虛擬社群線上行為模式之實證研究－線性結構模式之交叉效度檢定

Abstract

隨著網路科技的普及，網路使用者的技術門檻需求大幅降低，改變了傳統的人際關係溝通模式，形成許多新的網路社群。虛擬社群經營者必需滿足網友間的互動，網友於社群間和他人溝通獲得滿足，進而加深其參與程度，社群經營者要如何吸引人潮、留住人潮也格外值得我們思考。本研究提出一「虛擬社群線上行為模式」因果模式，採問卷調查模式以台南市東區網咖店為樣本母體，有效問卷回收516份，並依亂數分類成二組樣本資料，經結構化方程式模型之建構與實證，並得致下列研究結果：(1)研究模式中各潛在依變項對整體模式的變異解釋力(R^2)分別為：行為(0.881)、行為傾向(0.386)、態度(0.691)，顯示本研究所提的修正計畫行為模式在關係型虛擬社群，為一可接受的模式。(2)在雙交叉驗証方面，以基準樣本為基礎模式時交叉效度指標(Cross validation index, CVI)為10.87，落在顯著水準α=0.05信賴區間(10.13~11.64)中；反之，以校準樣本為基礎模式時交叉效度指標為8.32，落在顯著水準α=0.05信賴區問(7.77~8.89)中；顯示研究二組樣本具有交叉效度，研究量表與研究模式之接受性相當高。

With the simplification of internet technology, the technical requirements necessary for internet users have drastically declined. This has changed the traditional communication model of interpersonal relationships and created many internet communities. A virtual community has to satisfy the desire for interaction among its visitors, fulfill their need to communicate with others in the community and furthermore deepen their degree of participation. How a virtual community attracts and keeps its visitors is worth pondering. This study presents a causal model of ”the online behavior model in a virtual community.” The questionnaires were filled out at internet cafes located in the east district of Tainan, 516 valid of questionnaires were returned and were divided into two groups according to random numbers. Through structural equation modeling, the results came out as follows: (1) In the research model, the variation R^2 of each latent dependent variable to the holistic model rated as behavior (0.881), behavioral intention (0.386) and attitude (0.691) respectively, which shows that the revised theory of planned behavior is an acceptable model in the relevant online virtual community. (2) In respect of the double cross test, the cross validation index (CVI) read as 10.87 and fell between a significant level (α=0.05) and a confidence interval (10.13~11.64) when the calibration sample was taken as the basic model; conversely, the cross validation index (CVI) read as 8.32 and fell between a significant level (α=0.05) and a confidence interval (7.77~8.89) as the validation sample was the basic model. The two sets of samples present cross validity; the research questionnaire and model are highly acceptable.