Usability of puzzle and gesture-based CAPTCHAs on smartphones via statistical analysis

Abstract

This paper investigates the usability of the CAPTCHA on smartphones. Four puzzle and gesture-based CAPTCHA schemes are tested: (1) TAPCHA version 2, (2) TAPCHA version 3, (3) FUN and (4) CAPY CAPTCHA. The response time to solve the CAPTCHA is employed as a usability parameter, while demographic and personal parameters such as gender, age, education, daily and yearly Internet experience are used as independent variables. Four hypotheses are examined via statistical analysis, in particular using the Kruskal-Wallis H test and the Mann-Whitney U test. It is proved that the solving time of TAPCHA schemes when solved on a smartphone depends on user's Internet experience, education, and age. On the contrary, the solving time of FUN and CAPY CAPTCHAs does not depend on any tested input parameter. The paper presents new insights on the usability of CAPTCHAs on smartphones. They can be adopted for designing CAPTCHA schemes that are more adjusted to different groups of Internet users.