Towards More Practical Spacing for Smartphone Touch GUI Objects Accompanied by Distractors

Utility of Crowdsourced User Experiments for Measuring the Central Tendency of User Performance: A Case of Error-Rate Model Evaluation in a Pointing Task

“…For target pointing tasks in PC environments, Komarov et al (2013) found that crowdsourced and lab-based experiments led to the same conclusions on user performance, such as that a novel facilitation technique called Bubble Cursor (Grossman and Balakrishnan, 2005) reduced the MT compared with the baseline point-and-click method. Yamanaka et al (2019) tested the effects of target margins on touch-pointing performance using smartphones and reported that the same effects were consistently found in crowdsourced and lab-based experiments, e.g., wider margins significantly decreased the MT but increased the ER. Findlater et al (2017) showed that crowdworkers had significantly shorter MTs and higher ERs than lab-based participants in both mouse-and touch-pointing tasks.…”

“…However, some mouse-users might use touchpads in actuality, as we had instructed to use mice. Similar concerns have been reported before: for touch pointing tasks with smartphones, researchers could not confirm whether workers tapped a target with their thumb as instructed (Yamanaka et al, 2019). Some other crowdsourcing platforms support an option that task requesters can ask workers to shoot a video when they perform a task, e.g., UIScope (http:// uiscope.com/en).…”

“…However, researchers are aware of the risk of conducting a user experiment with a small sample size; e.g., the statistical power is weak (Caine, 2016). Therefore, using crowdsourcing services to recruit numerous participants has recently become more common, particularly for user experiments on graphical user interfaces (GUIs), e.g., (Komarov et al, 2013;Matejka et al, 2016;Findlater et al, 2017;Yamanaka et al, 2019;Cockburn et al, 2020).…”

Utility of Crowdsourced User Experiments for Measuring the Central Tendency of User Performance to Evaluate Error-Rate Models on GUIs

“…For research involving user experiments on graphical user interfaces (GUIs), it has recently become more common to recruit workers through crowdsourcing services (Cockburn et al 2020;Findlater et al 2017;Komarov, Reinecke, and Gajos 2013;Matejka et al 2016;Yamanaka, Shimono, and Miyashita 2019). Previous studies have typically focused on designing better GUIs or conducting user experiments to evaluate novel interaction techniques compared with baselines to demonstrate that a proposed method is statistically better than a baseline.…”