In this paper, we investigate the effect of context on usability evaluation. The focus is on how children behave and perform when they are tested in different settings. Two most commonly applied usability evaluation methods: the think-aloud and constructive interactions are applied to the children in different physical contexts. We present an experimental design involving 54 children participating in two different configurations of constructive interaction and a traditional think-aloud. The behavior and performance of the children in two different physical contexts is measured by evaluating the results of application of think-aloud and constructive interaction. Finally, we outline lessons on the impact of context on involving children in usability testing.