Americans' confidence in government is low by historical standards. We fielded a national telephone survey to examine influences on government confidence and whether public discontent was affected by altering the salience of specific government operations. We used a question order experiment where we alternated between first asking a general question about confidence in government and first asking about confidence in specific government operations. We found that posing the specific policy questions first elevated general confidence more than the reverse. The largest and most noticeable effects were observed for Republicans and those most knowledgeable and attentive to politics. Findings of context effects and the partisan subgroup differences are consistent with existing theories on public opinion change as well as with previous question order experiments. But, evidence that the cognitive elite was especially influenced by question order challenges previous theories that presumed that its sophistication made it immune to influence. We suggest that education and political knowledge expand rather than contract the opportunities for priming.