Replies

Yeah, I read it and saw right through it. You should know by now that there are not too many people on Planet Earth that are more knowledgeable about statistics than I. And that means I don’t respect Trump’s favorable polls either but they are picking up a signal, but none of it is statistically valid.

The first thing I noticed about your link to a ‘poll’ was that the website was directed to ‘Arkansas Matters’. Arkansas? And the poll was not even cited anywhere. Looking closely it appears to be a CNN poll but there is no citation and no link to it anywhere; at least my tired worn eyes did not see it.

The next thing I noticed was “likely republican voters”. Those data come off of voter rolls showing who voted last time. They also come off of asking a series of questions about the likelihood of voting next primary or election, etc.

‘Likely Voter’ was a useful indicator 20 years ago. And since it has descended into a nothing-indicator.

Today, landlines are used by fewer and fewer people. Those that still use them are retirees, elderly, welfare queens (but Obamaphones are making a dent there), and ‘others’ who find themselves sitting at home these days more than they would like to.

A large portion of the ‘Others’ is coming from nearly 100 million out-of-work Americans that are sitting home. This is why Trump ‘consistently’ leads in the polls because the polls are at least managing to pickup that signal although not reliably.

Note pollsters, and the statisticians behind them, are the lowest of the low when it comes to analysts. To real scientists, a political pollster is an embarrassment to science. Rejects and losers usually find a job in such venues. The reason for this is because every last political poll, no matter how touted as ‘respected’, is invalid on its face.

In an MSA of 1,000,000 voters, only about 0.01% (0.0001) will pickup and actually answer a poll. That would be 100 people. Pollsters will then depart from their scientifically prescribed sampling plans and start to use backup lists to get about 1000 respondents so that they can magically pin an MOE of +/-X % to make it look ... (shaking head) ‘respectable’. Kind of like a 3rd world military officer with 40 pounds of medals and decorations on his chest, political polls, all of them, are a farce.

The sampling plan execution is ruined in such settings. Census takers do a great job because it is backed by law. But there is no law that says a person must respond to a political poll, thank goodness.

Where signal is picked up is not in polls, it is in how many independents and dem crossovers show up. Trump commands those, Cruz does not.

Californian republicans, independents, and Reagan democrats are not so enamored with Cruz that they will turn out for him, and they detest Romney and Ryan. The turnout will not be driven by a fairly popular governor as it was in Wisconsin nor will the MoneyPACs of the Party of Romney be able to generate turnout like they did in Wisconsin. Because Californian republicans are a dispirited bunch and are angry with the national party.

A good portion of them will turn out for Donald Trump because he represents the truest outsider and that’s what they really want.

They are not going to listen to the same old same old from the national party. Cruz’ platitudes fall flat. His bible belching will not be received well. They simply don’t like him.