本研究以性質上等同「自結盈餘」之「盈餘概估」資料（以下均逕稱為「自結盈餘」），探討「自結盈餘」與「公告盈餘」間之差異是否受公司規模、負債比率、董監持股比例、好消息及審計品質等因素之影響。透過不同的差異衡量（「差異達百分之十」、「差異達百分之二十」與「連續型態的差異比例」等三種研究設計），本研究並檢視現行「百分之二十」與原訂「百分之十」之控管門檻，是否能對公布「自結盈餘」公司，發生識別分群之效果，達成預期的控管作用。實證結果顯示，就整體迴歸結果觀之，以連續型態衡量應變數之OLS模型，主要觀察變數之係數皆未顯著異於零，解釋力遠遜於以間斷型態衡量應變數之Logistic結果。顯示在「自結盈餘」與「公告盈餘」之差異對公司之識別分群效果上，存在重大性水準門檻，亦即位於此門檻上下兩方之公司始有顯著因素不同；現行法令中控管門檻之設置，與此狀況相符。而在以「差異達百分之十」與「差異達百分之二十」之間斷型應變數差異衡量下，本研究實證結果一致顯示：公司之負債比例與差異成正向關係；公司之董監持股比例與差異成負向關係；且擁有好消息之公司其差異較小。但公司規模僅於以「差異達百分之十」定義差異時，與差異成顯著負向關係；審計品質則僅於以「差異達百分之二十」定義差異時，與差異成顯著負向關係。

Based on the Taiwan's listed companies from 1992 to 2000, we analyze whether the difference betweenn audited earnings and unaudited earnings, which will be required to be disclosed, is affected by firm size, debt ratio, management ownership, positive news, and audit quality. The difference between unaudited earnings and audited earnings are measured by two-tier (discrete) method and continuous method, and the Logistic and OLS regressions are applied, respectively. The empirical results are summarized as follows: the large the firm size is, the smaller the difference between unaudited earnings and audited earnings is. The lower the debt ratio is, the smaller the difference between unaudited earnings and audited earnings is. The heigher the management ownership is, the smaller the difference between unaudited earnings and audited earnings is. The company with positive news exist the smaller difference between unaudited earnings and audited ernings. The better the quality of the CPA auditing the audited earnings is, the smaller the difference between unaudited earnings and audited earnings is. In addition, we find that the results of OLS models are not significant statistically. However, those of Logistic models are. It may mean that there exists a material threshold, such as 10% or 20%, on the difference between unaudited earnings and audited earnings.