Look, it's fine to do theory-free experimental research. Like that MT study. If they found a treatment effect, what explains it? Being reminded of an election taking place? Name recognition b/c the names of the candidates are on the mailer? The fact that the mailer looked "official" rather than from a campaign?
Or, like the authors claimed, it could only mean that learning about ideology increases turnout? If you look at their research proposal, that's what they claimed their treatment "effect" would measure.