Monday, September 19, 2016

Minnesota, migrants, and a migraine for Hillary

All of the the weekend's attacks are rib kicks to Hillary as she tries to pick herself back up off the phlegm-covered ground--New York because it's New York, New Jersey because it's within Trump's reach and along with a flipping a smaller state like Iowa or New Hampshire from 2012 would allow him a clear path to electoral victory without having to win Pennsylvania, but most of all Minnesota because it packs so many transgressions against the CultMarx narrative into a single incident.

Consider, Dahir Adan, the attacker, apparently asked at least one of his victims whether or not he was Muslim before stabbing him. A Muslim terrorist targeting Midwestern-nice white Christians? It's supposed to be Core America oppressing Fringe America, not the other way around. 

The attacker struck from a distance so close to his first victims that the only possible defense said victims would have been able to utilize was to profile him ahead of time. A young black guy of east African descent, probably Muslim? If you're aware of your surroundings, he should be on your radar as soon as he comes into view. The Talk could've saved some unfortunate mall-goers some pain and suffering.

He wielded a knife. While not necessarily as deadly as a gun--none of the victims died, after all--it's psychologically even more horrifying for the victims experiencing it than a firearm is. There's a reason why slasher horror movies involve antagonists armed with hand-wielded sharp objects instead of projectiles.

The attack was stopped by an off duty cop who shot the Muzzie dead. This incident will go down as a single fatality from a firearm. 

Taken in isolation, of course, that stat gets the story exactly backwards. The application of the second amendment kept this incident from being as deadly as it otherwise would have been. These scenarios don't show up in the statistics (or if they do, they make the US look more violent rather than less so, ceteris paribus), allowing misleading and unfavorable comparisons to be made between the US and [insert Northwestern European country] when it comes to violence, particularly gun violence. 

But the reason the US is so homicidal in the aggregate is because the US has so many people of African--and, Amerindian, and increasingly MENA--descent inside its borders. White Americans are as well behaved as other Northwestern Europeans are.


Cicatrizatic said...

Good observations. Another case of private gun ownership to the rescue - not surprisingly, the media avoided that angle entirely.

I had forgotten that MN was just a single digit margin in the last election. A MN poll released last week has Clinton at +6.

Rasmussen has a new NV poll out today: Trump +3.

Audacious Epigone said...


Good point re: Minnesota. The same potential dynamic applies here as in New Jersey, although I think the latter is more winnable for Trump than Minnesota is.

Anonymous said...


O/T question:

Is this an example of a "Push Poll"?

At the bottom of the article it states poll results were based on a nonprobability survey. I Googled the term and found this:

I think NBC's purpose is to a) discourage Trump voters and depress their turnout and b) flip the votes of sub-sentient voters having a compulsive need to vote for the winner. In other words, given the recent hits Hillary has taken in the polls, the MSM wants to re-establish the "Hillary Is Inevitable" meme.

If true, then why would they admit to using a nonprobability survey? Do they count on people not reading the fine print?

Just curious, if you have time.

Jokah Macpherson said...

"Do they count on people not reading the fine print?"

That's rhetorical, right?

Anonymous said...

"That's rhetorical, right?"


TangoMan said...

Sean Trende wrote an excellent 4 part essay on the Sailer Strategy. Here is a graph on Minnesota voting patterns over time. Everyone reads about how states turn purple and then blue, but not much analysis on how some states are turning red.

Nineteen states have moved at least a point toward Democrats, while 25 have moved toward Republicans by a similar amount. If you weight the shift in each state by electoral vote, it actually works out to a slight shift toward Republicans overall.

Cicatrizatic said...

I'll be interested to see AE's thoughts regarding the SurveyMonkey poll.

Note that the SurveyMonkey poll is an opt-in survey of people who log on to the SurveyMonkey platform and choose to take the poll. I think it is a poor methodology, and generally unrepresentative. Why would the universe of SurveyMonkey users be representative of the general voting public? Note that YouGov uses a similar methodology.

In SurveyMonkey's latest poll results (from 9/12 to 9/18), Clinton gained 3 points from the previous week. During Clinton's worst week yet, where other tracking polls had Trump gaining 3-5 points, and the state polls swung violently to Trump, they have her gaining. Note that generally, this is typical of the SurveyMonkey poll. It is non-responsive to general trends that the majority of other polls pick up. That's a huge red flag.

Note that opt-in panels like LA Times/USC and People's Pundit Daily create a panel of possible voters to continuously poll - but they gather the panel from the general public, they don't do it by just accepting people who show up to a website and opt-in.

SurveyMonkey also asserts that the MOE on their poll is 1.2. A margin of error is impossible for an opt-in, non-random poll. Opt-in panels like LA Times/USC give a confidence interval, not a margin of error. This alone should disqualify SurveyMonkey from being included in RCP. Even Nate Silver grades them at a C-

Cicatrizatic said...

Another note on poll trends this week:

Trump has declined two days in a row in the LA Times poll, but this is entirely due to his % with blacks crashing back to normality. It had reached 20%, but in last two days has fallen to 6%.

So the sharp two-day decline in the top-line result is a little misleading. It's really just the black vote correcting itself.

The good news is that his % with whites - easily the most reliable cross-tab in the entire poll due to its large sample size - has been steadily increasing for 10 days. He is now at 56.5% to Clinton's 32%, a 24% margin that is better than Romney's margin (20%) and that will likely increase further in the next 6 weeks.

Audacious Epigone said...

Cicatrizatic is a much better source for understanding the vagaries of polling samples than I am. I have no idea how or why RCP allows the NBC/SurveyMonkey poll to show up as having the smallest margin of error of all the polls considered when, as he says, that's impossible to calculate given the methodology used. It's an opt-in and the distribution is based on Census demographic data rather than actual likely voter variables. The sample size is large but the self-selection problems aren't diluted at all from that.

The LA Times/USC poll has a similar opt-in problem, but the user base is constant. The benefit there is that it gives us a window into which way the electoral winds are blowing. We still don't know what the true starting point is though. LA Times/USC leaves me feeling reasonably comfortable that the Trump-Clinton matchup has gotten about 7 points better for Trump over the last month. But I have very little confidence in where the starting point a month ago or the current spot now are. Hillary up by 5 and now down by 2; Hillary up by 12 and now up by 5, Trump up by 1 and now up by 8? Beats me.

The NBC/SurveyMonkey poll looks similar. It's always among the worst for Trump, but the latest iteration is better for him than a month ago.

From SurveyMonkey's notes on methodology:

The NBC News|SurveyMonkey Weekly Election Tracking data for the week of September 12, 2016 through September 18, 2016 was conducted online among a national sample of 13,230 adults aged 18 and over who are likely to vote. Respondents for this survey were selected from the nearly three million people who take surveys on the SurveyMonkey platform each day. Data for this week has been weighted for age, race, sex, education, region, and voter registration status using the Census Bureau and Bureau of Labor Statistic’s Current Population Survey to reflect the demographic composition of the United States.

An additional smoothing parameter for ideology based on previous Election Tracking interviews is included. The smoothing parameter was derived from aggregated data on ideology by region collected from previous weeks (May 1-June 26) of the Weekly Election Tracking poll. Because the sample is based on those who initially self-selected for participation rather than a probability sample, no estimates of sampling error can be calculated. All surveys may be subject to multiple sources of error, including, but not limited to sampling error, coverage error, and measurement error.

To assess the variability in the estimates and account for design effects, we create a bootstrap confidence interval to produce an error estimate. The bootstrap confidence interval for this survey is plus or minus 1.2 percentage point for likely voters.

Audacious Epigone said...


I'd forgotten about that. The white decline from '08 to '12 was disproportionately due to Trump voters not voting in '12. Trende presaged Trump without intending to.

Anonymous said...

i have a simple question.

according to reuters/ipsos and la times, trump is losing among college educated whites by a substantial margin. if you plug a clinton win among this demo into nate silver's "swing o matic", even a small win of 51-49, while leaving all other voter splits and turnouts the same, Clinton is ahead in the popular vote by ~10%. if you crank non-college educated white turnout up to 75%, and shift the black turnout closer to its 2000-2008 levels, you can get Clinton down to ~+6 in the popular vote, but there are no plausible scenarios where Trump is winning the popular vote if he's losing among college educated whites. it isn't even close.

so, what gives?

Cicatrizatic said...


Well, it largely comes down to turnout levels across all categories. Very likely that college white turnout will be down, and non-college white turnout will be up. They may end up being close to even in turnout levels.

As you note, black turnout will probably be less.

I think Trump may do marginally better with Hispanics than Romney did. While Trump is hovering around 25-33% with Hispanics - right where Romney was - Hillary is only at 50-60%, far less than Obama's 71%. For example, in the Gravis poll released today, Hispanics are 50% Clinton, 33% Trump. I think Trump will end up with about 30-33% Hispanics, slightly higher than Romney.

Lastly, although they are only 4-5% of the vote, the Asians/Other category is decently close, although the cross-tabs on this demographic have a huge MOE. It looks like the Asian vote will be 60D/40R this time, as opposed to 70D/30R, like last time.

If you plug these turnout levels and splits into the 538 module, I'm guessing you'll get something close to a tie.

Anonymous said...

thanks for your reply.

but as mentioned in my original post, even if you crank non-college educated white turnout way up to 75%, a massive and implausible increase (57% in 2012), and lower black turnout to 55% and give Trump 10% of 'em, and increase GOP support among Latinos, Clinton is still +4 in the popular vote when college educated whites lean ever so slightly Dem. he *needs* that demo to come back to the GOP or he cannot win.

still, there are polls where Trump is +5 while losing among college educated whites by 10%+. what sort of voter turnout are these pollsters predicting?

Audacious Epigone said...


In '12 those without a college degree made up a majority of voters. Reuters-Ipsos, however, surveys more college-educated respondents than non-college educated ones which seems suspect given Trump's appeal relative to Romney's (and Hillary's appeal relative to Obama's for that matter, at least among non-blacks).

As a rough sketch, Trump wins 2:1 among non-college whites and he and Hillary draw among college whites. That's 70%+ of the electorate there, with Trump up. That's Trump at roughly 42% and Hillary at 30%. Hillary gets 4/5ths of the remaining 30%- non-white, putting her at 52% and bumping Trump up to about 48%, which is right where the R-I poll is with the undecideds/won't vote removed.

Am I missing something?

Cicatrizatic said...


In addition to increasing non college white turnout, you need to increase their % for R from 62% to 67-70% R.

Mil-Tech Bard said...

Audacious Epigone,

This is a change election.

All the structural markers have pointed to the party out of power taking the Presidency away from the incumbant presidential party. Example -- in the last three elections the GOP has taken pretty much every non-presidential Federal office and most state level offices that it is demographically possible to take.

Hell, the Bi-partisan Fat Cat Oligarchs cleared the field of viable Democratic candidates against Hillary and a nut-ball independent socialist from Vermont got a 43% protest vote.

These same Bi-partisan Fat Cat Oligarchs had the GOP primaries wired to splinter the conservative vote and get Jeb.

Epic Fail, that.

Trump walked right in and "flipped the script" by uniting the majority of the 45% 'change caucus' early in the GOP primaries against a fragmented "Fat Cat" candidate field until it was to late in February this year.

Frankly, it was Trump's full throated roar against Muslim Terrorism and tying it to open borders immigration that sank the rest of the GOP field. The other
16 candidates attacking Trump for saying the non-PC truth on Muslim immigrant behavior marked the rest of the field as "establishment candidate in a change election" politicans.

It was that leadership that branded Trump as "The most change candidate" available and won him the GOP nomination.

Effectively, everything the political campaigns do or don't do since the conventions means bupkus. Trump is in the structural election cat bird seat. He would have to be assasinated not to win.

The real issue here isn't Trump's impending victory. It is that the issues up for public debate after his election are all Fat Cat vital interest money issues that they have bought the bi-partisan political class over the last 50 years to keep out of the public eye.

The Congress and the Federal Courts will do their best to hamstring Trump doing anything effective about Muslim terrorism because the only way to deal with it effectively is to mass deport non-citizen Muslims and cut off family reunification immigration laws for any Muslim citizens.

If you do that with violent radical Muslims immigrants, you can do that to violent Mexican, or any other immigrant group.

This means that the populist wave of this elect isn't going away.

The people ticked off and winning with Trump will come back again and again until enough politicos lose despite the Fat Cat money. And their replacements listen to voters, not money.

All Trump has to do to destroy the Bi-partisan Fat Cat Oligarchs is keep trying to do effective things while Muslim immigrant terrorism rises, and damn the bought and paid for politicians, bureaucrats and judges as corrupt bloodyhanded killers of American citizens.

This will keep the change caucus turning out in disproportionate numbers until the politic class listens.

It will be a war of attrition of money interests versus a populist middle/working class on the vital interests of both.

Mil-Tech Bard said...

Via Scott Adams, with a good reason why Trump went down with Blacks in the daily trcking polls --

Trump’s “stop and frisk” comment will haunt him for a few weeks, but it comes in the context of outrage about an African-American policeman killing an African-American citizen. According to the pundits on TV, that changed the frame from a problem of white-versus-black to a question of police training. And even Trump is asking why the police shot a man in Tulsa who appeared to be surrendering.

Trump is consistent in staking out whatever is the most bad-ass sounding position on all matters of security. Later, following his well-observed pattern, he negotiates down to something that doesn’t violate the Constitution so much. So I wouldn’t worry about “stop and frisk” becoming a thing. States will figure out that stuff on their own.

Trump is -not- playing election polics here. He is laying markers for a change of governance after election day.

IOW, he is making people 'think past the sale.'

Audacious Epigone said...

Mil-Tech Bard,

Great analysis regarding the Republican primaries. By playing as Trump's wingman early on in the first few debates and in a joint rally, Cruz was able to stay standing while all the other perceived insider candidates got mowed down. Cruz got too ambitious though, and he to ended up going down. It was after the Northeast went and he struck the corrupt deal with Kasich in an attempt to win Indiana that he was abandoned by his anti-Establishment supporters and the nomination was Trump's.

What I'm worried about: Vote rigging. Who in the political class is going to try to stop it? Who is going to take allegations of it seriously? I can see a bipartisan agreement to simply steal the election from Trump going down. I'm not talking about voter fraud, either--that only matters at the margins--I'm talking about electoral fraud, the actual vote counting.

Mil-Tech Bard said...

Audacious Epigone,

There was not enough advanced planning for that to be pulled off.

The Bi-partisan Fat Cat Oligarchs didn't plan for Trump.

Bi-partisan Fat Cat Oligarchs could not agree on an alternate GOP candidate to Trump until it was far too late.

And they didn't take Trump seriously enough as a General Election candidate after the conventions.

As a result, the partisan functionaries just don't have a bi-partisan "go to hell plan" ready to pull that sort grand election steal.