A new study casts doubt on the effectiveness of robo-polls, raising the question of whether the results of automated polls in the 2012 Republican primaries may have been adjusted to match live-interview surveys.