CNN  — 

The 2021 California gubernatorial recall election is in the final stretch, and there is a clear trend in the data.

Democratic Gov. Gavin Newsom’s position (i.e. the “no” option on the recall) is strengthening significantly. While there are a few gubernatorial elections featuring a polling error as large as Newsom’s advantage, it would take one of the largest polling misses in the last 23 years for him to be recalled on Tuesday.

Take a look at the recent polling in the recall race. The average of the polls has “no” on the recall ahead of “yes” by about 15 points. “No” has widened its lead from about 10 points a week ago and from single digits in mid-August.

The edge that the “no” position has over the “yes” position in the recall has, in other words, more than tripled in the average since last month.

It shouldn’t be too surprising that the movement has been against recalling Newsom. While any election is a choice between candidates, a recall is by its very nature more about the incumbent’s standing in the minds of the voters than a regular election. Newsom’s approval rating has regularly and usually comfortably been above 50% in the polls.

Now, obviously number crunchers like myself are cautious after polling in a number of recent high-profile situations underestimated the conservative position, which in this election would be the “yes” side.

An examination of those misses reveals that the polling now is nothing like it was then.

Most of the biggest misses in last year’s presidential election occurred in the Midwest. President Joe Biden held a high single-digit lead in Wisconsin, while then-President Donald Trump held slight leads in in Iowa and Ohio.

Biden ended up winning Wisconsin by less than a point, as Trump carried Iowa and Ohio by 8 points each.

Depending on how you calculate the averages, the polling underestimated the result by about 7 points in these three states.

“No” is ahead on the recall by double the size of these 2020 polling misses.

The other thing to note about 2020 that is especially relevant here is that there was not a significant polling miss in California. The polls pretty much perfectly nailed Biden’s 29-point win in the state. The polling in 2016 and 2018 actually underestimated the Democrats’ wins in the key races that year of president and governor, when Newsom was first elected.

Indeed, Republicans have tended to be underestimated in states where the majority or clear plurality of voters are Whites without a college degree such as in the aforementioned Midwestern states. In California, these same voters make up only about a quarter of the electorate.

The longer arc of polling history is against the Republicans too. I looked at 243 gubernatorial elections since 1998 with at least two polls conducted in the final three weeks. There were 4 of these 243 races in which an average of the polls over the final three weeks were off by 15 points or more. That’s only about 2% of the time.

This 2% could underestimate the chance the polls are off in this year’s recall because of its unique nature (i.e. it’s an off-year election in which voters aren’t deciding between a Democrat and a Republican initially but first on whether to recall Newsom). Still, this 2% is a pretty good baseline to understand that the type of polling error the recall effort needs to succeed happens very rarely.

Keep in mind, too, that any polling error could go against the recall effort. The polls could be underestimating the opposition to the recall, meaning the “no” vote is higher than it looks – a potentially under-discussed possibility.

Neither of the two major gubernatorial recalls of the last 20 years (the 2003 California recall or 2012 Wisconsin recall) had polling misses anywhere what Republicans hope for now, either.

One way the polls could be seriously off would be if pollsters misjudged which voters would ultimately turn out to vote.

So far, there really isn’t much of a sign to expect a vastly different turnout by party than in 2020, when Biden easily won the state.

As of late last week, Democrats made up about 28 percentage points more of voters who cast pre-Election Day ballots than Republicans, according to Political Data Inc., a firm that does work for Democratic candidates, progressive organizations and nonpartisan campaigns. At this point in the 2020 election, the Democratic advantage was closer to 30 points.

This slightly more Republican leaning electorate than you might get in a presidential year is about what pollsters expect.

It’s obviously possible that Republican turnout in the final days is so great that the ultimate voting electorate is much more Republican than 2020 and much more Republican than the pollsters calculated.

Republicans have, after all, been more inclined than Democrats to vote in-person on Election Day since the coronavirus pandemic began. It just seems unlikely at this point.

For Newsom to get recalled, something would need to dramatically change in the final two days before the election. Again, that’s plausible and wouldn’t be unprecedented. It just would be a real big surprise.