Republican presidential candidate Donald Trump models a hard hat in support of miners during his rally at the Charleston Civic Center in Charleston, W.Va. (Mark Lyons/Getty Images)

A reporter asked me this question today. Here’s a paraphrased version of our exchange.

Would an early ad campaign against Trump be effective?
The starting presumption about the effectiveness of any early ad campaign — against Trump, against whoever — should be skepticism. Ads can certainly shift people’s opinions, but the impact tends to wear off quickly. Most of the impact disappears after a few days. See this analysis of ads in 2000 presidential election and in 2006 down-ballot elections by James Lo and colleagues. Lynn Vavreck and I found the same thing in our book about the 2012 presidential election.

This isn’t to say that an early ad campaign couldn’t have some impact on Trump’s poll numbers in the moment. But would the impact of ads aired in the summer still be felt in November? Most likely not.

But many analysts thought the early advertising in 2012 helped define Romney as a wealthy businessman who built his fortune at the expense of regular people.

We could not find any evidence of this in our book. Between January and November 2012, we found virtually no change in whether people perceived Romney as “caring about” the wealthy, middle class, poor or “people like me.” The trend lines were flat — including after the early ads, the news coverage of Bain Capital, and the 47 percent remarks. If he was being “defined” by those ads, it wasn’t evident to us. Here’s a graph from the book:

Obama and Romney were perceived like previous presidential candidates — perhaps because these questions about who “cares” about whom tap into decades-long partisan stereotypes. This could be one reason why these perceptions were difficult to change, including with an early ad blitz.

If the Democrats tried an early ad campaign against Trump, would it matter even in the short run? Or is Trump immune because he’s such a different candidate who is already defined?

Trump isn’t immune to attacks. Perceptions of even well-known candidates can change. For example, as I described here, in the fall of 2000 Al Gore’s honesty ratings dropped. There, the culprit wasn’t just ads per se, but a broader round of GOP attacks that was echoed frequently in the news media. That suggests a best-case scenario for those who want to attack Trump: hope that attack ads engender media coverage of related issues that extends for weeks if not months.

But it’s hard to imagine that early ads will create a “narrative” that persists unchanged for months. Inevitably, there will be new story lines.

Let’s go back to Romney. The attacks on Romney’s time at Bain Capital in the summer ads were actually not accompanied by consistent news coverage. There was just a short burst of coverage brought on by the Boston Globe story saying he had some decision-making authority at Bain Capital for longer than he had claimed.

The same thing happened after Romney’s comments about “the 47 percent.” That was clearly a big story at the time, but it was arguably overtaken by the first debate. We went from people saying some version of “Romney’s campaign was over” to people wondering whether Obama had just “thrown the election away.”

If it’s tough to control the “narrative” for weeks on end — and the impact of early ads depends on the resulting narrative — then it’s even harder for early ads to have a lasting impact.

To be clear, an early ad campaign against Trump could hurt his poll numbers. Just don’t expect those ads to matter several months later.

People think campaign ads are like nuclear radiation. Once you see the ads, they stick with you forever. The evidence suggests they’re more like Tylenol.