事件研究之所以可以被用來探討事件是否對股票報酬有影響，其中一個基本原則是所計算的CAR的平均值應為零。本文指出當母體參數有改變而使用文獻上所建議的移動β法：領先法和對稱法，有可能違反比原則。使用Monte-Carlo試驗法，本文發現在母體參數有改變的情況下，領先法和對稱法均有可能誤估情形，而本文提出的正確法則可完全掌握事件的影響，如因為樣本不足而不能使用正確法，領先法亦勉強可用，至於對稱法和固定係數法都會產生極大的誤估。

The use of CAR to study how an event affects the stock price IS well known as the event study. The underlying assumption behind the event study is that the expected value of CAR should be zero. This assumption holds when the parameters are fixed. When the pa-rameters are changing over time, the assumption is violated either leading or symmetric methods, two popular methods suggested in the literature, is employed. An appropriate method is discussed in this paper when the parameters are not constant. A Monte-Carlo experiment is conducted to compare their performance.