Two experiments investigated observer bias in the attitude attribution paradigm. In the first, subjects assessed the usefulness of a constrained essay prior to attributing the writer's attitude. Contrary to the prevailing view that subjects are inattentive to situational forces, a majority indicated that the essay was not useful for making an attribution about the writer. Subsequent attitude attributions, however, revealed a pronounced bias toward correspondent inference. That the same subjects judged the essay to be 'not useful' yet proceeded to make a biased attribution suggests that there are strong pressures in the paradigm itself to use the essay to make the required judgment. A second experiment indicated that considerably more favorable impressions were formed about an attributor who had made a relatively extreme attribution regarding the writer's attitude. Even when the essay had been assigned to the writer, a discounting judgment was viewed in relatively unfavorable terms. The context of this experimental paradigm appears to sanction the informational value executed, the paradigm may prevent a majority of subjects from expressing reservations concerning the value of the constrained essay.