It may be time to question the polls that Real Clear Politics include in their averages. Some of these polls are completely unrealistic in my opinion. the latest poll by Reuters/IPSOS is a perfect example. The outfit released a poll today (10/12/2016) showing Hillary Clinton with a 7 point lead over Donald Trump. Theit last poll prior to today was released on 10/5/2016 and showed Clinton with a 6 point lead over Donald Trump. (Note: Both polls include 3rd party candidates Stein and Johnson.)
So looking at these two most recent polls, Clinton was +6 on 10/5 (42% to 36%) and is now +7 over Trump a week later (44% to 37%). And what a tumultuous week it was with the release of a 2005 tape that had Trump using some locker room language and then the debate Sunday night where fireworks flew and most saw the debate as a clear Trump victory.
So the conclusion one might draw from these two polls is that the tapes, obviously a help to Clinton and the debate, a plus for Trump almost canceled each other out. Clinton gained a point.
But wait! Take a look at the internal numbers behind the polls. And for reference, according to the Roper Center at Cornell University, in 2012 38% of the voters were Democrats and 32% were Republicans. 29% were Independents. And according to the same report, 50% of the Independents voted for Romney and 45% for Obama.
Reuters/Ipsos Poll 10/5/2016
On October 5, 2016 which was prior to the release of the Trump tape and prior to the debate, 43% of the people sampled in the Reuters/Ipsos poll were Democrats and 35% were Republicans. Independents were only 13% of the sample. Compared to the actual results of the 2012 election, this poll over sampled Democrats by 11%, they over sampled Republicans by 6% and under sampled Independents by a whopping 16%. And while Romney had a 5% advantage with Independents in 2012, this poll shows Trump with an 11% advantage over Hillary with Independents which dilutes votes for Trump.
And with this poll that I consider seriously skewed towards Democrats, Hillary only has a 6% advantage.
Note that the figures I used include “leaners”.
This poll, besides under representing Independents, also has a +8 skew toward’s Democrats. And when you look at the enthusiasm and massive rallies Trump holds compared to the small croeds Hillary attracts, it is, in my opinion, unrealistic to believe that Democrats will turn out in greater numbers than they did in 2012 for Obama. (Obama was +7 in 2008.)
Latest Reuters/Ipsos Poll 10/12/2016
So let’s take a look at the latest poll by Reuters/Ipsos and see what is really behind the 7% Hillary lead.
This poll samples 47% Democrats and only 31% Republicans. A whopping +16 difference in the Democrats favor and a full 10% more Democrats than Obama drew in 2012.
And astonishingly, they sampled even fewer Independents at 12%! The poll does show Independents go for Trump slightly less than the 10-5 poll with 8%, down from 11%. But in 2012, Independents made up 29% of the electorate.
So this poll adds 4% more Democrats than the previous poll, and takes away 4% of the Republicans sampled in the 10-5 poll.
And in spite of this poll moving from +8% Democrats skew (compared to the 10-5 poll) to a +16% Hillary advantage, she only gains 1% from last week’s poll.
So, forgetting for a minute that pretty much everything in this poll fails the “reality” test, what if they had not added those 8 points to Hillary? What if the poll used the same percentages as the one last week? If we take 4% from Hillary AND add 4% to Trump, the poll would then show Hillary with 40% (not 44%) and Trump with 41% (not 37%).
So, does this poll show Hillary gained a point over last week as it appears? Does this poll show that the Trump tapes hurt the Republican Candidate more than a poor debate hurt Hillary?
And you have to ask yourself, is a poll that samples 8% more Democrats than the previous week actually trying to measure public opinion, or are they trying to do something else entirely?
But the biggest question is why RCP would not insist on some quality control and standardization especially in polls by the same outfit in order to be included in the RCP Average?
The RCP average is used as absolute truth by most news media as a reflection of the state of the Presidential race. And even worse, nobody will hold a polling company accountable for these wild swings in their methods poll to poll. the only one anyone will remember or use to measure which poll does the best job is the one on November 7th. Until then, we are stuck having to look at the numbers behind the results to see if this is a decent poll, or if it is one we must take with a grain of salt – or not at all.
what Real Clear Politics should do is publish the percentage of Republicans, Democrats and Independents used in every poll. And any poll that does not release their internal numbers should be rejected by RCP.
In addition, this poll is an online poll. So one must consider the source. And they admit in their results:
Statistical margins of error are not applicable to online polls.
All sample surveys and polls may be subject to other sources of error, including, but not limited to coverage error and measurement error.
And yet, this poll is part of the average at RCP.