A comparison of perceived and real shoulder-surfing risks between alphanumeric and graphical passwords

Previous research has found graphical passwords to be more memorable than non-dictionary or "strong" alphanumeric passwords. Participants in a prior study expressed concerns that this increase in memorability could also lead to an increased susceptibility of graphical passwords to shoulder-surfing. This appears to be yet another example of the classic trade-off between usability and security for authentication systems. This paper explores whether graphical passwords' increased memorability necessarily leads to risks of shoulder-surfing. To date, there are no studies examining the vulnerability of graphical versus alphanumeric passwords to shoulder-surfing. This paper examines the real and perceived vulnerability to shoulder-surfing of two configurations of a graphical password, Passfaces™[30], compared to non-dictionary and dictionary passwords. A laboratory experiment with 20 participants asked them to try to shoulder surf the two configurations of Passfaces™ (mouse versus keyboard data entry) and strong and weak passwords. Data gathered included the vulnerability of the four authentication system configurations to shoulder-surfing and study participants' perceptions concerning the same vulnerability. An analysis of these data compared the relative vulnerability of each of the four configurations to shouldersurfing and also compared study participants' real and perceived success in shoulder-surfing each of the configurations. Further analysis examined the relationship between study participants' real and perceived success in shoulder-surfing and determined whether there were significant differences in the vulnerability of the four authentication configurations to shoulder-surfing. Findings indicate that configuring data entry for Passfaces™ through a keyboard is the most effective deterrent to shouldersurfing in a laboratory setting and the participants' perceptions were consistent with that result. While study participants believed that Passfaces™ with mouse data entry would be most vulnerable to shoulder-surfing attacks, the empirical results found that strong passwords were actually more vulnerable.