線性結構模式在國民小學學生自然科學習因素之分析與應用

The Application of Linear Structure Model in Analyzing Primary School Students’ Science Achievement

本論文旨在依據相關文獻及理論，提出國小學生認知型式、空力、玩玩具行為與自然科成績之結構模式分析。線性結構關係（Linear Structure Relation）簡稱Lisrel，係由Jöreskog發展而成，近年廣為一般社會為及行為科學研究所使用。它結合了迴歸模型（regression model）和因素分析模型（factor analysis model）兩個理論，適用於驗證性因素分析（confirmatory factor analysis）的情境，且能克服常用的徑路分析（path analysis）所受的限制。例如，Lisrel能分析交互因果（reciprocal causation）現象，而Lisrel配合SPSS統計套裝程式，則更能解決較複雜變項的情境。\r另一方面，本篇文章就國小學童認知型式、空間能力、玩玩具行為與自然科學成績之四個變項，探討其理論基礎及相關研究文獻，提出此四個變的Lisrel分析模型，以供參考，分析結果則因受篇幅所限，將在另文中討論。

The aim of this paper is to propose the Linear Structure Relation analysis model of the cognitive style、spatial ability、toy playing and science achievement， by reviewing the related literature and theories.“Linear Structure Relation”， which is briefly named as Lisrel， was developed by Jöreskog. It has been widely used by social and behavioral science researchers recently. It has combined the theories of regression model and factor analysis model， and is suitable for use in the situation of “confirmatory factor analysis”. Above all， Lisrel overcomes the limitation of path analysis. For example， it can analyze the condition of reciprocal causation， which can not be resolved by path analysis. By matching with SPSS statistical computer program， it can resolve the condition of more complex variables.On the other hand， this paper will also investigate four variables：cognitive style、spatial ability、toy playing and science achievement. By this way， we raise a Lisrel model of the four variables aforementioned. And this is only to give reference， the results of statistical analysis will be described in another paper.