Empirical Comparison of Task Completion Time between Mobile Phone Models with Matched Interaction Sequences

CogTool is a predictive evaluation tool for user interfaces. We wanted to apply CogTool to an evaluation of two mobile phones, but, at the time of writing, CogTool lacks the necessary (modeling baseline) observed human performance data to allow it to make accurate predictions about mobile phone use. To address this problem, we needed to collect performance data from both novice users' and expert users' interactions to plug into CogTool. Whilst novice users for a phone are easy to recruit, in order to obtain observed data on expert users' performance, we had to recruit owners of our two target mobile phone models as participants. Unfortunately, it proved to be hard to find enough owners of each target phone model. Therefore we asked if multiple similar models that had matched interaction sequences could be treated as the same model from the point of view of expert performance characteristics. In this paper, we report an empirical experimental exercise to answer this question. We compared identical target task completion time for experts across two groups of similar models. Because we found significant differences in some of the task completion times within one group of models, we would argue that it is not generally advisable to consider multiple phone models as equivalent for the purpose of obtaining observed data for predictive modeling.