本研究以 Ajzen(1985)的計劃行為理論(theory of planned behavior)為基礎加以修正，建構台灣民眾至金門觀光行為傾向的因果模式。並經結構方程式模型實證獲得下述之研究結論與研究。本研究修正 Ajzen(1985)「計畫行為理論」，將「態度」修正為中介構面，發現台灣民眾對金門觀光的「形象認知」、「主觀規範」、「成本效益」、「行為控制知覺」對其「行為傾向」存有間接效果。據此，建構具有穩定性的台灣民眾至金門觀光行為傾向之因果模式，可以解釋民眾從事金門觀光的行為傾向。說明如下：1.在顯著水準α=0.01時，民眾的「主觀規範」、「成本效益」等潛在自變項對金門觀光「行為傾向」具有正向顯著影響；將顯著水準放寬至α=0.1時，民眾至金門觀光的「態度」對其「行為傾向」具有顯著正向影響，但民眾的「行為控制知覺」對金門觀光的「行為傾向」，並無顯著影響。2.在顯著水準α=0.01時，民眾至金門觀光的「態度」主要受其「形象認知」、「行為控制知覺」、「主觀規範」、「主觀規範」、「成本效益」等正向影響。本研究所建構出台灣地區民眾至金門觀光的行為傾向模式，進行測量模式與結構模式的穩定度與交叉效度驗證，得到居住台灣不同地區之樣本其測量模式與結構模式具有不變性、穩定度與交叉效度。政府主管單位及觀光產業經營者欲增進觀光客至金門觀光的態度，應從觀光客的成本效益及提升金門觀光形象認知兩方面來考量，並促使觀光客與主觀規範參考群體間的互動，提昇觀光客對金門地區的觀光態度，增進對金門觀光的行為傾向，並在評估成本效益時，獲得正向回饋。

Kinmen has been a popular tourist destination in Taiwan area and received 350,776 tourists in 2000. Kinmen National Park and many historical battlefields form the pillars of these islands’ tourism industry. After a long high economic development time of Taiwan, Taiwanese have aspiration for a better life about tourism. Since the government of Taiwan began to implement the policy of two-day-off weekends in 2001, the demands for tourists’ traveling to Kinmen have been increasingly rising. In response to the booming development, it is necessary to make research concerning tourists’ behavioral tendencies of Kinmen tourism. The purpose of the study is to explore the tourists’ behavioral tendencies, whether they tend to go vacationing in Kinmen. Furthermore, based on the planned behavior theory by Ajzen (1985), it is desirable to understand what are the key factors to influence tourists’ behavioral tendencies in Kinmen and what are the relations of these factors. Then the study will construct the behavioral tendency model of tourist in Kinmen and provide the tourism industry in Kinmen with beneficial information for making business strategies. The study uses confirmatory factor analysis that is based on the planned behavior theory and other relative theories. The factor analysis model is hypothesized that six factors which include image perceived, perceived behavioral control, subjective norm, attitude, cost-benefit and behavioral tendency, then collects data using these six factors with 25 indicators of the questionnaire. The structured questionnaire contained seven parts. Part 1 was design to assess Kinmen tourists’ attitude with 5 indicators, on a 5-point Likert scale, ranging from “strongly agree” (5) to “strongly disagree” (1). Part 2 assess Kinmen tourists’ behavioral tendency on a 5-point Likert scale, ranging from “strongly likely” (5) to “strongly unlikely” (1) with 4 indicators. Part 3 assess Kinmen tourists’ image perceived, on a 5-point Likert scale, ranging from “strongly agree” (5) to “strongly disagree” (/) with 3’ indicators. Part 4 assessed Kinmen tourists’ subjective norm on the same 5-point Likert scale with 5 indicators. Part 5 assessed Kinmen tourists’ perceived behavioral control on the same 5-point Likert scale with 4 indicators. Part 6 assessed Kinmen tourists’ cost-benefit on the same 5-point Likert scale with 4 indicators. Part 7 collected social-demographic data about the respondents. The population for the study was sampling from Taiwanese tourists in Kinmen in the period from July 27 to August 31, 2001 at Kinmen Airport. A questionnaire was used to collect the opinions and perceptions from the Taiwanese tourists traveling in Kinmen. Of the 1,765 questionnaire received, 1,607 were completed and usable. Representing a response rate of 91% Completed and usable questionnaires show that 56.44% of the respondents were female, 43.56% of the respondents were male. Most respondents were in the age categories 21-40(52.46%). These data analysis were used to construct model dimensions and theory models. These data analysed include follow three stages: (a) to use chi-square incremental value to test model fitness and to test discriminant validity; (b) to use three samples which came from the three different regions of Taiwan to test the invariance of these measurement models and these structural models; (c) to construct the theory model by merged data from the above three regions of Taiwan. The study used LISREL 8.30 to analyse these models. Several findings are acquired as the following: (a) according to the perceptions from the tourists touring Kinmen, it shows high consistence between measurement models and structural models; (b) at the significance level of 0.01, the independent variables, subjective norm and cost-benefit, have positive influence on respondents’ behavioral tendencies; (c) there lies positive correlation between respondent’s attitude and behavioral tendency at the significance level of 0.1; (d) level; (e) at the 0.01 significance level, image, perceived behavioral control, subjective norm and cost-benefit all have positive influence on respondents’ attitude. The research findings revealed that Kinmen travel agencies should focus more effort on improving Kinmen tourist perceives about the traveling is valuable. Kinmen travel agencies will have to good reputation of travel agent. The region government of Kinmon should make better tourism advertising and hold series agencies should put the cost-benefit and Kinmen’s image into core marketing competencies while they make strategic planning for enhancing tourists’ behavioral tendencies toward Kinmen.