Lies, damned lies, and then there are polls: Beware any poll that samples just registered voters.

[SirLinksALot]

SOURCE: AMERICAN THINKER

URL: http://www.americanthinker.com/blog/2016/08/lies_damned_lies_and_then_there_are_polls.html

by Andrew Grant Whit


Poll problems.  I'm not talking about adjustments made for improper questions or demographic sampling techniques, though those do tend to bias the poll results in favor of one particular candidate.  (Guess who.)  No, I'm talking about the folks actually sampled, specifically registered voters vs. likely voters.

Most are aware of the RCP rolling average poll and the famous chart showing Clinton cleaning Trump's clock (though not by as much as she needs to at this point in the electoral cycle):



Problem is, I've noticed anecdotally for some time not just that the various polls in this rolling poll average were odd (even ignoring the poll construction flaws), but that those using registered voters (R.V.) only seem to usually favor Clinton – by a lot.  So I took the RCP poll history; deleted all R.V. polls, which still leaves a lot of likely voter (L.V.) polls; and plotted the result.



Tells a different story, no?  Bottom line: using L.V.-only polls, Clinton is currently 44% and Trump is currently 42% with a similar 3% margin of error, not the 7% difference reported by the RCP rolling average poll.  Moreover, both of them are increasing right now, perhaps taking back some of the defections to third-party candidates of recent.

Moral: Beware polls in general – they are all flawed.  But especially beware any poll that samples just registered voters.  They tell you things like "Hillary's bounce remains firm."  Do you see a bounce stability in my chart?  So how can a statement like that be supported?   Because today ABC and the Washington Post released a new poll showing Clinton 8% ahead based on 815...registered voters.  Hmmmmm.

[sinkspur]

Likely voters are not factored into polling until September and October, when election day draws near.  Almost all polling firms will switch to LVs then.

I love how these desperate conservative sites are clinging to any thread to save Trump.  The FACT is, he is losing every single poll, except that bogus LA Times poll, and the trend has been away from Trump for the last two weeks.

Instead of Trump and his supporters focusing on why Trump should win, they're concocting "rigged" election fantasies and contorting polls to show that Trump is not losing by landslide proportions.

[Just_Victor]

I think all bets are off regarding who are the likely voters in this election cycle.  Likely voter models are probably not going to produce more accurate results.

[Bigun]

Polls! 

I personally don't trust ANY for public consumption poll!  They are two easily manipulated and to often are!

[ABX]

I think all bets are off regarding who are the likely voters in this election cycle.  Likely voter models are probably not going to produce more accurate results.

I completely agree. We have one of the largest voting blocks, Millennials, coming of age, many to vote for the first time. The boomer demographic is fading away. We have the old media being pushed aside for pop culture driven new media. We really don't know. RV versus LV still look at past registrations and is leaving a lot out. This one is hard to predict. State by State, the EV count looks rather lopsided, but both candidates have very high negatives. Trump has an advantage for the proverbial 'low information voter' due to his pop culture recognition. At the same time though, he is losing the new, giant Millennial voting block by a significant amount (and Hillary isn't picking that up either so that's a big unknown right now).

[sinkspur]

I think all bets are off regarding who are the likely voters in this election cycle.  Likely voter models are probably not going to produce more accurate results.

I disagree.  We thought the same about the Republican primaries:  that Democrats voting would skew the results, that the polls couldn't possibly track all those first-time voters.

Well, the polls in the primaries were mostly on target (the Dem Michigan polls were wrong), especially when it came to predicting the winners.  Margins were off a point or two, but usually underestimated the performance of the winner.

And such will be the case in the general election.  These pollsters do this for a living, folks.

[Bigun]

  These pollsters do this for a living, folks.

That's EXACTLY right! And can produce whatever result anyone is willing to pay for!

[ScottinVA]

Yes, I remember "unskewed polling."

How accurate was that?  I forget..

[Suppressed]

I love how these desperate conservative sites are clinging to any thread to save Trump.  The FACT is, he is losing every single poll, except that bogus LA Times poll, and the trend has been away from Trump for the last two weeks.

I'm very much against the charlatan, but I think it's important to consider two things. 

Trump does better on LV over RV.  That indicates a more solid core to count on for turnout. 

Secondly, he does better on robopolls than live-questioner polls.  There's likely a chunk of Trump voters who aren't going to show up in polls, but will vote for him when the curtain is closed.

[sinkspur]

I'm very much against the charlatan, but I think it's important to consider two things. 

Trump does better on LV over RV.  That indicates a more solid core to count on for turnout. 

Secondly, he does better on robopolls than live-questioner polls.  There's likely a chunk of Trump voters who aren't going to show up in polls, but will vote for him when the curtain is closed.

I'd agree, except those same objections were put forth in the primaries, and proved to be wrong.  If anything, Trump underperformed his polls in the primaries.

[Just_Victor]

I disagree.  We thought the same about the Republican primaries:  that Democrats voting would skew the results, that the polls couldn't possibly track all those first-time voters.

Well, the polls in the primaries were mostly on target (the Dem Michigan polls were wrong), especially when it came to predicting the winners.  Margins were off a point or two, but usually underestimated the performance of the winner.

And such will be the case in the general election.  These pollsters do this for a living, folks.

You are correct that the primary models did not differ dramatically for the actual outcome.  But I'm pretty sure that primary voter models are not the same as general election voter models.  Primary voters are a fairly consistent group of party faithful.  When a pollster questions, "Are you going to vote in the primary?" they can expect a black or white answer.  In the general election, we all know that we're supposed to vote, but ~half of us don't.  And who it is that doesn't bother to show up and why (or who show up and refuse to vote for the major party candidates) is going to be very different in this election.  Maybe the pollsters have already figured it all out.  It is their job after all.