Rigged polls

(Nick Anderson)

Josh Marshall calls this the "Universal Derp Implosion." He's not wrong:

Donald Trump is giving a raging, rambling speech with accusations against everyone and everything. But there's one thing he flagged that everyone needs to understand the details about.

Here's what he said ...

Wikileaks also shows how John Podesta rigged the polls by oversampling democrats, a voter suppression technique. That's happening to me all the time. When the polls are even, when they leave them alone and do them properly, I'm leading. But you see these polls where they're polling democrats. How is Trump doing? Oh, he's down. They're polling democrats. The system is corrupt, rigged and broken. And we're going to change it. [ Cheers and applause ] ...

There are several levels of nonsense here. Let me try to run through them.

You'll note for starters that that the email is from 2008 and Podesta is neither the sender nor the recipient. But that's just a footnote. More importantly, what Tom Matzzie is talking about is the campaign/DNC's own polls. Campaigns do extensive, very high quality polling to understand the state of the race and devise strategies for winning. These are not public polls. So they can't affect media polls and they can't have anything to do with voter suppression.

Now you may be asking, why would the Democrats skew their own internal polls? Well, they're not.

The biggest thing here is what the word 'oversampling' means. Both public and private pollsters will often over-sample a particular demographic group to get statistically significant data on that group. So let's stay you have a likely voter poll with 800 respondents. The number of African-Americans in that sample is maybe going to be 100 people, maybe less. 800 people is a decent sample for statistical significance. 100 is not. So if you're trying to draw conclusions about African-American voters, levels of approval, degree of opposition or support of a candidate, demographic breakdowns, etc. you need to get an 'over-sample' to get solid numbers.

Whether it's public or private pollsters, the 'over-sample' is never included in the 'topline' number. So if you get 4 times the number of African-American voters as you got in a regular sample, those numbers don't all go into the mix for the total poll. They're segmented out. The whole thing basically amounts to zooming in on one group to find out more about them. To do so, to zoom in, you need to 'over-sample' their group as what amounts to a break-out portion of the poll.

What it all comes down to is that you're talking about a polling concept the Trumpers don't seem to understand (or are relying on supporters not understanding), about polls that are by definition secret (campaign polls aren't shared) and about an election eight years ago.

So, either Donald Trump and all of his campaign staff are too ignorant to know what "oversampling" means, and too dumb to know the difference between internal campaign polls and public polls, or they think that his supporters are, so they're deliberately lying to them.

I don't know which it is, but it's definitely one or the other.

Of course, Trump has been flailing about all over the place, desperate to excuse what's likely to be a big loss next month, even if he can't prevent it. Being a "loser" is more than Trump's pathetic little ego can stand.

I say likely loss, because polls mean nothing. Voting means everything. If we're so lazy that we don't vote, or so dumb that we throw away our vote on a third-party candidate with zero chance of winning, we may actually get President Trump for the next four years.

That can happen. The polls aren't showing that it can't. Indeed, most of the battleground states are very close. And the Republican Party has been working hard to suppress the vote. If we don't care enough about America to vote for Hillary Clinton - and for Democrats up and down the line - we could be looking back at President Bush with nostalgia.

I'm reasonably optimistic, but it's not over until it's over.

(Tom Toles)