ABC election poll is off the charts and should be excluded from list of Real Clear Politics’ results

hillary-pollsThe average of all polls measuring voter sentiment in the 2016 Presidential race, except ABC, showed former Secretary of State Hillary Clinton ahead today of Donald Trump by 45.3 to 39.9, or a spread of 5.4. The poll results are updated daily by Real Clear Politics.

For the past two days ABC has reported a spread of 12 points – more than twice the average of all the other polls.

Examining the ABC methodology for interviewing potential voters, one thing stands out – an apparently odd decision to ignore middle age and senior voters in favor of the young.

Numbers are called multiple times during the field period in multi-night polls; the standard for full-length ABC/Post polls is a minimum of six calls to each number. Interviews are conducted via a computer-assisted telephone interviewing (CATI) system. SRBI’s professional interviewers, and their supervisors, are extensively trained in interviewing practices, including techniques designed to achieve the highest possible respondent cooperation. For landline respondents, interviewers ask to speak with the youngest male or youngest female at home. Cell-only respondents are screened for age eligibility (18+). Cell-only respondents are not offered compensation, but a reimbursement check is offered if use of minutes is raised as an objection. Cell sample respondents’ place of residence is checked and their Census region adjusted accordingly if necessary.

ABC conducted this survey in both English and Spanish and asked several questions about Trump and Clinton controversies:

(ASKED OF LIKELY VOTERS) Do you approve or disapprove of how [ITEM]? Do you feel that way strongly or somewhat?

  • Trump is handling questions about his treatment of women
  • Clinton is handling questions about her use of personal email while she was secretary of state?
  • Trump has said he thinks the election might be rigged so that Clinton could be elected through vote fraud. Do you think (this is a legitimate concern), or do you think Trump is trying to make excuses in case he loses the election)?
  • Trump has refused to say whether or not he would accept the election’s outcome if Clinton wins. Do you approve or disapprove of Trump indicating that he may not accept the election outcome?

smug-mugI’ll leave it up to the reader to decide if a total of three out of the four questions on “policy” for one candidate was appropriate, and if some of the questions were assumptions or push polling. One thing is for sure – if respondents were given a choice to change their vote after those questions, many would.

Overall, those polled said that Clinton won the third debate by 52 – 29, which is far from other polls showing it tied or Trump ahead.

What difference does it make that the ABC poll is skewed? The following is the current polling chart by Real Clear Politics, allowing newscasters to toss around a 5.4 lead by Clinton as near proof the election is over.

Polling 25-Oct
Date Sample MoE Clinton Trump Clinton Ahead
RCP Average 10/13 – 10/24 45.30 39.90 5.4
Rasmussen 10/20 – 10/24 1500 LV 2.5 43 42 1
IBD/TIPP 10/19 – 10/24 815 LV 3.6 42 41 1
CNN/ORC 10/20 – 10/23 779 LV 3.5 49 44 5
ABC News 10/20 – 10/23 611 LV 4.5 50 38 12
NBC News/SM 10/17 – 10/23 32225 LV 1 46 41 5
Quinnipiac 10/17 – 10/18 1007 LV 3.1 47 40 7
Economist 10/15 – 10/18 925 RV 3.9 42 38 4
FOX News 10/15 – 10/17 912 LV 3 45 39 6
Bloomberg 10/14 – 10/17 1006 LV 3.1 47 38 9
Reuters/Ipsos 10/13 – 10/17 1190 LV 3.3 42 38 4

By simply removing the  ABC poll, the total spread in favor of Clinton drops to 4.7.

Polling October 25, 2016 Without ABC poll
Date Sample MoE Clinton Trump Clinton Ahead
RCP Average 10/13 – 10/24 3.00 44.78 40.11 4.7
Rasmussen 10/20 – 10/24 1500 LV 2.5 43 42 1
IBD/TIPP 10/19 – 10/24 815 LV 3.6 42 41 1
CNN/ORC 10/20 – 10/23 779 LV 3.5 49 44 5
NBC News/SM 10/17 – 10/23 32225 LV 1 46 41 5
Quinnipiac 10/17 – 10/18 1007 LV 3.1 47 40 7
Economist 10/15 – 10/18 925 RV 3.9 42 38 4
FOX News 10/15 – 10/17 912 LV 3 45 39 6
Bloomberg 10/14 – 10/17 1006 LV 3.1 47 38 9
Reuters/Ipsos 10/13 – 10/17 1190 LV 3.3 42 38 4

To properly determine the current status of the election, those polls ending more than a week ago should be eliminated, because they are outdated. When we do that, the spread in favor of Clinton drops to 3.8 with a 2.74 margin of error (MOE).

Polling October 25, 2016 Without ABC poll or old polls
Date Sample MoE Clinton Trump Clinton Ahead
RCP Average 10/17 – 10/24 2.74 45.40 41.60 3.8
Rasmussen 10/20 – 10/24 1500 LV 2.5 43 42 1
IBD/TIPP 10/19 – 10/24 815 LV 3.6 42 41 1
CNN/ORC 10/20 – 10/23 779 LV 3.5 49 44 5
NBC News/SM 10/17 – 10/23 32225 LV 1 46 41 5
Quinnipiac 10/17 – 10/18 1007 LV 3.1 47 40 7

Finally, NBC is projecting a 1% MOE, which on the face of it seems ridiculous and impossible. By adjusting that MOE to a more reasonable number of 2.5, we end up with still a 3.8 spread but a 3.04 MOE. Using that methodology, the candidates are either ahead or behind by less than 1% from the margin of error.

In my opinion the race is far from over for either candidate, and my conviction remains that numbers can be shifted many ways for different results.


Also see: New Podesta Email Exposes Playbook For Rigging Polls Through “Oversamples”

One response

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s

%d bloggers like this: