My Photo

Ordering Information

Tomi on Twitter is @tomiahonen

  • Follow Tomi on Twitter as @tomiahonen
    Follow Tomi's Twitterfloods on all matters mobile, tech and media. Tomi has over 8,000 followers and was rated by Forbes as the most influential writer on mobile related topics

Book Tomi T Ahonen to Speak at Your Event

  • Contact Tomi T Ahonen for Speaking and Consulting Events
    Please write email to tomi (at) tomiahonen (dot) com and indicate "Speaking Event" or "Consulting Work" or "Expert Witness" or whatever type of work you would like to offer. Tomi works regularly on all continents

Tomi on Video including his TED Talk

  • Tomi on Video including his TED Talk
    See Tomi on video from several recent keynote presentations and interviews, including his TED Talk in Hong Kong about Augmented Reality as the 8th Mass Media

Subscribe


Blog powered by Typepad

« Countdown to No More Trump: 8 Weeks to Go. Hillary up by 2.5%, Leads in 3 of 4 states that Trump has to capture | Main | Debate Preview: Trump vs Hillary: The Championship Bout. Its possible that the election is decided on Monday »

September 20, 2016

Comments

Tomi T Ahonen

Hi Everybody

Gosh this is brilliant. NY Times invited 4 respected pollsters to view their data before it was published and give THEIR count of what the polling data 'revealed'. So this is the SAME data but 4 experts who would conduct the survey and interpret the results. Boy is this revealing... NYT has it here

http://www.nytimes.com/interactive/2016/09/20/upshot/the-error-the-polling-world-rarely-talks-about.html?_r=0

So in addition to the 4, there is the group who DID the polling who also reported, so there are 5 interpretations of the SAME DATA. How much to professionals agree on the SAME DATA - 5 experts, four SEPARATE results, the furthest apart were 6 POINTS apart. One found a +5% for Hillary and the other -1% for Trump. A six point swing from the SAME data, depending on who interprets it.

The average of the five results is +1.75% for Hillary. The 'official' reported finding was +1% that is published and shown for example by RCP. But one respected analyst found from that same data that Trump is ahead by 1% and one found from the same data that Hillary is ahead by 5%. In the same data!

Why? Because of WEIGHTING the sample to expected voter turnout. We do not poll the ACTUAL voters on election day. We poll a tiny sample. A sample which may be correctly balanced male/female but most certainly will not be perfectly balanced by age, or by race, or by education, or by region of the state, or by profession/employed/unemployed/student or marital status or whatever. And THAT has to be done to such data. The un-weighted data is ALWAYS going to be far more old than young, because older people have more time on their hands (retired people) and often lonely, happy to answer a phone and do a survey. The poorest are far less likely to HAVE required communciations to do the survey (but can still vote) or have the time to take a poll (working bad shifts at some night-time job) or afraid to answer the phone (fearing a bill collector) etc.

The polled results ALWAYS have imperfect data on the population that will show up to vote. So the pollster has to do the weigh-ing of the data. What percent of respondents WERE black who responded. What do we EXPECT to be black turnout in this case, in the state of Florida. Etc.

I have a few times been upset at some polls with their samples (the CNN poll recently that was a clear outlier showing Hillary behind nationally in 4-way race). There are many more that I don't bother to bitch about here on the blog but when I see the result, I go dig through the 'cross tabs' and see oh, this guy only sampled half the Hispanic turnout that is expected, no wonder Trump is doing so well - etc; or another that had a turnout model of 50/50 men/women which in all elections is about 47/53 and this election is far more likely to be 45/55 or even more for women.

But its nice to see it 'measured' this way. Four professional pollsters (by the way, the 'Republican' pollster was not the one who got the -1% for Trump. The GOP pollster of the sample got a +1% finding for Hillary, same as the actual result as published but a Democratic-leaning pollster did get the best-Hillary result).

It comes to 'best way is to average polls' to try to eliminate bias. As I said, if those 5 analysts are averaged, they result in a 1.75% finding for Hillary lead in Florida, based on that data. The reported number was 1.0%. The extremes -1% for Trump or +5% for Hillary seem to be outlier opinions of unlikely voter turnout models or ratings.

NOW that said - it DOES matter of course what the turnout is. Republicans want the turnout down. Democrats want the turnout up. A negative campaign usually means suppressed voter turnout and this should become the nastiest campaign ever - although remarkably, so far, it hasn't yet been so. And scaring voters to the polls is perhaps a variation/change to that tactic and both sides are attempting to scare their side to show up. That should help Democrats if it pushes up turnout. BUT it may SKEW the polling - if unusually high numbers of white racist voters turn up (many models have the white turnout up) and whether the Hispanic vote wave emerges (some think Hispanic enthusiasm is down) and if the black surge of Obama votes will pass and be down compared to the last 2 elections. Those all matter to the outcome.

So my professional view - haha, I AM a professional pollster too, by background - is that Youth vote will be down vs 2012. Black vote may be down slightly but will still be above its historic average. Hispanic vote will be up significantly vs 2012. White vote thus will be down at least as % of total vote. The male/female balance will be exceptionally tilted to women. So for voter TURNOUT model, Hispanics & women up. Blacks about same. White vote down, youth vote down.

THEN we get the internal shifts within the model, how do women vote, more for Hillary than voted for Obama? (I think so). Black vote? About same for Hillary as Obama. Hispanic vote, more against Trump than was against Romney. Those produce a wave that there are not enough white male votes to counter and stop the election. BUT note. The POLLS as reported, do not necessarily CAPTURE that insight - because of exactly the issue NYT now proved. Five analysts looking at the same data can find 4 separate results that can be 6 POINTS apart !!! Pennsylvania could be tied. Or Texas could be tied. Thats a 6 point swing if the recent couple of polls from those states were by accident reported by analysts who happened to have a bias that was at the extreme. The more polls we get, the more we can be sure the problems are averaged out. Pennsylvania at a 6% lead for Hillary is PRETTY DARNED safe, so many PA polls are out. But TX a 6% Trump lead, it might be secretly a 3% race that has a few misleading polls out (or Arizona or Georgia) and remember, it goes both ways. Texas could seem like a -6% race and actually be six points THE OTHER WAY, in reality -12% for Trump (or Hillary be up by +12% in PA haha).

Back again to our 50 state model. You see in the model the bolded states. They have 3 or more in-state polls conducted within the month of August (up to first week of Sept). Those we can take as reasonably accurate. The unbolded states, they are less confident and the actual state could be more off than say one or two points.

..isn't this fun?

Tomi Ahonen :-)

tz

Republicans want the high turnout. Mainly the "Monster" (Deplorable) vote. Those who haven't voted in years, but are at Trump rallies which overflow stadiums holding 10k.
Meanwhile, Hillary seems to have 3rd stage Parkinsons, gives brief speeches to dozens not filling high school auditoria.
http://www.dangerandplay.com/2016/09/17/the-enthusiasm-gap-sick-hillary-clinton-donald-trump-polls/

grouch

Reposting:

Do-nothing Congress gets an earful from a fed-up Obama in weekly address
By Susan Gardner

"' The Republicans who run this Congress aren’t doing their jobs.

Well, guess what? Congress recently returned from a seven-week vacation. They’ve only got two weeks left until their next one.'

"President Obama let loose with some sarcasm and a laundry list of issues that need to be addressed at the Republican do-nothing Congress in this morning’s weekly address, leading off with Zika funding, resources for Louisiana after its devastating flooding, and approving Merrick Garland for the Supreme Court. But that was just the beginning."

http://www.dailykos.com/story/2016/9/17/1571091/-Do-nothing-Congress-gets-an-earful-from-a-fed-up-Obama-in-weekly-address

Winter

@tz
"Meanwhile, Hillary seems to have 3rd stage Parkinsons, gives brief speeches to dozens not filling high school auditoria."

Sounds like a plausible Swift Boat attack. Baseless accusations that are difficult to defend against.

grouch

The Many Scandals of Donald Trump: A Cheat Sheet
http://www.theatlantic.com/politics/archive/2016/09/donald-trump-scandals/474726/

176 Reasons Donald Trump Shouldn't Be President
http://www.gq.com/story/176-reasons-donald-trump-shouldnt-be-president-olbermann

How the Trump Organization's Foreign Business Ties Could Upend U.S. National Security
http://www.newsweek.com/2016/09/23/donald-trump-foreign-business-deals-national-security-498081.html

What We Know About Donald Trump's Scandal-Plagued Charity Foundation
http://www.vice.com/read/donald-trumps-charity-foundation-scandal


grouch

Trump used $258,000 from his charity to settle legal problems
By David A. Fahrenthold

https://www.washingtonpost.com/politics/trump-used-258000-from-his-charity-to-settle-legal-problems/2016/09/20/adc88f9c-7d11-11e6-ac8e-cf8e0dd91dc7_story.html

"Donald Trump spent more than a quarter-million dollars from his charitable foundation to settle lawsuits that involved the billionaire’s for-profit businesses, according to interviews and a review of legal documents.

Those cases, which together used $258,000 from Trump’s charity, were among four newly documented expenditures in which Trump may have violated laws against “self-dealing” — which prohibit nonprofit leaders from using charity money to benefit themselves or their businesses."

Tomi T Ahonen

Hi Everybody

Haha the pollsters give and the pollsters taketh away. Today the long enduring pain of Hillary being at less than 300 EV votes - a total of six days at that - came to an end. While Trump picked up a lead in the RCP 'no tossups' map in both Nevada and North Carolina, Hillary took back Florida. She's up to 301 EV votes again. BUT it does suggest right now, the fight is back-and-forth in 5 states (arguably only 4) ie:

Florida
Ohio
North Carolina
Nevada
Iowa (arguably no longer contested, Trump winning this one)

But as we know, even if Trump wins all 5 of those states, Hillary wins the election. Trump needs at least one state more (Colorado or Pennsylvania) or New Hampshire AND the one district in Maine. And Hillary is pretty safe in all those three states.

BTW California, haha, 17% for Hillary. New York, 21% for Hillary. So much for Trump and his original stupid idea of somehow flipping solidly blue states.

Tomi Ahonen :-)

grouch

'I've never encountered anything so brazen': Report details 'shocking' new revelations about Trump Foundation

http://www.businessinsider.com/donald-trump-foundation-2016-9

grouch

Tomi:

Princeton Election Consortium still shows Clinton 296 as of this moment:

http://election.princeton.edu/electoral-college-map/

grouch

There Is So Much Wrong With Donald Trump Jr.'s Skittles Tweet
http://m.motherjones.com/politics/2016/09/everything-wrong-skittle-trump-jr-tweet

grouch

Trump Campaign Offices in the West Bank Offer Perfect Symbol of Trump’s Idea of America
By Ed Kilgore
http://nymag.com/daily/intelligencer/2016/09/trump-campaign-offices-in-west-bank-are-his-idea-of-u-s.html

Wayne Borean


OK, let's go back to Georgia. Tomi says that campaigning to get another 5% of the black vote would not be a good use of campaign resources. But...

What about gaining 10% more of the college educated white vote? That could flip the state, especially with the backing of Jimmy Carter, the former president who is still popular in Georgia. Carter also has an impact with Evangelicals.

So yes, I think Georgia could be blue. The changing demographics, and the unpopularity of the GOP candidate could swing it. Could already have swung the state.

Clinton would not have bothered to have any staff in the state if it was deep red, like California is deep blue, but she does have staff there. If you look at the RCP polling for Georgia it indicates that the state was in Clinton's hands for most of August, and has swung back to Trump. Given the volatility shown in polling, a swing back to Clinton would not be surprising.

http://www.realclearpolitics.com/epolls/2016/president/ga/georgia_trump_vs_clinton-5741.html

Winter

It is clear that the PotUS elections in the US are decided on voter turn out. And voter turn out is what is evidently worst predictable.

I remember how Tomi analyzed Obama's get out the vote efforts as decisive for the land slide results.

Winter

HuffPo warns us not to expect anything from the debates, at all:

There's No Debate
http://www.huffingtonpost.com/bill-moyers/theres-no-debate_b_12090380.html

Asko

@Tomi
It'd be nice if you could visit the iPhone 7 launch blog post comments as we'd love to hear your view on calendar Q4 sales expectations.
(feel free to delete this comment when you do.)

talvi

Democrats' debate advice to Clinton: Let Trump screw up

http://www.politico.com/story/2016/09/clinton-debate-advice-228431

Winter

@talvi
It is always a bad idea to let the initiative slip. The best advice is to MAKE Trump screw up.

talvi

@Winter

I agree. The danger is in this debate is that Hillary might win the debate at technicalities which will not be that obvious ordinary people. Therefore Hillary must win clearly the debate(s) because a draw is basically a victory for Trump who is seen as the underdog.

I guess that Trump's tactic will be to interrupt as much as possible Hillary's talking or even talk over her (that is basically bullying Hillary) and throw those punch-lines (which I find very stupid but many people like them).

Hillary's job is actually pretty difficult because of "how do you debate with an smart-ass who does not follow rules of logic?". I guess that probably Hillary's strategy is to bury Trump with facts whilst playing the "teacher's role" (by pointing his grammar errors, pointing to his repetitions, etc.). Rubio has tried a little bit of this but he was pretty bad at it.

talvi

So the best advice for Hillary is to get Trump angry. A shrink might have some very good tips here.

Winter

@talvi
"So the best advice for Hillary is to get Trump angry."

That is the real price. If Hillary get Trump to lose his temper, he is gone. On the other hand, if Trump succeeds in making look Hillary as "weak" and "lacking energy", it is the end for her.

The comments to this entry are closed.

Available for Consulting and Speakerships

  • Available for Consulting & Speaking
    Tomi Ahonen is a bestselling author whose twelve books on mobile have already been referenced in over 100 books by his peers. Rated the most influential expert in mobile by Forbes in December 2011, Tomi speaks regularly at conferences doing about 20 public speakerships annually. With over 250 public speaking engagements, Tomi been seen by a cumulative audience of over 100,000 people on all six inhabited continents. The former Nokia executive has run a consulting practise on digital convergence, interactive media, engagement marketing, high tech and next generation mobile. Tomi is currently based out of Helsinki but supports Fortune 500 sized companies across the globe. His reference client list includes Axiata, Bank of America, BBC, BNP Paribas, China Mobile, Emap, Ericsson, Google, Hewlett-Packard, HSBC, IBM, Intel, LG, MTS, Nokia, NTT DoCoMo, Ogilvy, Orange, RIM, Sanomamedia, Telenor, TeliaSonera, Three, Tigo, Vodafone, etc. To see his full bio and his books, visit www.tomiahonen.com Tomi Ahonen lectures at Oxford University's short courses on next generation mobile and digital convergence. Follow him on Twitter as @tomiahonen. Tomi also has a Facebook and Linked In page under his own name. He is available for consulting, speaking engagements and as expert witness, please write to tomi (at) tomiahonen (dot) com

Tomi's eBooks on Mobile Pearls

  • Pearls Vol 1: Mobile Advertising
    Tomi's first eBook is 171 pages with 50 case studies of real cases of mobile advertising and marketing in 19 countries on four continents. See this link for the only place where you can order the eBook for download

Tomi Ahonen Almanac 2009

  • Tomi Ahonen Almanac 2009
    A comprehensive statistical review of the total mobile industry, in 171 pages, has 70 tables and charts, and fits on your smartphone to carry in your pocket every day.

Alan's Third Book: No Straight Lines

Tomi's Fave Twitterati