| Summary: | 碩士 === 淡江大學 === 資訊管理研究所 === 84 ===
This study first attempts to investigate the effects of data models and training degrees on data representations produced by end users. Two different data models (i.e., extended entity relationship medel vs. relational model) and two different degree of training courses (i.e., high degree vs. low degree) are used to explore the similarities and differences in the quality of end users' data representations. The quality are evaluated through five constructs (i.e., entity/relation, relationship, identifier, descriptor, and category) and six facets of relationship (i.e., unary one-to-one, unary one-to-many, binary one-to-one, binary one-to-many, binary many-to-many, and ternary many-to-many-to-many) in a data model. Results indicate that data representations' quality significantly exists differences in the construct of category, relationship, binary one-to-many, and binary many-to-many between two data models. there are dissimilarities appearing in constructs of data model between two different training degrees except entity/relation, descriptor, unary relationship, binary one-to-one relationship, and ternary relationship. No differences of users' performance exist in the interactin between data medel and training degree.
The second purpose of the study is to investigate the impact of training degrees on end users' self-efficacy and the relationship between users' self-efficacy and their performance of data representations. The low degree and high degree training courses are used to explore the changes in end users' confidence. The statistical relationship between users' belief and their performance is then tested through the analysis of correlation. A set of computer self-efficacy questions are revised to measure end users' perceptions of their ability on database scheme design. As compared to the low degree of training course, the high degree of training course does not imposingly improve users' confidence of their ability. The result also depict that no significant relationship exists between users' self-efficacy and their performance of data representations.
|
|---|
