簡易版刺激尋求量表的發展及驗證

目標：一、是以臺灣大學生為調查對象，用探索式因素分析，檢驗Zuckerman的刺激尋求量表在臺灣的適用性：二、是將此實證分析結果與過去的研究結果進行彙整、分析、比較，以建構出適用於美國文化以外的簡易版刺激尋求量表；最後再以結構方程模式中的驗證式因素分析，檢定比常模的配適度。方法：由421位研究參與者填寫SSS-V量表及人口變項相關問題，研究者再隨機將374份有效問卷分成二組，一組資料以探索式因素分析，檢驗SSS-V在臺灣的適用性：另一組資料則以驗證式因素分析，檢測此常模的配適度。結果：40題的SS-V題目中有30題出現在其所屬的因素，但仍有10題出現在其不所屬的因素、3題有因素負荷量過低及6題有多重負荷的問題。有鑑於此，本研究以在5個跨國研究中皆正確出現在預期面向、因素負荷量較高及無多重負荷問題的題目為選取標準，由SSS-V中挑出8題以建構簡易版SSS-V。簡易版SSS-V之因素負荷量最低為0.45，最高到0.93；四個因素之組合信度分別為.75、.60、.61、.5 7；收斂效度分別為.61、.43、.45、.40，顯示量表的信效度佳。從整體模式配適度之建議指標評估，其結果皆顯示模式配適度佳，表示假設模式與樣本資料契合度良好。結論：簡易版的SSS-V量表提供配適度高的刺激尋求測量，尤其是在大型調查研究中，研究情境無法允許過多的題目時，特別適用。

關鍵字

刺激尋求；刺激尋求量表；探索式因素分析；結構方程式；驗證式因素

並列摘要

Objectives: The study examines the cross-national replicability of the underlying structure of the Sensation Seeking Scale Form V (SSS-V) to develop a brief form of SS-V (BSSS-V) based on the empirical results of cross-national comparisons, and to investigate the reliability and validity of the BSSS-V Methods: A total of 374 college students completed a self-report questionnaire consisting of 40 SSS-V items. The sample was subsequently divided randomly in to 2 subsamples. The factor structures for half of the data were examined using exploratory factor analysis (EFA). The EFA result and the results obtained from samples in other nations were used to construct a new SSS form with 8 items. The factor structures of the BSSS-V for the remaining half of the data were examined using confirmatory factor analysis (CFA). Results: The EFA result Supports the existence of the 4 dimensions identified by Zuckerman, and the percentage of total variance accounted for was higher than those reported in other studies (43.47%). However, similar to the findings of previous research, there were items that did not load significantly on the expected factors, loaded on the wrong factors, or cross-loaded on more than one factor. The BSSS-V factor loadings ranged from .45 to .93, and the composite reliability indicators of its 4 subscales were .75, .60, .60, .57 with respective AVEs of .61, .43, .45 .40, indicating good scale reliability and validity. A shorted 8-item SSS-V scale based on the result of the cross-national comparison was tested with satisfactory results for all fit indices, with their corresponding recommended values, providing evidence for a good model fit between the proposed model and the obtained empirical data. Conclusions: In light of limitations with the SSS-V, the modified, shortened version of SSS-V may provide a more appropriate measure of sensation-seeking, especially when the goal is to minimize the number of indicators in large-scale surveys.