The truth about polls and why we very rarely cover them

Political Polls

If I were to comb through the archives of the various websites I’ve published on over the years, I could probably find half-a-dozen articles about why polls are corrupt, manipulative, inaccurate, and really shouldn’t be considered newsworthy until literally the day before any given election. I’m going to pretend like I’ve never written about it and start fresh.

We won’t be covering them very often here. That puts us in a bit of a disadvantage with other news aggregation sites because they love polls. Heck, Drudge runs their own polls to add to the misinformation madness. We ran a poll once, but it wasn’t political. It asked our readers about whether they wanted to see more links, fewer links, or if we’d hit the Goldilocks moment on our homepage. Those types of polls are useful. Political voting preference polls are worthless four months before election day. In many ways they’re worthless the day before the election.

Here’s why:

All pollsters are manipulative

There is no such thing as an unbiased political poll. It’s impossible when there are humans involved. The pollsters who generate the questions often (though not always) position them in a way that favors one candidate or another. It’s about psychology. How are the questions worded? Where and how will the poll be delivered? What results will be accepted and what criteria will be used to throw out certain responses?

Then, there are the people delivering the polls. If you think that phone polls are safe, think again. The people making the calls or delivering the robo-message will tend to lean in one direction or another. The way they adjust their tone or inflections can affect the responses. It’s subtle but effective.

Online polls? Worthless. Where and how the questions are asked can make a huge difference. There’s a reason that Drudge’s post-debate polls showed that Donald Trump won all of them by a landslide even in the debates where he was completely destroyed by the moderators and his opponents. Again, the order and wording of the questions and response options makes a difference.

Many people being polled are manipulative

Many polls showed Trump was within striking distance or even beating Hillary Clinton in the May. Trump had vanquished his opponents while Bernie Sanders was still haunting Hillary. This made a huge impact on the polls because many of Bernie’s supporters claimed that they would vote for Trump in an effort to make Clinton look unelectable. There’s no way the majority of them would even consider voting for Trump, but they definitely didn’t want to help their biggest roadblock, Hillary, in her quest to lock up the nomination, so the manipulation happened.

Even today with Bernie gone, there are ways that those being polled manipulate the results. Some people will claim to be “likely voters” even if they probably won’t vote. Others will claim to be Independents to boost those numbers when they’re actually lifelong Democrats or Republicans.

The effects of sentiment within a particular polling segment can have enough of an impact on results to send the wrong message.

Tortured data

To paraphrase Ronald Coase, if you torture the data long enough, you can make it confess to anything.

Polls are often a case study in tortured data. Then, journalists and pundits take their run at the results to torture the data even more in a way that will fit their agenda. It’s human nature to want things to say what we want them to say. There was a recent poll that showed Clinton beating Trump by a certain percentage. The journalists reading the poll rightfully pointed out that the percentage of Democrats more than Republicans polled was equal to the gap between Trump and Clinton. Then, another publication pointed out that the gap was driven partially by a large percentage of Republicans who were voting for Clinton. I’m sure we could take it a few steps further and highlight other data points back and forth that yielded an advantage in both directions. The bottom line is this: polls are awful.

We won’t be posting very many stories about them on The New Americana. It will make our job harder; there are days when every other story is about a poll. So be it. They’re worthless and we won’t waste your time on them.

JD Rucker

JD Rucker is Editor of this site as well as Soshable, a Conservative Christian Blog. He is a Christian, a husband, a father, and founder of both Judeo Christian Church and Dealer Authority. He drinks a lot of coffee, usually in the form of a 5-shot espresso over ice. Find him on Twitter, Facebook, and Pinterest.

  1. Great article. It’s interesting that all news channels report polls without any context like: Who was polled by political affiliation, type of person polled (registered, likely, etc.), historical bias of the polling organization, etc.

    Although I understand why you do not want to amplify the noise generated by reporting poll results I would encourage you to continue reporting on your perspective and interpretation of various polls and the historical accuracy and partisanship of the major pollsters.

    Malcolm Baldrige said “Without data everything is just an opinion”. I would add to that “Without data honestly interpreted within the context of the bias of the pollster and the characteristics of the people polled – polls are just a tool for manipulating the uninformed”.

    A Washington Post or NYT’s poll is worthless given their historical bias, subjective reporting and blatant disregard for truth. Most other pollsters have the same track record. Enough said.

    1. Hi Dan,

      The closer we get to the election, the more the polls will mean to the end result. You pointed out the biggest problem – honest interpretation. I’ve found myself struggling in the past to not put a spin on polls and I am aware of the issues. Journalists who are either unaware of their own nagging bias or who simply don’t care will certainly spin. The funny part is that the margins in the polls are becoming so wide, I don’t think we can avoid reporting on them much longer.

Leave a Reply