Wednesday, August 3, 2016

Using Google Search Results as a forecasting tool

OK, I think this popped on twitter. Donald, You Ignorant Slut by Joe Bob Briggs.

It is a reasonably well written, though long, restatement of the objective observation that the media leans far left of the public and that that leaning, whether intentionally or not, shapes both what news is covered and how it is covered. Briggs has a spirited writing style intended to assuage the offended and offend the assuaged. He starts with:
NEW YORK—All over America IKEA futons are groaning with the restless insomnia of journalists—tossing, turning, cursing the impotence of their melatonin capsules—burdened with the future of the Republic. Long nights of torment, and then . . .

Morning resolve! Before they’ve even microwaved their second Jimmy Dean Sausage Sandwich, they know that this will be the day of reckoning. They will fire up the Kia Sedona and take the long way to work, giving them more time to think about the epic 1,500 words that will make the difference between chaos and civilization.

Yes, they tell their wives, It’s time for my “Donald Trump is a Dickwad” column.

Let me make it clear here that I’m not talking about lesbian-rights vegans who organize fair-trade coffee boycotts at Maxwell House and agitate for medical marijuana in The Nation. Nor am I thinking of tweed-jacketed professors of sociology at Montana State submitting articles to the Journal of Spanish-American Diacritical Marks. Think-tank analysts at the Institute for Pan-Arab Non-Alignment are most certainly churning out white papers on why Donald Trump is a dangerous threat to the Maghreb treaty on fish hatcheries, but I’m not discussing them either. I’m not talking about intellectuals or activists or experts.

No. I’m talking about the guy who enrolled at McNeese State in the nineties and fell into deep reverence for Professor Rusty Naugahyde, the legendary teacher whose Newswriting 312 workshop was almost as inspirational as Lou “The News Is Sacred” Grant on The Mary Tyler Moore Show. Our starry-eyed undergraduate buys a safari jacket, takes an oath of objectivity and resolves never to be a member of a political party. After that it takes years of struggle to become the lead Metro columnist at the Lotus Tree, Kansas, Daily Arapaho, days spent chronicling the brutal fights over county bond issues needed to repair the Lost Frenchman Bridge. But now that he’s a 32nd-degree Mason and chairman of the Little League committee on maintenance and parking, he knows that it’s his responsibility, and his privilege, to tell the people of Lotus Tree that Donald Trump is a narcissistic disagreeable soulless callous rude arrogant authoritarian vicious egotistical vulgar braggart and megalomaniac, possibly a lunatic, definitely a psychopath, perhaps a fascist.

This goes against everything in the journalism rule book. Elections are the ultimate on-the-one-hand, on-the-other hand story. You get the League of Women Voters to interview each candidate and then you make sure every biography is exactly the same length as the one before it and the one after it. You never say that one campaign platform is better or worse than another campaign platform. You occasionally note “controversy” over “remarks perceived to be off color,” but you limit your commentary to observations like, “It now appears that Beaver County Sheriff Judd will bring his traditional voting bloc to the side of Culpepper, while District Judge Monahan will side with the Democratic challenger.” The last thing you ever do is suggest, much less state, that someone is a pathological liar, because there’s a strong likelihood that more than 50 percent of the people you write about in the course of a lifetime will be, in fact, pathological liars.
All this is the set-up to Briggs real question and topic:
What is it about Donald Trump that makes journalists go insane?
A fair question.

In addition to his energetic and colorful writing, though, Briggs throws down some data. It isn't especially scientific or rigorous. But we already have reasonably robust academic research supporting his main contentions. Briggs instead proffers much more common sense data that is very easy to engage with and which his readers are likely to easily replicate. Number of Google Results. What is the Google Search Results count when you pair "Donald Trump" with a range of insults such as "insane."

He googles a large number of insulting and vituperative insults heaped on Donald Trump by journalists. The numbers are very large in terms of results. He starts out with a list of forty:
puerile 85,000 (not a popular term in the Midwest)
boorish 115,000
soulless 240,000
patronizing 248,000
egotistical 253,000
megalomaniac 261,000 (this one would score higher if it were not six syllables, therefore rendering it unusable in headlines)
vengeful 304,000
callous 395,000
grandiose 406,000
combative 407,000
birther 425,000
condescending 452,000 (could be combined with patronizing for a higher score)
mean-spirited 453,000
misogynistic 479,000 (misleadingly low because “woman hater” gets over a million)
foul-mouthed 500,000
disagreeable 503,000
arrogant 509,000
lunatic 524,000
immature 525,000 (should probably be combined with puerile)
xenophobic 532,000
fascist 554,000
authoritarian 571,000
braggart 586,000
obnoxious 606,000
narcissistic 635,000
boring 643,000 (wow, this one really shouldn’t make it past the fact-checkers)
haughty 645,000
obsessive 707,000
superficial 713,000
psychopathic 784,000
thin-skinned 785,000
vulgar 967,000
It's an interesting approach but not particularly meaningful on its own. First, you need to compare these numbers with the associated count for Hillary Clinton. For example, here are the results for both of them on the first few insults.
puerile - DT 85,000; HC 113,000
boorish - DT 115,000; HC 81,000
soulless - DT 240,000; HC 184,000
patronizing - DT 248,000; HC 233,000
egotistical - DT 253,000; HC 152,000
I did this for a dozen or so and generally, they are within 30% of each other. I am not sure why Clinton would be seen as more puerile than Trump, but thems the numbers. It makes much sense, at least to me.

Briggs settles on eight terms with the biggest numbers
bully 1.4 million
self-obsessed 2.5 million
vicious 9.1 million
rude 13.3 million
cruel 13.3 million
liar 16.2 million
angry 19.3 million
And the winner—drum roll, please:
idiot 20.5 million
OK, let's do the same thing again, this time with Hillary Clinton's numbers
bully - DT 824,000; HC 720,000
self-obsessed - DT 330,000; HC 320,000
vicious - DT 573,000; HC 566,000
rude - DT 808,000; HC 843,000
cruel - DT 823,000; HC 815,000
liar - DT 11.1 million; HC 10 million
angry - DT 46.1 million; HC 34.6 million
idiot - DT 22.8 million; HC 1.1 million
People seem to think that Trump is much more of an idiot than Clinton but, interestingly, that Clinton is much angrier than Trump.

All very interesting, but Briggs' original contention is that the mainstream media journalists are far more insulting of Trump than Clinton. General search will capture all sorts of blogs and other non-MSM materials. Fortunately you can Google search just the News. Here are the respective numbers for just the media.
bully - DT 299,000; HC 640,000
self-obsessed - DT 5,750; HC 3,750
vicious - DT 320,000; HC 180,000
rude - DT 486,000; HC 421,000
cruel - DT 543,000; HC 497,000
liar - DT 347,000; HC 262,000
angry - DT 3.4 million; HC 855,000
idiot - DT 272,000; HC 184,000
There's a lot of very interesting things going on with these numbers but it is hard to see an overarching pattern.

In general searches, Trump draws 70% more insults than does Clinton. Among journalists, however, he draws 86% more insults than does Clinton, supporting Briggs' contention that journalists hate Trump, perhaps not insanely, but at least more than the public at large.

The score for journalists associating Clinton with bullying is 89% of that of the public, i.e. the public and the MSM seem to see pretty eye-to-eye on that charge. There is a wide divergence between the public and the journalists on most of the Trump items. The closest similarity is on the charge of cruelty, journalist see Trump at 66% of the rate as do the public.

My inclination is that there is not much that can be concluded from these numbers - it is too crude a test. However, the overall results seem to conform with both Briggs argument and the academic research as well, the media views Republicans with a more jaundiced eye than do the public.

I did one more final test. Brigg looked at negative charges. What about positive attributes? Here are the results for the at large searches:
Kind - DT 90.5 million; HC 252 million
Strong - DT 75.9 million; HC 65.9 million
Generous - DT 1.5 million; HC 703,000
Honest _ DT 25.8 million; HC 23.9 million
Leader - DT 85.1 million; HC 72.3 million
Trustworthy - DT 502,000; HC 541,000
And for the mainstream media:
Kind - DT 12.4 million; HC 60.4 million
Strong - DT 9.5 million; HC 8.7 million
Generous - DT 501,000; HC 304,000
Honest _ DT 816,000; HC 710,000
Leader - DT 16.1 million; HC 9.3 million
Trustworthy - DT 39,300; HC 60,500
Well that's an interesting alternative perspective. The general public sees Trump as stronger, more generous, more honest, more of a leader than Clinton. And in all those instances, the mainstream media agrees with them. Clinton is seen as kinder and more trustworthy than is Trump, and again, the media agrees.

If the election hinges on negative attributes, Trump attracts more opprobrium and Clinton will take it. If the election hinges on positive attributes, Trump will take it.

If you believe that Google Search Results is a reliable forecasting proxy. Which I don't.

No comments:

Post a Comment