新產品判斷性預測準確性之員工行為衡量量表建構

Abstract

在新產品上市前，如何有效預測市場反應一直是企業成敗的關鍵。其中，判斷性預測因不仰賴歷史銷售資料，成為經理人面對新產品時最常依賴的預測方式之一，因此，判斷性預測在新產品預測中扮演著重要角色，若經理人預測出現偏誤，將可能導致新品上市決策失準，本研究即旨在發展一套適用於企業於新產品上市預測階段之自我檢測量表，協助企業檢視其判斷性預測是否具偏誤與改進空間。本研究旨在發展一套適用於企業於新產品上市預測階段所使用之判斷性預測自我檢測量表。量表涵蓋三大概念：問題定義、判斷性預測行為及其新產品預測潛在偏誤，六個構面：問題定義、判斷方法、判斷性預測偏誤、事後審查偏誤、倡導偏誤、樂觀偏誤，目的在於協助企業於完成新品銷量預測後進行反思與調整，並提升整體預測準確性與決策品質，研究同時強調，藉由此量表可評估經理人是否因判斷性偏誤而導致預測失準，進一步引發新品上市失敗風險。本研究於前測階段進行項目刪減與構面重組，最終完成一份包含15題、3構面之正式量表。雖然驗證性因素分析結果顯示部分構面之收斂效度尚未達標，惟因本研究為初步量表開發，顯示日後仍可透過題項修訂與資料累積進行優化，具備後續深化潛力。此外，實證分析亦發現不同職級（高階、中階、基層）在預測行為上存在顯著差異，顯示此量表除作為評估工具外，亦可作為企業設計教育訓練課程之參考，特別有助於中階管理層預測能力之培養與提升。

Accurately forecasting market reactions before the launch of a new product is a critical determinant of a firm’s success. Among various forecasting methods, judgmental forecasting—which does not rely on historical sales data—is one of the most commonly used approaches by managers when facing new products. Given its prominent role in new product forecasting, biases in judgmental forecasting may lead to flawed decision-making and ultimately to product launch failures. This study aims to develop a self-assessment scale for enterprises to examine the potential biases in their judgmental forecasting during the new product pre-launch phase.The proposed scale comprises three concepts: problem definition, judgmental forecasting behavior, and potential biases in new product forecasting and six dimensions: problem definition, judgmental method, the post-decision audit bias, the advocacy bias, the optimism bias. The goal is to help companies conduct structured reflection and adjustment after forecasting product sales, thereby improving the accuracy and quality of their decision-making. The scale is also designed to assess whether managerial judgment may have been affected by biases, potentially leading to inaccurate forecasts and launch failure.Through a pretest phase involving item reduction and factor restructuring, a finalized scale with 15 items across 3 dimensions was developed. Although confirmatory factor analysis (CFA) results indicated that some dimensions did not fully meet the criteria for convergent validity, the study’s nature as an initial scale development implies room for future refinement through iterative item revisions and larger sample collections.Moreover, empirical results revealed significant differences in forecasting behavior across managerial levels (senior, mid-level, and junior staff). This suggests that the scale may also serve as a reference for corporate training programs, particularly in enhancing the judgmental forecasting abilities of mid-level managers.