本文應用解構式計畫行為理論， 並結合貝式學習架構（ Bayesian learning framework），以受限分量迴歸進行正、負向風險認知，對不同農藥 使用程度的農業生產者之農藥使用量決策分析。結果顯示，當對農藥使用 的正、負向風險認知程度越高，農業生產者會有越高的機率使用法令規定 許可的農藥。其中正向風險認知，雖然不管對高、低使用量的農業生產者 之農藥使用量都會造成影響，但隨著使用量的增加，影響的係數與顯著性 也隨之增加。而負向風險認知，則是對低使用量的農業生產者的農藥使用 量有顯著的負向影響，但對高使用量的農業生產者，影響卻不顯著。

This study uses the decomposed theory of planned behaviour, combined with a Bayesian learning framework to determine positive and negative risk perceptions using censored quantile regression. With this we will perform a pesticide use decision analysis of the usage of different pesticides. The result shows that agricultural producers have a higher chance of using pesticides with regulations permitted when these producers have higher positive or negative risk perceptions of pesticide use. Positive risk perceptions have an effect on the pesticide use of the agricultural producers regardless of high or low usage. However, the influence coefficient and significance increase when total usage increases. Negative risk perceptions have a significant negative effect on the pesticide use of agricultural producers with low total usage, but they do not significantly affect the pesticide use of agricultural producers with a high total usage.