Psychometric security scales can enable various crucial tasks (e.g., measuring changes in user behavior over time), but, unfortunately, they often fail to accurately predict actual user behavior. We hypothesize that one can enhance prediction accuracy via more comprehensive scales measuring a wider range of security-related factors. To test this hypothesis, we ran a series of four online studies with a total of 1,471 participants. First, we developed the extended security behavior scale (ESBS), a high-coverage scale containing substantially more items than prior ones, and collected responses to characterize its underlying structure. Then, we conducted a follow-up study to confirm ESBS’s structural validity and reliability. Finally, over the course of two studies, we elicited user responses to our scale and prior ones while measuring three security behaviors reflected by Internet browser data. Then, we constructed predictive machine-learning models and found that ESBS can predict these behaviors with statistically significantly higher accuracy than prior scales (6.17%–8.53% ROC AUC), thus supporting our hypothesis.