本研究之主要目的是利用結構方程模式來同時檢定管理當局非預期銷貨及盈餘、財務分析師非預期銷貨及盈餘、異常報酬率、及異常交易量等四項變數之關係。本研究中設立了兩種模式。其中模式1假設管理當局非預期銷貨及盈餘及財務分析師非預期銷貨及盈餘有交互影響之關係。而模式2則假設管理當局非預期銷貨及盈餘對財務分析師非預期銷貨及盈餘有正向影響之作用。 吾人利用LISREL軟體來分析上述2個模式,實證結果顯示模式1及模式2之衡量模式(measurement model)皆支持各項指標(indicator)擁有收斂之效度(convergent validity),亦即每一指標皆能有效地代表其潛在變數(latent variable)。同樣的,結構模式測試結果發現:模式1中之管理當局非預期銷貨及盈餘與財務分析師非預期銷貨及盈餘有交互地影響作用。而模式2中之管理當局非預期銷貨及盈餘對財務分析師非預期銷貨及盈餘有正向之影響。除此之外,模式1及模式2之異常報酬率和異常交易量皆有相互影響作用。此兩個模式經適合度指標檢定後,認定是個適宜之模式。 Previous research supported that managers' and analysts' earnings forecasts were associated with stock price reactions. While few research tried to examine stock price with revenues and earnings forecasts at the same time. Also, until now, no research has examined the relationship among stock price and managers' and analysts' revenues and earnings forecasts simultaneously. The purpose of this study is to use a structural equation framework to examine the relationships among those variables together. In order to examine the relationships, this research developed two models. Model 1 proposed hypotheses involving the stock price reactions to both managers' and analysts' in revenues and earnings. Model 2 tested hypotheses involving the analysts' forecasts as an intervening variable for the relationship between managers' forecasts and stock price reactions. Using structural equation models analysis (Lisrel analysis), the testing results of measurement models supported convergent and content validity for every indicator in all constructs. Also, the tests from structural models supported that management unexpected revenues and earning were positively related with analysts' unexpected revenues and earnings. And, management information had positive effects on analysts' information. In addition, cumulative abnormal returns and cumulative abnormal trading volumes were interactive. In addition, the overall modeling tests for Model 1 and Model 2 were indicative of good fit.