We present results from meta-analyses of 68 experimental studies on digital learning prompts, examining what features constitute an effective prompt for learning achievement. First, a systematic review reveals the different features of prompts used across various studies. Second, a quantitative meta-analysis using a random-effects model shows that digital prompts significantly enhance learning achievement (d = .394), though adjustment for publication bias yielded a more conservative estimate (d = .220). Their effectiveness is largely moderated by prompt design features. Based on meta-regression models, we find the highest efficacy for action-based prompts (rule-based AI) (d = .465), prompts designed for specific learner groups (d = .513), and combinations of generic and directed prompts (d = .571). Regional differences were pronounced, with studies from East Asia showing substantially larger effects than European settings. The effectiveness of learning prompts is further moderated by the learning domain and target group. Our findings reveal that prompts are not a universal solution but require thoughtful implementation. We recommend implementing action-based prompts triggered by learner behavior, using log data to tailor prompts to expertise levels. Designers should keep prompts concise to minimize cognitive load. We advise combining generic and directed prompts based on learning goals. Finally, it is essential to ensure learners are familiar with prompt use. These evidence-based guidelines can help optimize digital learning prompts to support diverse learner needs.
扫码关注我们
求助内容:
应助结果提醒方式:
