With the New York Times running a giant analysis about how the conventional wisdom about the demographics of the electorate is wrong because exit polls exaggerate the number of minority voters, especially Hispanic voters, I’d like to point out that I’ve documented that bias repeatedly since 2001. Here are excerpts from the new NYT article and below is my July 24, 2001 UPI article on pretty much the exact same subject. And then I follow up with my 2013 VDARE article on how the exit polls got 2012 turnout wrong, which means the GOP Establishment’s support for “comprehensive immigration reform” was justified by innumeracy.

There Are More White Voters Than People Think. That’s Good News for Trump.

Nate Cohn @Nate_Cohn JUNE 9, 2016

… New analysis by The Upshot shows that millions more white, older working-class voters went to the polls in 2012 than was found by exit polls on Election Day. This raises the prospect that Mr. Trump has a larger pool of potential voters than generally believed.

The wider path may help explain why Mr. Trump is competitive in early general election surveys against Hillary Clinton. And it calls into question the prevailing demographic explanation of recent elections, which held that Barack Obama did very poorly among whites and won only because young and minority voters turned out in record numbers. This story line led Republicans to conclude that they had maximized their support from white voters and needed to reach out to Hispanics to win in 2016.

Those previous conclusions emerged from exit polls released on election night. The new data from the census, voter registration files, polls and the finalized results tells a subtly different story with potential consequences for the 2016 election.

… Latino voters did not put Mr. Obama over the top, as many argued in the days after Mr. Obama’s re-election. He would have won even if he had done as poorly among Latino voters as John Kerry.

This is all good news for Mr. Trump. There’s more room for him to make gains among white working-class voters than many assumed — enough to win without making gains among nonwhite or college-educated white voters.

But Mr. Trump’s narrow path could close if he loses ground among well-educated voters and alienates even more nonwhite voters than Mitt Romney did four years ago. …

When you hear about the demographic challenges facing the Republican Party, almost all of the data comes from exit polls: surveys conducted with tens of thousands of voters at precincts across the country on Election Day, along with a supplemental telephone survey with early voters.

The exit polls are excellent surveys. But like any survey, they’re imperfect. The problem is that analysts, including me, have treated the exit polls like a precise account of the electorate.

“There are campaigns and journalists who take the exit polls as the word of God about the shape of the electorate and their voting propensities,” said Michael McDonald, a political scientist at the University of Florida who researches voter turnout. “They’re meant to tell us why people voted. They’re not designed to tell us much about the demographic profile of the electorate.”

The exit polls have a series of subtle biases that depict a younger, better-educated and more diverse electorate. Mr. McDonald tentatively reached this conclusion in 2005, and the pattern has been seen in a broader set of data.

The evidence for a whiter, less-educated and older electorate comes from two main sources.

The first — and longest-standing — source of alternative data is the Current Population Survey, known as the C.P.S. Conducted by the Census Bureau, it is the same monthly survey that yields the unemployment report. After elections, it includes a question about whether people voted. …

But for many experts in the field, these issues pale next to those facing the exit polls. For Bernard Fraga, a professor of political science at Indiana University, there is “no question that the exit poll is not as accurate.” He added, “It’s clearly much more reliable to look at the C.P.S. or even better to look at the voter file-based work.” Today, virtually all major campaign polling, voter targeting and election law litigation is conducted using voter file data.

… Hispanic voters played only a modest role in Mr. Romney’s defeat. They cost him Florida — a must-win state for Republicans, but also the closest contest. Elsewhere, Mr. Obama would have easily survived even if Mr. Romney had equaled George W. Bush’s 2004 share of Hispanic voters.

I’ve written about this endlessly for almost 15 years. Here’s an early example from the summer of 2001 pointing out how the exit poll report that Hispanics cast 7% of the vote in 2000 Bush-Gore election was undermined by the Census Bureau analysis that pegged the Hispanic share at 5.4%.

Analysis: Mexican-Americans and the vote

By STEVE SAILER, UPI National Correspondent

LOS ANGELES, July 24, 2001 (UPI)

Pundits have been hailing as a “political masterstroke” the Bush Administration’s reported proposal to offer legalization to many, although not all, illegal Mexican immigrants. (Democrats have countered by suggesting programs should aid undocumented workers from all countries, not just Mexico.) Yet, new data now available on the Census Bureau’s website, although not yet formally issued to the press, shows that the voting strength of Mexican-Americans remains surprisingly limited. People identifying themselves as being of Mexican ethnicity cast only 3.0 percent of the vote in the 2000 election.

In contrast, the Census Bureau found that African-Americans accounted for 11.5 percent of voters, making them almost four times as numerous as Mexican-American voters. Even Hispanics with roots in countries other than Mexico comprised 2.3 percent of the electorate, not much less than the Mexican total.

Non-Hispanic whites dominated voting with 80.7 percent. Anglo whites cast almost 27 times more ballots than did Mexican-Americans.

Although it’s still small, the Mexican-American share has been growing steadily. It’s up from 2.6 percent in 1996. If recent trends continue, it should reach somewhere around 3.5 percent by 2004.

Similarly, the total Hispanic grew from 3.6 in 1988 to 4.7 percent in 1996 to 5.4 percent in 2000. It likely will be about six percent or slightly higher in 2004. The Voter News Service exit poll claimed that Hispanics comprised seven percent of the 2000 vote, but that was based on a sample only one quarter as large as the Census Bureau’s.

The Census Bureau survey’s main weakness is that respondents can falsely claim they voted. Yet, this would only bias the results reported here if some ethnic groups lied more than others. Exit polls, however, have a hard time handling absentee voters.

Although in the long term, Mexican-Americans – and Hispanics in general – are likely to wield massive influence, they probably will not play an outsized role in the 2004 election – in particular, they are unlikely to offer much aid to Bush’s expected re-election bid.

Numerous commentators have uncritically repeated the claim made by Bush pollster Matthew Dowd in The Washington Post that, “As a realistic goal, we have to get somewhere between … 38 to 40 percent of the Hispanic vote,” compared to the estimated 35 percent Bush earned in 2000. Yet, simple math shows that if Bush boosts his share of the Hispanic vote from 35 percent to 40 percent, and Hispanics cast six percent of the votes in 2004, then Bush will gain a mere 0.3 percentage points overall.

“Obviously, all this is trivial compared to such things as the economy,” one left-of-center election analyst commented on condition of anonymity. “If I were George W. Bush, I would just spend the next four years doing things for Florida and Pennsylvania, rather than playing Hispanic games.”

Further, the Electoral College, not the national popular vote, decides presidential elections. (Otherwise, Al Gore would be President.) In 2004, Mexican-American voters are likely to have even less influence there than in the popular vote.

That’s because 72.3 percent of Mexican-American voters in 2000 lived in just two states: California and Texas. Neither one is expected to be up for grabs in the next election.

Bush won his home state of Texas with 59.3 percent last year. He picked up 72 percent of Texas’ white vote. If he were to need in 2004 a higher share among Mexicans just to hold on to Texas, that’s a sure sign he would be doomed to lose nationwide.

In contrast, in California Bush took only 41.7 percent. Even if Bush had won 100 percent of the Mexican-American voters in California last November, he still would have lost California by around 400,000 votes.

In the extremely close 2000 race, 12 states were decided by less than five percentage points. Only seven percent of all Mexican-Americans voters lived in those states. Mexican-Americans accounted for merely 0.8 percent of the all the ballots cast in those 12 states.

Harry P. Pachon, head of the Tomás Rivera Policy Institute in Claremont, Calif., said, “The only countervailing facts are that in 23 of 50 states, Latinos are now the largest minority.” Pachon defended the political utility of Bush’s pro-Mexican-American push: “If 2004 is another squeaker, then it makes a lot of sense.” Pachon also pointed out that the U.S. House and Senate are almost evenly balanced, and the Mexican-American vote could decide which party controls them.

Where does this little-known data come from? Right after every national election, the Census Bureau supplements its monthly Current Population Survey of 50,000 households with additional questions about voting. Jennifer Day of the Bureau said it is not planning to issue a press release on the 2000 election findings for another half year or so, but the actual data is currently available to anyone willing to use FERRET (Federal Electronic Research and Review Extraction Tool), a data query tool on the Census.gov website.

This widespread assumption that policies benefiting Mexican-Americans are crucial to Bush’s reelection chances seems to have been fueled by the Census Bureau’s surprising announcement in March that Hispanics had overtaken African-Americans in numbers, and now comprise exactly one out of eight residents (12.5 percent).

Yet, only one out of 33 voters was Mexican-American. That low share is largely for two reasons.

First, not all Hispanics are Mexican-American. While this should be obvious, observers sometimes seem to conflate the two categories. The Bureau’s survey found that Mexicans comprised just 56 percent of all Hispanic voters in 2000.

Residents of Puerto Rican and Cuban ethnicity were more than twice as likely to vote as Mexican-Americans. Unlike Mexicans, Puerto Ricans are all U.S. citizens. Cubans tend to be older and more politically active than Mexicans.

This distinction between Mexicans and other Hispanics is particularly important in assessing the Bush Administration’s plan to strike an immigration deal to aid Mexican nationals with President Vicente Fox. Some non-Mexican Hispanics have complained that such a plan would be ethnically discriminatory against their peoples. Senate Majority Leader Tom Daschle (D-S.D.) has emphasized that the Democrats’ alternative will not be restricted to just Mexicans.

Thus, because the non-Mexican Hispanic voters (2.3% of the electorate) and Asian-American voters (1.8%) combined outnumber Mexican-American voters (3.0%), Bush’s Mexican-only initiative might end up actually costing him votes among immigrant groups.

Second, residents of America who identify as being of Mexican heritage are only three-eighths as likely to say they voted in November as the average American. There are several causes. Residents of Mexican descent are less likely to vote because they are less likely to be citizens. Further, they tend to be younger, poorer, and less educated than the typical American, all of which correlate with low turnout.

If there is little political gain in courting the Mexican-American vote, why the seeming focus on it? One political observer suggests a symbolic but politically potent reason: “I think it goes back to Bush being nice to Hispanics to help him with suburban moderates, who don’t like Republicans who are too mean spirited.” Because non-Hispanic whites cast four out of every five ballots, that might be the most sensible explanation.

This doesn’t mean the Brimelow-Rubenstein observation of 1997 that mass immigration is massively bad for the GOP in the long run is wrong. It just means that this future won’t go through the formality of taking place as quickly as the conventional wisdom assumes, leaving an opening to do something about it.

Nor does it mean that Trump will necessarily win. He has his a lot of work cut out for him to convince enough whites to vote for him.

But these observations raise major red flags about the innumeracy of the formulators of the conventional wisdom.

A standard topic for punditry these days is Why Trump? What did the Republican Establishment do wrong to let Trump become the nominee?

Here’s an answer you haven’t heard too often: The innumeracy of the Republican Establishment Brain Trust is a major reason why the Republicans didn’t field candidates who could stop Trump. They all believed in the Hispanic Tidal Wave Theory that meant they had to Hispander like crazy.

That’s why the GOP big money made Jeb Bush the 2015 frontrunner: they thought his message that he loves Mexicans more than he likes Americans was dictated by the numbers! That’s why in 2013 Marco Rubio wrecked his 2016 chances by deciding to be the face of the Democrats’ amnesty bill. That’s even one reason why Cruz, blessed with a Hispanic surname, became a major force, despite his evident shortcomings (although that’s minor compared to the major screw-ups of Jeb running at all and Rubio choosing the wrong side on amnesty).

Back on June 3, 2013, I wrote in VDARE:

Ever since last November’s election, we’ve been hearing that Hispanics comprised a record 10 percent of the vote—which therefore obliges Republican Congressmen to pass “comprehensive immigration reform” a.k.a the Schumer-Rubio Amnesty/ Immigration Surge bill RIGHT NOW.

…

But what if these nice, round turnout numbers provided by the Edison exit poll company weren’t true? What if the “Comprehensive Immigration Reform” clamor is based on exit poll error?

What if in 2012 the Sleeping Giant of the Latino Vote didn’t actually awake—but instead rolled over and started a new siesta?

In short, what if the Main Stream Media exaggerated the Hispanic share of the 2012 vote by a factor of almost 20 percent?

Well, we now have the numbers. We now know that the suppositions behind these awkward questions are true.

After every national election, the Census Bureau conducts a massive survey of voter turnout. Then it bureaucratically mulls over the results for months—while the conventional wisdom congeals around whatever slapdash numbers the exit poll firm emitted in the early going.

In contrast to the Census Bureau survey, though, exit polls aren’t designed to measure turnout. Heck, exit polls aren’t even very good at figuring out who won the election—just ask President John F. Kerry.

Exit polls can’t be based on the random samples that would be needed to measure turnout accurately, because the exit poll company has to bake a forecast of the electorate’s demographics into its plan of which precincts to send workers to cover. Not surprisingly, it tends to get back the results it anticipated.

Moreover, Hispanics are both of interest to sponsors and difficult to survey (they can need Spanish-speaking pollsters). So their needs are typically given more weight in planning the exit poll. The result:  national exit polls have overstated the Hispanic share of the vote at least since 2000.