Well, the data is available from other studies done on the influence of advertising in an election—party advertising—so we can infer what government advertising would be. That is, it's largely short term and it is largely around confirming as opposed to negating existing beliefs.
The study that I quoted was really around the last five elections, excluding 2015, and showed that government spiked in its spending on advertising in the year before. It varies because, of course, the timing of the elections varied as well.
You've also put your finger on, I guess, a dirty little secret, and that is we really do not know how effective advertising is in influencing the attitudes and behaviours of citizens. We just don't know.