使用試題反應理論來估計消費者的產品知識

Application of IRT to Estimate Product Knowledge

目的：使用試題反應理論來估計消費者產品知識。產品知識程度分組採用古典測驗中原始總分的加總，再以四分位法來區分高、低產品知識兩組，這樣的估計和分組方式，結果會導致同分者過多，分組人數不等，屬於總加評分量尺，它是根據古典測驗理論「重複測量」和「測驗加總」為基礎，必須符合各題加權相等和測驗量尺視為等距量尺具可加性的假定。方法：本研究以台北地區已購買健身俱樂部會員卡或有購買意願者的447名國小教師為研究對象，以自編消費者產品知識測驗為研究工具進行測驗，使用Xcalibre程式進行三參數對數模式分析，從資料和模式的適合度分析，試題符合性檢定，估計試題參數，估計受試者能力，結果：顯示三參數對數模式能更精確估計消費者產品知識。結論：為利用試題反應理論來估計消費者對於健身俱樂部會員卡的產品知識，將原始分數先轉換為洛基分數（logit score）；將可解決分組人數不等和各題權重不等和等距量尺的問題，精確估計出消費者的產品知識，使分組更加有效，更可以有效提昇產品知識測驗的品質。

Purpose: The purpose of this study is using IRT to estimate Product Knowledge. Methods: The subjects of product knowledge were divided into two groups, according to the sum of the raw score of the classical test theory. The study uses the total sum to discriminate two groups: high involvement (the former 25%), low involvement (the later 25%). Two problems are gourd in the theory. First , there will be too many subjects with the same score. Second, the number of the subjects in the two groups is not equal. It aimed at 447 primary school teachers who have already purr chased a health club member card or have the intention to buy one. The method to measure product knowledge was 'The Consumer Product Knowledge' Scale. Xcalibre was used as an estimation tool for model-data fit analysis, item difficulty estimation, and person ability estimation. Results: The results we got show that three-parameter logistic model can provide a suitable and solid base for estimating consumer product knowledge. Conclusion: The item response theory can estimate the product knoeledge precisely.