Into the wild

Field trials of experimental systems in the wild have developed into a standard method within HCI - testing new systems with groups of users in relatively unconstrained settings outside of the laboratory. In this paper we discuss methodological challenges in running user trials. Using a trial of trials we examined the practices of investigators and participants - documenting demand characteristics, where users adjust their behaviour to fit the expectations of those running the trial, the interdependence of how trials are run and the result they produce, and how trial results can be dependent on the insights of a subset of trial participants. We develop three strategies that researchers can use to leverage these challenges to run better trials.