Republican Mike Garcia’s victory in California’s 25th congressional district was noteworthy on many levels. Garcia was the first Republican to flip a Democratic seat in California since 1998. As the son of Mexican immigrants and a former Navy pilot, he becomes immediately a high profile hispanic Republican nationally. Finally, Garcia dramatically outperformed the polling. Indeed, if there were ever a race that highlights that political polling can be wrong, it was this race.
There were three polls conducted for this congressional race. All three polls showed the race for the open seat to be a close race between Garcia and his Democratic opponent Christy Smith, with two of the three polls showing Smith in the lead. The final poll was conducted May 6-10 by DCCC, a Democratic polling firm. DCCC indicated that it contacted 675 likely voters showing the Democrat Smith leading the Republican Garcia by a 2% margin, 48% to 46%. 48 hours later, on March 12, the actual election took place primarily by mail-in ballots and Garcia won 56% to 44%. That is a +14% swing in the vote from the final poll which was conducted just days beforehand.
So what explains this huge gap. First, there could be a huge enthusiasm gap. Republicans could be more energized and ready to turn out and vote. It’s also true that turnout was 146,992 which is down by almost a third from the 2018 election when Democrats turned out in record high numbers. At same time, this is actually a pretty good turnout for a special election, particularly one taken during a pandemic. Second, the coronavirus pandemic could be hitting Democratic constituents harder than Republican voters. There is some evidence that Democrats are more scared of venturing outside than Republicans. Third, the polling could have been conducted poorly, driven by over sampling of Democratic constituencies. Still, there were three polls and all three showed the race to be close. Finally, voters could be afraid to tell pollsters of how they are actually going to vote. Again, there is some evidence that minority voters — in this case hispanic voters — are hesitant to tell pollsters their position if it is against the conventional expectation for that minority group. Thus, black and hispanic voters who lean Republican may not be willing to tell pollsters who they are voting for.
None of these explanations are great for Democrats. JHK Forecasts currently suggests that the Democrats have a 63% chance of winning the seat in November as does the Cook Report which had this as a toss-up. Given the results this week, we believe that Garcia has to be considered the favorite. For Garcia to win this seat in such a convincing manner (+12% margin) suggests that there was a powerful break between the polling and what actually occurred. To see how impressive Garcia’s victory was, we can compare it to prior elections in CA-25. In 2018, Democrats won by +10% and in 2016 the Republican candidate had won by +6%. The bottom line is that electorate that voted was dramatically different from what the polling had predicted. An election is the only poll that really matters and until the vote is counted, we should remain skeptical of polling or predictions that give an unduly high likelihood that a certain outcome will occur.
NewEnglandPolitics.com 