In which the contributors to BS and friends of the contributors discuss data analytics
Dramatis Personae:
Arran Gold, numbers guy; Rebel Yell, chief scientist; Duke of Toronto, financial doomist; Dalwhinnie, Tory squire and Avatar of Enlightenment.
Arran Gold started it by sending the following article to the BS list. It went as follows:
https://www.nationaljournal.com/s/646194?unlock=O0PSAHTAHF7G58Y1
How Analytical Models Failed Clinton
Her campaign was so confident in its data that it opted not to do tracking polls in states that decided the election.
The November elections pitted Democrats against Republicans, conservatives against liberals, Trump-style populists and tea partiers against the establishment and conventional politicians. Another contest, followed mainly by political aficionados, matched traditional pollsters against newly fashionable analytics wizards, some of whom—pretentiously in my opinion—called themselves “data scientists.”
It was well known that traditional polling was having problems. The numbing effect of billions of telemarketing calls and the advent of caller ID and voice mail had reduced response rates (the percentage of completed interviews for every hundred attempts) from the 40s a couple of decades ago to the high single digits. As they struggled to get truly representative samples, pollsters “weighted” their data more than ever before, making assumptions of what the electorate would look like on election days that were weeks, months, or even a year or more away.
Dalwhinnie responded:
I am sure Rebel Yell and his statistician William Briggs will have something to say about this. For my part, the bias of this report is that better numbers could have told Clinton something and would have helped her make better decisions about ad-buys.
No. This election was not won or lost on polls directing ad-buys.
In an important sense, this article doesn’t “get it”. Clinton lost because pride went before a fall, because Democratic votes are excessively concentrated in urban areas , and enough people voted in the right states in the right proportions for Trump to win, and that was because they had someone to believe in.
This is just more shouting at the bar.
Arran Gold responded in his characteristically polite way:
She never went to Wisconsin. The best explanation I have seen is that Putin put a cloaking device on that state so her campaign never saw it. She did go to Arizona and there was even talk of her taking Texas. That is just dumb.
That article was written by Charlie Cook of Cook Political Report who is non-partisan so there is no “get it” because his job is to analyze political campaigns. Is HRC a born loser, yes, but what should she have done differently? That is the question Cook is addressing.
AG
He added:
Here is another example of Clinton’s approach being wrong.
Hillary Clinton’s Vaunted GOTV [ get out the vote ] Operation May Have Turned Out Trump Voters
A focus on big data over people may have backfired.
=====================================
Clinton-was-gonna-lose-anyways doesn’t address the question on how to improve on a data driven campaign strategy next time around.
AG
Rebel Yell then interposed:
Of course their approach was wrong, but it was little to do with data. Look, a chunk of the electorate is going to vote Republican, whatever, and a chunk Democrat, whatever.
http://www.edisonresearch.com/hidden-group-won-election-trump-exit-poll-analysis-edison-research/
The Hidden Group that Won the Election for Trump: Exit Poll Analysis from Edison Research
The story gets even more pronounced when we look at the states that swung the election to Trump. In each of the cases in the table below, the votes gained by people who said: “I don’t like Trump but I’m going to vote for him anyhow” is greater than his total margin in these states. In other words – it was the “Neithers” who pushed Trump over the top in these states and ultimately won him the election.
State | % “Neithers” | Trump | Clinton |
Wisconsin | 22% | 60% | 23% |
Pennsylvania | 17% | 56% | 31% |
Michigan | 20% | 50% | 29% |
Florida | 14% | 61% | 24% |
North Carolina | 15% | 63% | 28% |
The “Neithers” are more likely to be men (61%) and are more likely to be age 30-44 than in the younger or older age groups. They are 78% white, as compared to the total electorate which is 70%. One of the most intriguing aspects of the “Neithers” is that a significant portion of those who were unfavorable to both Clinton and Trump were favorable to President Obama. Nearly half of those who didn’t like either of this year’s two major candidates do have a favorable impression of President Obama – and a significant portion of this group voted for Trump. The 2016 election was unique in so many ways. One distinguishing characteristic is just how many people had an unfavorable impression of both of the major party candidates. To be sure, some of these people decided not to vote for either – Gary Johnson and Jill Stein combined for 18% of the vote among the “Neithers.” However in the end, far more people who liked neither candidate chose Donald Trump and that provided him with his margin of victory in the battleground states.
The last mentioned article is both persuasive and relevant. It asked the right questions, and gets the right answers. I particularly like the fact that many of the Trump votes liked Obama. The same nation elected them both.