AI 中文总结
比较大学微积分基于标准Moodle和STACK的两种逐步测试形式，基于经典测试理论框架，通过人工验证评分矩阵对比，结果显示两种形式总体性能高，基于STACK的测试在人工校正、内部一致性等方面表现更优。
AI 中文摘要
本文比较了大学微积分在线评估的两种形式：基于标准Moodle的逐步测试和基于STACK的逐步测试。两种测试都评估分部积分，并将解决方案分为连续的响应字段，但验证机制不同。标准测试依赖预定义评分模式，而基于STACK的测试使用符号验证规则。比较基于两个学生队列的最终人工验证评分矩阵，遵循经典测试理论框架。结果表明两种形式总体性能高且有上限效应。基于STACK的测试所需人工校正更少，内部一致性更高，解决方案步骤与总分之间的关系更连贯。
英文摘要
This paper compares two formats of online assessment in university Calculus: a standard Moodle-based step-by-step test and a STACK-based step-by-step test. Both tests assess integration by parts and divide the solution into consecutive response fields, but they differ in their validation mechanisms. The standard test relies on predefined scoring patterns, whereas the STACK-based test uses symbolic validation rules. The comparison is based on final manually verified scoring matrices from two student cohorts and follows a Classical Test Theory framework, including score distributions, reliability estimates, response-field-level indicators, and correlation-based measures. The results show high overall performance and ceiling effects in both formats. However, the STACK-based test required fewer manual corrections, showed higher internal consistency, and produced a more coherent relationship between solution steps and the total score.
