Recently, researchers have started to investigate using app distribution channels, such as Apple’s App Store or Google’s Android Market to bring the research to the users instead of bringing the users into the lab.
My colleague Niels, for example, used this approach to study how people interact with touch screens of mobile phones. But, instead of collecting touch events in a boring, repetitive task he developed a game where users have to burst bubbles by touching them. And instead of conducting this study in the sterile environment of a lab he published the game on the Android Market for free, so it was installed and used by hundreds of thousands of users. So, while these users were enjoying the game they generated millions of touch events. And unlike traditional lab studies, this data was collected from all over the world and many different contexts of use. The results of this study were reported at MobileHCI ’11 and were received enthusiastically.
Since my work is on pedestrian navigation systems & conveying navigation instructions on vibration feedback, lab studies are oftentimes not sufficient. Instead we have to go out and conduct our experiments in the field, e.g. by having people navigate through a busy city center.
So, if we can bring lab studies “into the wild” can we do the same with field experiments?
My colleague Benjamin and I started addressing this question in 2010. We developed a consumer-grade pedestrian navigation application called PocketNavigator and released on the Android Market for free. Then, we developed algorithms that allow us to infer specific usage patterns we were interested in. For example, these algorithms allow us to infer whether users follow the given navigation instructions or not. We also developed a system that allows the PocketNavigator to collect these usage patterns along with relevant context parameters and send these to one of our servers. On a side-note, the collected data does not contain personally identifiable information, so it does not allow us to identify, locate, or contact users.
With this setup we conducted a quasi-experiment. Since my research is about the effect of vibration feedback on navigation performance and the user’s level of distraction, we compared the usage patterns of situations where the vibration feedback was turned on versus turned off. Our results show that the vibration feedback was used in 29.9 % of the trips with no effect on the navigation performance. However, we found evidence that users interacted less with the touch screen, looked less often at the display, and turned off the screen more often. Hence, we believe that users were less distracted.
The full report of this work has been accepted to the prestigious ACM SIGCHI Conference on Human Factors in Computing Systems (CHI ’12) and has been presented in May 2012 in Austin, Texas.
The paper can be downloaded from here.
