There has been significant interest in indirect measures of attitudes like the Implicit Association Test (IAT), particularly because of the possibility that these techniques may uncover implicit prejudices. We derive a set of qualitative predictions for people's performance in the IAT based on random walk models. Three experiments comparing clearly positive or negative categories to nonwords support the general claims of the random walk model. They also provide evidence that subjects shift their response criterion when doing the IAT. Because of these criterion shifts, a response pattern in the IAT can have multiple causes. Thus, it is not possible to infer a single cause (such as prejudice) from IAT results. A surprising additional result was that nonwords were treated as though they were evaluated more negatively than obviously negative items like insects suggesting that low familiarity items may generate the pattern of data previously interpreted as evidence for implicit prejudice.