User Test on Text-Based CAPTCHA: A Letter Case Examination

ABSTRACTCompletely Automated Public Turing Test to tell Computers and Humans Apart (CAPTCHA) is a security application requiring a user to input a string of text to verify that the user is human. Each text string has distinctive visual features, making it perceivably different from others. No empirical work has examined which text strings included in text-based CAPTCHA have high correct response rates from users. The current study hopes to fill this gap. A user test on 8,928 iterations for which both uppercase and lowercase letters were used in 1,844 CAPTCHA tests confirmed (a) users' correct response rates to lowercases were higher than those to the uppercases, although both rates are acceptable, (b) the letters with straight line features had higher correct response rates than those without these features, and (c) the diagonal or the slender features had no impact on the correct response rate. The findings offer both theoretical and practical contributions.