This is the focus of a lot of my research. One conditioning factor we think matters in whether an ad is successful or not is the prior knowledge and opinions that voters have of the candidates running. Meaning, Hillary Clinton is unlikely to change a lot of minds with her political ads, but an unknown candidate might make a lot of headway with voters across the political spectrum. Of course, the quality of the ads, the issue focus, and other external factors will matter too. But the difference between being a well-known and unknown candidate might matter quite a lot. For example, imagine taking an ad by Hillary Clinton, like the one below (with a still image of the final shot).
Now replace the images and references of Clinton with an unknown candidate. For example, everyone knows that Hillary ran for president, so the above ad from her doesn’t even mention that fact. This allows us to show people an ad from a different candidate, Paula Clark (who is made up; we used stock footage of a different woman), and ask people what they think of her (without having to pretend she ran for president; the office Clark ran for can be left unstated). See below.
The ad is identical; the similarities in appearance between Clinton and Clark are clear. Everything is the same but for the actual person running. This gives us a clear and simple comparison of ad effects in races with candidates that have well-known backgrounds vs. candidates who are newcomers and for whom voters are generally unaware.
Think about this in terms of 2020. Will Pete Buttigieg’s ads do more work for him than Joe Biden’s?
What about swapping out the sponsor of the ad to be an outside group instead of a candidate? See the disclaimer at the bottom of the below still shot.
This gives us a lot to investigate. What impact does the Clinton vs. Clark versions have on Republican voters, for example? Are they more open to liking and perhaps voting for Clark than they would ever be for Clinton? The message of the ad is identical, remember, so in principle the candidate shouldn’t really matter. More soon as I and my co-authors work on this kind of question.