[论文解读] Validating an Instrument for Teachers' Acceptance of Artificial Intelligence in Education

Shuchen Guo, Lehong Shi+1 more|arXiv (Cornell University)|Jun 15, 2024

Online Learning and Analytics被引用 5

一句话总结

这个研究开发并验证了一个27项的量表，用于衡量教师对教育中AI接受的五个维度，在274名师范生中通过CFA和项目分析显示出较强的效度和信度。

ABSTRACT

As artificial intelligence (AI) receives wider attention in education, examining teachers' acceptance of AI (TAAI) becomes essential. However, existing instruments measuring TAAI reported limited reliability and validity evidence and faced some design challenges, such as missing informed definitions of AI to participants. This study aimed to develop and validate a TAAI instrument, with providing sufficient evidence for high psychometric quality. Based on the literature, we first identified five dimensions of TAAI, including perceived usefulness, perceived ease of use, behavioral intention, self-efficacy, and anxiety, and then developed items to assess each dimension. We examined the face and content validity using expert review and think-aloud with pre-service teachers. Using the revised instrument, we collected responses from 274 pre-service teachers and examined the item discriminations to identify outlier items. We employed the confirmatory factor analysis and Cronbach's alpha to examine the construct validity, convergent validity, discriminant validity, and reliability. Results confirmed the dimensionality of the scale, resulting in 27 items distributed in five dimensions. The study exhibits robust validity and reliability evidence for TAAI, thus affirming its usefulness as a valid measurement instrument.

研究动机与目标

• 基于TAM构建的教师对教育中AI接受度（TAAI）的理论框架。
• 在五个维度上发展条目：感知有用性、感知易用性、行为意向、自我效能感，以及焦虑。
• 通过专家评审和think-aloud访谈来确立表面/内容效度。
• 使用现场数据的确认性因素分析和信度分析评估心理测量属性。
• 为TAAI提供一个健壮、经过验证的测量工具，并具备强效度证据。

提出的方法

• 通过文献和基于TAM的框架识别TAAI的五个维度。
• 开发一个32项的工具，并提供阅读刺激使受试者熟悉教育中的AI。
• 通过专家评审和与师范生的think-aloud访谈确保表面/内容效度。
• 将工具对274份有效问卷进行管理，并进行项分析以检测离群值。
• 使用CFA(WLSMV)和Cronbach’s α评估构念、聚合效度、判别效度和信度；剔除不合适条目以实现良好拟合。

实验结果

研究问题

• RQ1 TAAI条目的心理测量特征是什么？
• RQ2 本工具条目在衡量教师对AI接受度方面的可靠性在多大程度上？
• RQ3 本工具条目在衡量教师对AI接受度方面的效度有多高？

主要发现

• 最终的TAAI工具包含27项，覆盖五个子维度。
• CFA显示模型拟合良好：RMSEA 0.061, CFI 0.981, TLI 0.979, SRMSR 0.051, χ2/df = 2.00.
• 梯度载荷（标准化载荷）在0.60到0.94之间，跨越各因子。
• 整体工具信度为0.92；子维度PU(0.88)、PEU(0.91)、BI(0.91)、SE(0.91)、AN(0.77)。
• 全部条目项判别显著（上27%对下27%）。
• 聚合效度得到AVE > 0.5和CR值达阈值的支持。