Computerized methods for collecting confidence ratings: Task influences on patterns of responding

Retrospective confidence ratings and other judgments frequently are collected in computer-based psychology studies, but little research has investigated whether the method with which these ratings are collected influences the resulting data. To explore whether different confidence rating entry methods elicit different responses, 96 subjects were tested in a recognition memory paradigm. To rate confidence in recognition decisions from 0 - 100, half of the subjects used the numeric keypad on the keyboard to respond whereas the other half used an on-screen slider. Notably, whereas subjects using the numeric keypad frequently chose to enter confidence ratings divisible by 5 and 10, subjects using the slider showed no such preference but instead were more likely to accept the slider default value (i.e., 50) for each trial. The method with which confidence ratings are collected may have unintended consequences on confidence rating data and their interpretation.

在基于计算机的心理学研究中，回溯性置信评分（retrospective confidence ratings）与其他判断类任务的数据收集工作已颇为普遍，但目前鲜有研究考察这类评分的收集方式是否会对最终所得数据产生影响。为探究不同置信评分录入方式是否会引发差异化的被试反应，本研究采用再认记忆范式（recognition memory paradigm）对96名被试开展了实验测试。在为再认决策的置信度进行0至100分量表评分时，半数被试使用键盘数字小键盘作答，其余半数被试则借助屏幕滑块完成评分。值得注意的是，使用数字小键盘的被试更倾向于输入能被5或10整除的置信评分；而使用滑块的被试则无此类偏好，反而更常在每一轮试次中直接采用滑块的默认值（即50）。综上，置信评分的收集方式可能会对置信评分数据及其解读带来意想不到的影响。