Unsupervised online assessments produce similar scores for students to in-person exams, according to a study.
Researchers found that cheating was either not widespread or ineffective at boosting scores, and determined that online exams are a “viable assessment tool”.
Authors Jason Chana and Dahwi Ahn say the transition of higher education to online classes has been going on since high-speed internet access became widely available, but the Covid-19 pandemic “dramatically accelerated” the process.
“Online classes make distance education affordable and accessible to millions of students, but a key concern is whether scores from online exams are meaningful,” they write.
Campus resource: Assessment tasks that minimise students’ motivation to cheat
They were trying to determine whether online exams without supervision provide “meaningful evaluative information about student learning”, but say it is not as important as one might think to check if online and in-person exams produce similar scores on average .
Instead, the Iowa State University team says that the critical question is whether they provide a similar evaluation of learning, so that students who perform well doing in-person exams also perform well online.
The onset of lockdowns in March 2020 meant that the spring term was neatly divided in two, allowing researchers to compare in-person exams with online ones across an entire university – a “virtually impossible task under any other circumstances”.
The results – which involved nearly 2,000 students across 18 courses – showed a positive correlation for every course, with little variation by the types of questions asked, the field of study, the course level, exam duration, or enrolment.
The findings, published in the Proceedings of the National Academy of Sciences, also showed little evidence that the online exams were disproportionately advantageous for students who performed poorly on the in-person exams, which suggests that either cheating was not widespread, or that cheating was ineffective at boosting exam performance.
However, more recent studies have found that rates of cheating in online examinations have hit a record high – and have increased since this study was conducted.
The researchers found that even in highly uncontrolled environments, online exams produced strikingly similar assessments of student learning relative to in-person exams.
“This encouraging finding, however, does not negate the importance of rethinking assessment for online delivery, especially in light of the explosive growth of generative artificial intelligence tools such as ChatGPT, which can answer complex questions in ways that are nearly indistinguishable from humans,” they write.
They say that given online exams can provide meaningful information student learning, educators can leverage their powerful advantages – such as instant feedback, easy assignment of a random subset of questions to different students, and easy rescoring of individual questions – to deliver better, more authentic assessments.
