On Wednesday I wrote about Sen. Ted Cruz’s claim that he got more than 40 percent of the Hispanic vote in 2012. No public poll shows a number that high (the range in scant public polling was 34 to 36 percent). Cruz’s office did make available his pollster, whom I spoke with on Thursday.

Chris Perkins of Wilson Perkins Allen Opinion Research told me he conducted a poll in December 2012. He is a pollster who works only for Republican clients. (He also works for more than “200 Christian and other not-for-profit organizations.”) Using voter rolls from the November 2012 election, he identified people with Hispanic surnames who cast ballots, accounted for geographic diversity, and by telephone interviews determined if any were not Hispanic. He got to a pool of 600, 132 of which could either not remember who they voted for or who said they voted for someone other than the only two candidates on the ballot. Of those who were Hispanic and could recall who they voted for, 38 percent said they voted for Cruz. That is where the polling stopped.

However, Perkins guessed — he was candid that this was an “inference” not based on any actual poll data — that of the people who couldn’t remember voting or who couldn’t say they voted for either of the candidates, Cruz really got another 2 percent, bringing his number up, wouldn’t you know, to 40 percent.

There are a raft of problems with this. For starters he never captured Hispanics without a Hispanic surname (e.g. Vicente Fox). Second, Post pollster Peyton Craighill said of the 2 percent bump, “You can’t make assumptions like that.” In getting Cruz up to 40 percent, Craighill said, “there is such engineering ” as to raise serious questions about the results. Moreover, Craighill explained, in order to get an accurate read on not just one number but two numbers (the Hispanic percentage for each candidate) you would need for this sample size of 600 a much greater gap between the two numbers (e.g. 20 and 40 percent) to come to an accurate number for each of the candidates.

But the dead giveaway is in another question in which the pollster asked respondents which of these they preferred: Grant them full amnesty, including path to U.S. Citizenship (35 percent); Give them work permits to allow them to work here legally but do not make them eligible for U.S. citizenship (46 percent); and No change in status for those currently here illegally (12 percent). Using the term “amnesty” and failing to give any option that suggests a pathway to citizenship with conditions, Craighill says, makes this “an advocacy poll.” He said that in his polling he would not use a buzzword like “amnesty.”

The funny thing is even with all that fiddling, the support for doing nothing on immigration (Cruz’s choice in voting against the Gang of 8) was only 12 percent. The no-citizenship approach got the backing of fewer than half of the Hispanics surveyed.

To recap, not even Cruz’s own pollster found he got “over” 40 percent. His methodology to get him to 40 percent (a figure no other public poll came close to) is so fraught with assumptions, inferences and sampling issues as to make it highly suspect.

This is not a trivial matter in the immigration fight. Cruz and others opposed to citizenship for illegal immigrants are trying to convince fellow Republicans that they will do fine with Hispanics by voting against immigration reform. There is a mound of data from polling, focus groups and election results that says anti-immigration reform rhetoric and votes matter quite a lot, and not only with Hispanics. If Cruz’s argument that all this data is wrong rests on an inaccurate assertion about his own experience, then Republicans should be wary.

This is certainly not the only time Cruz’s assertions have been faulty. His unfounded insinuation concerning Chuck Hagel’s possible receipt of $200,000 from North Korea set off howls on the left and dismayed those challenging Hagel’s nomination (who realized how badly he overreached). His dramatic rhetoric and hyperbolic language may be fine for talk radio or even the campaign trail, but as a United States senator he should adhere to a higher standard. Those in national office who play fast and loose with the facts are playing with fire and risking their own reputations and political viability.