Semantic Distance And the Alternate Uses Task: Recommendations for Reliable Automated Assessment of Originality
Semantic Distance And the Alternate Uses Task: Recommendations for Reliable Automated Assessment of Originality
复制标题
DOI：
10.1080/10400419.2022.2025720
复制
发表时间：
2022-01-23
影响因子：
2.6
通讯作者：
Forthmann, Boris
中科院分区：
文献类型：
--
作者：
Beaty, Roger E.;Johnson, Dan R.;Forthmann, Boris
语义距离越来越多地用于发散性思维任务的独创性自动评分，如替代用途任务（AUT）。尽管语义距离有一些心理测量支持，包括与人类创造力评级的正相关，但还需要进一步的工作来优化其可靠性和有效性，包括为AUT管理识别最可靠的项目（对象）。我们根据系统的物品选择策略（皮带、砖、扫帚、桶、蜡烛、时钟、梳子、刀、灯、铅笔、枕头、钱包、袜子）确定了一套13件AUT物品。这个项目集产生了可接受的信度估计，并被发现与人类创造力评级和创造性人格因素有适度的关系（研究1）。这些结果在一个新的参与者样本中得到了重复（研究2）。最后，我们提出了以下建议：1)基于理论/实践考虑做出选择；2)管理（部分或全部）本研究中的13个项目；3)如果必须使用其他项目，避免使用复合词作为AUT项目（如吉他弦）；4)在时间允许的情况下尽可能多地包括AUT项目；5)指导参与者“要有创意”；6)解决混淆想法数量和质量的流利性混淆（例如，通过Max评分）。
Semantic distance is increasingly used for automated scoring of originality on divergent thinking tasks, such as the Alternate Uses Task (AUT). Despite some psychometric support for semantic distance - including positive correlations with human creativity ratings - additional work is needed to optimize its reliability and validity, including identifying maximally reliable items (objects) for AUT administration. We identify a set of 13 AUT items based on a systematic item-selection strategy (belt, brick, broom, bucket, candle, clock, comb, knife, lamp, pencil, pillow, purse, sock). This item-set resulted in acceptable reliability estimates and was found to be moderately related to both human creativity ratings and a creative personality factor (Study 1). These results replicated in a new sample of Participants (Study 2). We conclude with the following recommendations for reliable and valid assessment of AUT originality using semantic distance: 1) make choices based on theoretical/practical considerations, 2) administer (some or all of) the 13 items from this study; 3) if other items must be used, avoid compound words as AUT items (e.g., guitar string); 4) include as many AUT items as time permits; 5) instruct participants to "be creative"; and 6) address fluency confounds that conflate idea quantity and quality (e.g., via max scoring).