提升信念與態度之關係一直是社會心理學家關切的問題之一，然而因為計分方式的不同導致了不少方法論上的爭議，但是真正的問題尚待未來加以解決。本研究就是在這種情況下產生的，主要的目的在於探討計分方式對於信念與態度之關係的影響，但同時考慮量表結構與加權方法在其中所扮演的角色。本研究所分析的對象是取材於四項研究方法非常相近的研究資料，並經事前的規劃產生總結資料，然後再經適當的資料轉換，最後進行統計方法分析。結果發現計分方式是影響信念與態度之關係的最主要因素，其次是研究主題、量表結構，而加權方法則否。另外。在排除不同研究主題的差異之後，發現結果信念與結果評價採取單極X單極或單極X雙極這兩種計分法都比採取雙極X雙極記分法，信念與態度之關係來得高。在量表結構上，當結果信念暨評價與態度這兩個量表都是多向度時，比結果信念暨評價或態度這兩個量表為單向度時，信念與態度之關係來得高。再者，計分方式對於信念與態度之關係的影響會因研究主題的不同而有所差異，但不會因量表結構或加權方法不同而有所差異。總之，多向度的結果信念暨評價量表採取單極X單極或單極X雙極計分，將可以提升信念與態度之預測效度，而且也可以解決過去若干理論上的爭議。

　　　　　It is one of the concerns for the social psychologists to improve the correlation between beliefs and attitudes; however, the scoring system exactly results in many methodological issues, which should be solved in the future. The major purpose of this study is to investigate the influence of scoring systems on the relationships between beliefs and attitudes toward the behavior, and simultaneously to consider the scale constructs and weighting methods. Four empirical data were served for the purpose of this study, and they were properly transformed into new data to be analyzed via the factorial analysis of variance. It was found that the scoring system was the most influential factor on the correlation between beliefs and attitudes, the types of study and the scale constructs were the next; however, the weighting method was not. After adjusting the influence of the type of study, it was found that the scoring procedure by using unipolar for beliefs and unipolar/bipolar for evaluations was significantly higher correlations between beliefs and attitudes than that by using bipolar for both beliefs and evaluations. The multi-dimensional construct for the scale of beliefs and evaluations was significantly higher correlation between beliefs and attitudes than the uni-dimensional construct. Again, the influence of scoring systems on the relationships between beliefs and attitudes was only determined by the type of study, but not by the scale constructs and weighting methods. In all, the multi-dimensional scale by using unipolar for beliefs and unipolar/bipolar for evaluations may not only improve the predictive validity between beliefs and attitudes, but can also solve several methodological issues.