Performance on divergent thinking (DT) tests varies by different testing conditions. Although many studies documented that DT performance increases when more time is provided, it still remains unanswered whether DT performance continues to increase linearly or follows a rather non-linear pattern as more time is allowed. The present study examined a potential curvilinear pattern in DT performance by synthesizing 237 effect size in 22 studies using a three-level approach to account for the nested structure of data. Results indicated that liberal time conditions provided significantly better DT performance than the restricted time conditions with the mean effect size (g = 0.666) being considered moderate to large. The effect size was small to moderate (g = 0.493) in a group of studies that merely compared “timed” and “untimed” conditions. The effect was larger (g = 0.901) in a group of studies comparing varying amounts of time. The quadratic term of time difference between the shorter and longer time conditions turned out significantly negative indicating the inverted J-shaped relationship between time-on-task and DT performance: the DT performance increases with more time, and the growth slows down at some point. Implications for testing DT are explored.

发散思维 (DT) 测试的性能因不同的测试条件而异。尽管许多研究表明，当提供更多时间时，DT 性能会提高，但当允许更多时间时，DT 性能是继续线性增加还是遵循相当非线性的模式仍然没有答案。本研究使用三级方法来解释数据的嵌套结构，通过综合 22 项研究中的 237 个效应大小，检查了 DT 性能的潜在曲线模式。结果表明，自由时间条件比限制时间条件提供了显着更好的 DT 性能，平均效应大小 ( g  =  0.666) 被认为是中等到大。效应量小到中等（g = 0.493）在一组仅比较“定时”和“不定时”条件的研究中。 在比较不同时间量的一组研究中，效果更大（g = 0.901）。较短和较长时间条件之间的时间差的二次项显着为负，表明任务执行时间与 DT 性能之间呈倒 J 型关系：DT 性能随着时间的增加而增加，并且在某个时间点增长放缓. 探讨了测试 DT 的意义。