Exploring economics, politics, technology, mathematics, philosophy, and psychology.

The Problem with Polling

The Problem with Polling

This isn't about the 2016 election. That polling, often denounced by Trump as a conspiracy against him by the media, did represent some systemic problems in the Midwest (Pennsylvania, Michigan, and Wisconsin), but was only a few points off. In the larger context of the election, these polls matched up with other errors to balance out the country's popular vote.

Other polling, though, is so outrageously wrong that it really does seem conspiratorial. In Michigan's Presidential Democratic primary, polls projected Clinton to win by anywhere from 13-37 percentage points. She ended up losing by 1.4%, or 17,000 votes.

Similarly, in the NY-14 Democratic primary, Joe Crowley's campaign sanctioned an internal poll that showed him leading Ocasio-Cortez by 36 points just three weeks before the primary vote. She certainly didn't flip the district 51 points in those three weeks, although it would be nice to believe. The polling must have been incorrect.

There are plenty of obvious problems with polls that everyone recognizes. They tend to center around phones and polling companies often arbitrarily decide which geographic areas they will poll. They also focus on adults who voted in the previous election, even when data suggests that might not be the best determiner of someone voting in the next one.

This is the key to Bernie's and Ocasio-Cortez's wins in Michigan and NY-14. They didn't only appeal to previous voters, but more importantly, to ones who hadn't voted before.

Time and time again, data suggests that the United States' ungodly low voting numbers come from a combination of low political efficacy and working class suppression. Voters are forced to make a mathematical calculation about voting in which they ask themselves: "Is it worth the drive, waiting in line, taking the day off work, and potentially having my vote thrown out to elect these candidates?"

For candidates like Ocasio-Cortez, Bernie Sanders, and Abdul El-Sayed, voters should decidedly answer their own question with "YES."

And they did, for the former two.

The most recent poll (it's worth reading and not very long) placed Abdul El-Sayed more than 20 points behind Gretchen Whitmer, a centrist Democratic candidate, and two points behind Shri Thanedar, a farcical democrat who didn't even know which party he would run in. Going through the results and process, this poll is deeply flawed, but not any more than others.

First, before getting to the questions about the primaries, the poll asked participants about Whitmer-Schuette and Thanedar-Schuette potential general election matchups. Not only did the poll leave out El-Sayed, but simply mentioning the names of the other two candidates put them ahead of El-Sayed in participants' minds.

This type of leading question is always discouraged in polling spheres because it's known to have extreme effects on the results.

After that, the poll asked another Whitmer-centric question, proposing that she is the most anti-GOP candidate (it just so happens that anti-GOP is very popular with Democrats right now):

whitmer poll.JPG

Again, this can have a serious effect on the following questions about Democratic primary preferences, and we'll never know how much this could've changed the results of the election (Whitmer is using this poll as part of her ad campaign).

The last fatal flaw with this poll is buried in the numbers. Every good poll asks questions about race, gender, and age and balances out the numbers for demographics if there's a large disparity. Here are the ages of the group they polled:

gov age.JPG

Most don't know the particulars of voting ages as far as elections go, but these numbers are wildly unrepresentative of the usual voting demographics. The best comparison of this governor's race is the 2016 Presidential primary in Michigan, so it's worth looking at those numbers for a comparison:

president age.JPG

Notice a difference? It's a little difficult to compare because the age groups don't match up, but it's clear that these age demographics have serious issues. Millennials were seriously underrepresented compared to how they actually vote (they almost voted at the same rate as the 30-44 and 45-64 designations), and El-Sayed's popularity with youngsters is duly noted.

As Whitmer runs as a centrist and the media continues to expose Thanedar, El-Sayed still has a good chance. And, his inspiring platform has the ability to turn out untold numbers of first time voters that could swing the election.

The latest Target-Insyght poll may very well be another large polling error that either costs El-Sayed the election or lays the groundwork for another historic upset.

The Secret History of the Catholic Church

The Secret History of the Catholic Church

Immigrants would rather face ICE than stay home

Immigrants would rather face ICE than stay home