Trump's convictions show no distinct effect on voter backing

With Donald Trump’s felony convictions inscribed in the historical record and the first debate of the presidential cycle about to take place, time is ripe for both more accountability and additional iconoclasm. 

First, accountability. 

I recently argued here against the proposition that voters would punish Trump if a jury of his peers convicted him of felonies.

How has that prediction fared? 

Listening to many of television’s talking heads, you’d think the sharp decline in Trump support predicted by the pre-trial polls had materialized in real life. The facts say otherwise. 

Instead of cherry-picking individual polls, start with changes detected by poll aggregators who average all the polls using somewhat different rules.

One baseline is just before the trial began on April 15. 538’s average gave Trump a margin of .8 points over Biden on April 14. 

The day after the verdict was rendered, Trump’s margin grew slightly to 1.4 points. 

Of course, during the weeks of the trial a variety of other events could have intervened to alter the race. What happens if we bound the dates more tightly?

The day before the verdict, Trump led by 1.2 points; the day after, 1.4 points; the week after, 1.2 points.

So, altogether, somewhere between no change and the tiniest movement toward Trump,

but the convictions produced no evidence of sharp decline in Trump support.

RealClearPolitics generates a slightly different average but yields similar results, with Trump very slightly improved from the day before the trial to the day after the verdict, though .4 of point lower the day after the verdict than the day before. Hardly meaningful in a polling context.

The Hill collaborates with Decision Desk HQ to produce yet a third average, which moved against Trump by 0 to .2 points, depending on your chosen start date. Taken as a group, these aggregated data reveal no consistent evidence of meaningful, verdict induced, punishment of Trump.

A second approach compares individual polls from the same company before the trial started and after the verdicts were rendered. I found six such polls. The change in Trump’s margin ranged from a decline of 2 points to an increase of 6 points, with an average of less than a 1-point movement toward Trump.

In only two of the six polls was Trump faring worse after the convictions than before them. 

Again, no clear evidence that the felony convictions hurt Trump. 

Finally, two pollsters conducted recontact surveys, reinterviewing voters after the verdict that they had interviewed previously. Both appeared to show about 2 percent deserting Trump. 

Some may debate whether that constitutes a sharp decline; however, the sample was so small that only seven people moved away from Trump and five of those seven shifted to undecided. Nearly as many—five people—moved away from Biden between their two surveys.

The New York Times/Siena Poll reinterviewed a much larger cohort, 1,897 voters they had interviewed in April and May nationally and in their swing state surveys. They found a 2-point shift in the margin, with Trump shedding 1 point and President Biden picking up 1 point. 

That’s a small movement at most. Moreover, since these respondents were interviewed in April and May and many reside in swing states that were subjected to intense campaigning, there is no reason to believe that all, or even most, of them switched because of the convictions.  

Add it all up and there’s a bit of evidence of tiny movement away from Trump, a bit of evidence of tiny movement toward Trump, a good deal of evidence of no impact and no evidence of a sharp decline in Trump support as a result of his felony convictions.  

Now the iconoclasm.

Commentators are expecting big things from the forthcoming Biden-Trump debate.

If one of the candidates drools into the camera for five minutes (or the functional equivalent), the effects could be substantial. Otherwise, history suggests not risking a big bet on big impact. 

Debates have probably never decided a presidential election. Michael Dukakis, Al Gore, John Kerry and Hillary Clinton all “won” their debates, but none won the presidency.

That does not mean debates have no effect — their effects are just usually small and not sustained.

For a host of reasons, it could be different this time, but the tendency to overestimate the impact of campaign events, anointing them “game changers,” seems as irresistible as it is inaccurate.  

Mellman is a pollster and president of The Mellman Group, a political consultancy. He is also president of Democratic Majority for Israel.