Okay, fine. So goes the theory, so goes some research. Then I stumble on this paper.
In contrast to previous research, we find little evidence that citizens are mobilized by or learn from presidential advertisements, but strong evidence that they are persuaded by them.
Drats. Data, once again, get in the way of good theory.
This is a unique real-world experiment by Huber and Arceneaux. "Advertising does a little to inform, next to nothing to mobilize, and a great deal to persuade potential voters." Ouch. What I can't tell from the analysis is whether the least interested were more affected, that is, was there an interaction effect for previous knowledge or attention or participation. I'll have to read deeper, and this is a unique single study versus a meta-analysis of several studies, but it does raise some interesting questions.