據最近教育研究趨勢的考察，筆者認為不宜排斥調查研究、相關研究、及其他 研究，而把研究設計窄化為實驗設計而已。有鑒於此，本文秉持寬廣的概念及彈性的作法， 擬針對五項重點，而予以探討。茲分述於下： （一）應用推論效度，以討論四類教育研究，即：敘述性實驗室研究、敘述性田野研究、實 驗性實驗室研究、以及實驗性田野研究。 （二）討論設計因素與設計類型的交互作用；這些因素包括：處理因素、試驗因素、區組因 素、組別因素、以及單元因素。 （三）檢討基本實驗設計與因素性設計，而分析此兩種類型不同模式與實例。 （四）分析選取良好研究設計的考量因素， 其重要者包括：（ 1 ）與研究問題切合的現有 知識基礎；（ 2 ）相關的研究設計的類型與推論性質；（ 3 ）研究者所掌握的資源；（ 4 ）不利於研究設計的效度之干擾因素；以及（ 5 ）複雜設計的適用性。 （五）綜合探討所得，提供考量教育研究設計的一些具體要領。

　　　　　From a survey of recent educational research, there is no point to reject the legitimacy of research designs in survey studies, correlational studies and other relevant studies. Experimental designs should not be considered as the only class of good designs in educational studies. Based upon broadened concepts and flexible practices used by a number of researchers in education, this paper is intended to conduct five principal tasks for research designs in education as follows: 1. Inferential validities are used to discuss four types of educational studies: descriptive laboratory studies, descriptive field studies, experimental laboratory studies, and experimental field studies. 2. Design factors are discussed in their interactions with design types. These factors are treatment factors, trial factors, blocking factors, group factors and unit factors. 3. True experimental designs and factorial designs are reviewed for their models and examples. 4. Some considerations are suggested for choosing a good research design. Among them are: (1) existing knowledge bases relevant to research questions; (2) types of research designs used and inferences made; (3) resources available to researchers; (4) specific threats to the validity of research design, and (5) the utility of complex designs. 5. Specific guidelines are recommended to consider the problems of selecting adequate designs in educational research.