Publication Details

Abstract

Outliers are participants whose scores are too extreme to be included in statistical analyses and are, hence, considered to be lost participants. In information behavior publications, descriptions of outliers and their handling are sparse or non-existent. This contribution suggests that outliers are not necessarily lost participants, but may, on the contrary, be an indication of normal events in the natural environment and thus an indispensable key to interpret information behavior. In this sense, outliers could even be better represent ations of real users than the statistically standard participant. Data for the outlier analysis come from an asynchronous remote user test that examined simple known-item search task completion in a laboratory and a natural environment test setting.