本文主要目的是在實証上檢驗由條件政策議決所評估之非市場財貨的嵌入效果，此類資料在分析上雖然較複雜，但這種探索可以擴展條件政策議決資料的應用範圍，採用複反應巢型羅吉模型，可用來估計多元議決資料中個別政策因素的邊際價值。實証分析資料來自一組評估美國俄亥俄州中部維護一生態保護區價值的調查。該政策包含維護改善多項自然環境資源，受政策影響的各項環境資源（政策因素）依序置放於政策評估的不同順序中。 實証檢定結果顯示，由多元議決條件評估所得到的個別政策因素之邊際價值，依該因素位於不同評估順序而有差異，也就是說，嵌入效果存在於條件政策議決之調查中。此結論與經濟理論上之預期一致，且與過去的研究探討私有財及開放式條件評估公共財觀察到的現象 是相同的。

　　　　　The embedding effect observed with contingent policy referendum of nonmarket goods is empirically examined. The investigation extends the application of contingent policy referendum data though the analysis is somewhat more complicated. Multiple-response nested logit model developed by Wu allows the estimation of marginal value of multiple-response data. In the empirical analysis, a manageable number of valuation sequences are chosen in such a way that each policy component (environmental attribute) appears in a different position in each particular sequence. A set of multiple referendum data is selected from a contingent valuation survey concerning proposed improvements in the environment of Big Darby Creek in central Ohio. Statistical tests based on the estimated parameters are found to provide evidence that marginal value derived from multiple-referendum contingent valuation is dependent of the sequence. That is, the embedding effect exists in contingent policy referendum survey. This result is consistent with standard economic theory and with the phenomenon observed with private goods and public goods evaluated by open-ended contingent valuation survey explored by other research.