The Empirical study on The Health Action Process Approach and The Stages of Change – Application of Logistic Regression and Neural Networks.

• Main Authors: Yu-Jen Huang, 黃宇溱
• Other Authors: Ching-Lin Wu
• Format: Others
• Language: zh-TW
• Published: 2013
• Online Access: http://ndltd.ncl.edu.tw/handle/7u3avk

碩士 === 國立中興大學 === 運動與健康管理研究所 === 101 === Purposes: The aim of this study was to investigate regular exercise behavior amount college student using the Health Action Processes Approach (HAPA) with the Trans Theoretical Model (TTM). The models were using to predict the regular exercise behavior, and to understand the behavioral intention and its mechanisms. Subsequently, we developed the possible model of the factors to influence regular exercise behavior. Method: 300 college students were randomized recruited from National Chung Hsing University who enrolled in Physical Education classes for the present study. Subjects completed self-developed structured questionnaires, which 254 valid questionnaires were used for analysis (the effective recovery rate 84.7%). The data was analyzed using Cluster analysis, one-way ANOVA, Correspondence analysis, Mediator Effect analysis, Decision Tree analysis, Logistic Regression analysis, Neural Networks, Path analysis. Results: (1) There were 135 subjects "without regular exercise" and, 119 subjects with " regular exercise". (2) The Stages of Change of exercise behavior, 37.8% of subjects were at the action stage, 33.1% at the preparation stage, 15.3% at the maintenance stage, 13% at the intention stage, and only 0.8% at the no intention stage. (3) The significant differences were found in “Health Promotion Exercise Behavioral Intention”, “Perceived Health Self-efficacy”, “Exercise Control Toward”, “Maintain Exercise Self-efficacy” , “Perceived Exercise Barriers” after Cluster analysis. (4) The “Exercise Control Toward ” was played the role of partial mediation between the “Health Promotion Exercise Behavioral Intention” and “Stages Change of Exercise Behavior”. The direct effect of “Health Promotion Exercise Behavioral Intention” on “Stages Change of Exercise Behavior” was considered at 0.647, and the indirect effect was at 0.207 mediated through “Exercise Control Toward”. (5) The model obtained the best learning result at 0.75 learning rate. The prediction accuracy were 90.9% and 86.4%; AUC values were 0.979 and 0.919; the extrapolation ability were 1.6% and 1%. (6) Predicting the students keeping in the maintenance action stage, the sensitivity was at 83.1% by Logistic regression analysis, and at 84.9% by neural network analysis. The overall prediction accuracy by Logistic regression analysis was 80.7%, and the neural network analysis was 85.4%. Conclusion: Collegiate students develops regular exercise habit is one of the major factor for maintaining regular exercise in the future. To develop regular exercise habit at the college stage may be able to obtain health benefit and also to reduce the risk of chronic diseases in the future.