Push Polling in the 2016 Election

I have been having some fun looking at polling samples in this election. This means looking into the raw data.
Here I have by far the most interesting one, as the sample is so far off base it is comical

Here is the poll, Arizona Republic/Morrison/Cronkite News poll showing hillary +6 in AZ.
Now lets look at the data




Now in AZ there are slightly more republicans registered than democrats, and the state leans red. As the data shows registered dems sampled 30 points higher than repubs. As for who they politically identify as it is 20 points. Now stuff like this is done in a lot of polls, most polls that have hillary up today are sampling dems at over +10

Now for the best part, a part not typically picked up on.

As we can see here they have sampled those with degrees to a truly ridiculous degree. There are 409 in the sample with a bachelors or higher. There are 103+201=304 with less than a bachelors. In AZ less than 1/3 have bachelors or above, the data has 57% with a bachelors. Hillary loses with those who have under a bachelors degree.

This is push polling or creating a narrative.
What do you think would happen if we normalized the data to Arizona's population?
+20 dem sample minimum, plus +27 bachelors and above sample.
It was difficult to find Arizona's raw data that would not be disputed so I did not attempt to adjust. I was going to simply adjust the education data to reflect reality.

Well 50/50 would give 50/50 results.
It's random dude. They didn't pick the dems, they picked a person who's answer was I'm a dem or republican.
The pollsters don't get access to voter registration records so they can't say oh hey she a democrat let's call her.
They call 100 people at random or however many samples they take.

a reply to: jellyrev


Just to add, and interesting video on a poll not being mentioned in the MSM.

originally posted by: Sillyolme
Well 50/50 would give 50/50 results.
It's random dude. They didn't pick the dems, they picked a person who's answer was I'm a dem or republican.
The pollsters don't get access to voter registration records so they can't say oh hey she a democrat let's call her.
They call 100 people at random or however many samples they take.

So if they did a poll and it showed Trump up by 15 points and it was found that 60 percent of poll participants regularly went to Drudge for news it would still be a scientifically valid result? When your polling data isn't close to the observed demographics then something is wonky about that poll and it's results should be dismissed or the margin of accuracy of the poll should be in the Dbl digits.

originally posted by: Sillyolme
They call 100 people at random or however many samples they take.

I hope it's a lot more than 100!

a reply to: Sillyolme

people with bachelors degree pick up the phone at 3 times the normal frequency of someone without a bachelors?

The more people who are sampled the closer to the mean a survey should become. That is the point of calling 100's of people.
What is the point of publishing random samples that are not representative? there is no point. Samples are supposed to be representative samples. If they are not representative of the population their is no point to doing the sample at all.
The people doing these surveys know full well this, as does anyone who is in the statistics, psychology or science fields.

They fail to adjust the data to the population if they were attempting to be honest. If I adjusted that education level to the population AZ would be hard red.

It's been said....

There are lies, damn lies and statistics.


I only trust bookies...


www.oddsshark.com...

a reply to: Sillyolme

How do you know that would give a 50/50 result... Maybe a registered dem might like Trump or maybe A Rep might like Hillary or another third party candidate.

Independents being polled more then Republicans is not right when the majority of voters are Dems and reps how can a poll be accurate when they survey more independents when like I said in the real world there's more democrats and Republicans then independents. Not to mention when you look at the polls where it's a more even sample size Trump is always winning.

Also I would like to add the methodology for the latimes poll is different then other polls and has been more accurate then majority of the other poll methods and Also the newest IBD/TIPP Tracking poll which has Trump up 2+ has been the most accurate poll in the most recent elections.

a reply to: jellyrev

I think polling and political spinsters should be banned. It sways voters opinions and their votes. Elections shouldn't have any outside influences swaying their opinions on who they vote for. They have their own eyes and ears, their decision and opinion should be their own.

originally posted by: olaru12
It's been said....

There are lies, damn lies and statistics.


I only trust bookies...


www.oddsshark.com...

Lol. Certainly, if you are indeed a gambler, you know heavy favorites go down every single day in every sport.

Vegas' existence depends on it!

originally posted by: six67seven

originally posted by: olaru12
It's been said....

There are lies, damn lies and statistics.


I only trust bookies...


www.oddsshark.com...

Lol. Certainly, if you are indeed a gambler, you know heavy favorites go down every single day in every sport.

Vegas' existence depends on it!

True but the house always wins....eventually!

I only bet on long shots when it's raining at the track.

a reply to: jellyrev

For a year now, we have been inundated by polls.

These polls have been used to manipulate how we vote. They have been fabricated, falsified.

But we're getting to a point now where the pollsters will have to salvage their reputation. So look for more honest polls in the coming days.

That way, after the election, they can say that they were correct and salvage some of their reputation.

a reply to: jellyrev

Great find, interesting analysis.

Are there any truly scientific polls, for AZ?

a reply to: pavil

Where in the world would results ever show that?
The only valid polls are random polls.

a reply to: jellyrev

Why are you asking me that?

a reply to: DBCowboy

There are trusted respected polls. There is valid information to be drawn from properly conducted polls.
The click polls on the internet are not valid or even trusted. Those are entertainment popularity contests open to the globe.

People who only trust the polls that support their viewpoint are just silly.
The campaigns certainly are paying attention to the polls.
They make decisions about where to campaign based on them.


And basically you're saying the polls are rigged donnie

a reply to: WeRpeons

Polls don't affect who I vote for. Or anyone else I doubt. Maybe some??
It just tells me where I stand against others.

originally posted by: Sillyolme
a reply to: DBCowboy

There are trusted respected polls. There is valid information to be drawn from properly conducted polls.
The click polls on the internet are not valid or even trusted. Those are entertainment popularity contests open to the globe.

People who only trust the polls that support their viewpoint are just silly.
The campaigns certainly are paying attention to the polls.
They make decisions about where to campaign based on them.
And basically you're saying the polls are rigged donnie

Polls should reflect the thoughts of the people at any given time.

instead, polls are now used to manipulate.

it's social engineering.

And in your case, it's worked quite well.

a reply to: Sillyolme

It affects a significant percentage of individuals. It's a preselection effect. This effect is used in a lot of human situations in order to evaluate and rank things/people they use what other people think especially those of higher expertise/status.

www.gsb.stanford.edu..." target="_blank" class="postlink" rel="nofollow">https...://www.gsb.stanford.edu/insights/how-polls-influence-behavior

The poll in question was not an online poll. It was conducted by a polling company.

originally posted by: Sillyolme
a reply to: pavil

Where in the world would results ever show that?
The only valid polls are random polls.

Exactly. Why give credence to a poll that doesn't actually reflect the demographics of The general population as a valid way of extrapolating accurate trends.?