Political Polls are useless in the age of Big Data

Political Polls are useless in the age of Big Data.

Last year, most of the polls said Trump would lose but every A.I. prediction said Trump would win. In this age of big data, it's hard to see how polls of 500 or 1,000 people can capture the true mood of the people like with A.I. and Big Data.

Using an artificial intelligence algorithm that mined social media, MogIA predicted Trump would win when almost no one else did.

In an era where self-driving cars promise to soon swarm down roads and Black Mirror is the hottest show on Netflix, it is only appropriate that an algorithm created by a Mumbai-based company MogIA called the election correctly on October 28.

Oh, and MogIA didn't just call this election correctly. It did so with both the Democratic and Republican primaries this year and with three of the last US Presidential elections (although there's been no public verification of those results).

As MogIA's creator Sanjiv Rai explained it to CNBC when he sent them his results in late October, MogIA (named after Rudyard Kipling's Mowgli) plumbs the vast and endless terrain offered by social media to make sense of national sentiments.

Specifically, it scans 20 million data points from public platforms such as Twitter, Google, and Facebook to come up with its predictions. Rai told CNBC that his AI system clearly showed that in any of the elections that MogIA had analysed, the winning candidate was the one that had leading engagement data.

"If Trump loses, it will defy the data trend for the first time in the last 12 years since internet engagement began in full earnest," Rai wrote in a report sent to CNBC on October 28. Not that anyone was listening. Rai also pointed out that Trump had managed to overtake even Barrack Obama's engagement number in 2008 by 25 percent.

www.zdnet.com...

It's just like the Polls about impeachment. The Data supports Trump. Look at his massive rally's and fundraising. The Polls overly sample Democrats and they don't ask any questions that might get an adverse response like:

Do you think Adam Schiff should lead impeachment trials?

Do you think Adam Schiff is compromised?

Do you think the President deserves due process?

Do you think the Republicans deserve the same subpoena rights as the minority parties during Nixon and Clinton?

You can illicit any information from polls. Networks should have a Data Scientist that follows Data trends also. Here's more:

From January 1 to November 6, 2016, Donald Trump and Hillary Clinton amassed a total following of 48,986,921 across Facebook, Twitter, and Instagram. On those three networks, Trump and Clinton’s posts attracted 495,120,770 engagements. Half a billion likes, comments, shares, retweets, and reactions. And most of them went to Trump.

On Facebook, Donald Trump collected a 208,099,876 Facebook engagements and 12 million Facebook fans. Hillary Clinton, who spent early 2016 fending off Bernie Sanders while Trump consolidated support, attracted 72 million engagements and 7.9 million Facebook fans. Her slick mostly image-based campaign won urban voters, but Trump’s grassroots, raw content crowdsourced from rallies was a bigger hit with his fans.

On Twitter, Trump stayed ahead with 89,459,006 total engagements to Hillary’s 41,572,396.

Which candidate won the election on social media? Trump and Clinton’s total followings are 24.3% apart. As of November 6, Donald Trump has 27,902,237 fans across Facebook, Twitter, and Instagram to Hillary Clinton’s 21,084,684.

If the general election follows the same pattern as the primaries, the social media data collected indicates that Donald Trump will be the next President.

medium.com...

These numbers are more important than any Poll.

Yep. politics polls are useless.

Next threAD?

a reply to: neoholographic

Polling, as it relates to statistical mathematic probability, is as accurate as it ever was, given the margin of error "fine print". Math doesn't lie.

I agree with your basic statement though, but it isn't because of the superiority of algorithms.

The biggest problem with polling lies with the fact that they are conducted by humans, who are, by nature, bias creatures.

Questions are often asked in a "leading" manner. Over/under sampling are commonplace, and ofter done purposely.

I've taken part in polls that were cut short when my first couple answers were not what they were looking for. If pollsters do that type of thing, the whole poll becomes invalid.

Now, I either just decline to take part or, if I'm in the mood for mischief, lie to "troll" them.

Devil's Advocate.

Polls, especially political ones, are extremely useful - normal people just don't know what their real purpose is.

The purpose of the poll is not to accurately predict, but to influence behavior by spreading the idea of the certainty of an 'expected' outcome.

This is why different polls with sponsors of different candidates (or ideas) have very different results, known as "cherry picking". NOT cherry picking the results - but picking the universe of respondents, targeting specific groups for specific outcomes.

As long as people seek and adopt the opinions of others instead of their own, pollsters will have an honored place in bringing these PSYOPS to our attention. Polls are basically PSYOPS.

OPs point about AIs are well taken. Political bias and predetermined polling results can be programmed into the AI, but then it would be just as inaccurate as humans. In my decades with big data, I found that regardless of the endeavor, bias is strong with assumptions underlying predictive models. Most mathematical models, not just polls, reek of bias and assumptive manipulation in academia and throughoutbusiness.

SO my position is that Polls are not useless, because they achieve the results their patron$ are looking for.
That's usually not in line with what the normal people reading the polls are looking for and don't expect.

ganjoa

a reply to: ganjoa

Excellent points as usual, ganjoa.

originally posted by: Mach2
a reply to: neoholographic

Polling, as it relates to statistical mathematic probability, is as accurate as it ever was, given the margin of error "fine print". Math doesn't lie.

I agree with your basic statement though, but it isn't because of the superiority of algorithms.

The biggest problem with polling lies with the fact that they are conducted by humans, who are, by nature, bias creatures.

Questions are often asked in a "leading" manner. Over/under sampling are commonplace, and ofter done purposely.

I've taken part in polls that were cut short when my first couple answers were not what they were looking for. If pollsters do that type of thing, the whole poll becomes invalid.

Now, I either just decline to take part or, if I'm in the mood for mischief, lie to "troll" them.

Good points!

I got called by a poll and I'm an Independent. When they found out I was a Conservative leaning Independent, they ended the call.

So in these surveys they say their polling independents but there's moderate independents, left leaning independents and right leaning independents. So you can skew a poll one way or the other just by saying you're polling independents but you don't break down which way independents lean.

You can poll 80 percent left leaning independents and 20% conservative leaning independents and skew the poll.

Last year, most of the polls said Trump would lose but every A.I. prediction said Trump would win. In this age of big data, it's hard to see how polls of 500 or 1,000 people can capture the true mood of the people like with A.I. and Big Data.

a reply to: neoholographic

Well, polling a 1000 respondents is, incredibly, accurate to the point of which polling 10 million people isn't noteworthily more accurate.

Really, it is the the wording, sampling and interpretation that matters.

You can have three diverging but equally accurate polls - they could even have the exact same sampling pool and still be vastly different even though all the respondents answered honestly in all three polls.

Odds are, that they are measuring slightly different things, and odds are that the polling firms have slightly different models for weighing the answers.

The likely would be no different using big data - be it conducted by humans or AI. Three different AI's would have three different predictions, because ultimately it IS a prediction, and ultimately the AI must in some way or another decide the likelihood of whether or not the respondent is actually going to vote come election day.


Also, lets not pretend deliberate bias isn't a thing - polls are a huge part of campaigning and if you as a polster has Part A as your largest recurring customer, you make rally really sure that your poll are not biased against said party.

No one will admit that of course.

a reply to: DupontDeux

Exactly true.

That is why all big, well run campaigns do their own internal polling.

Add the polling about Brexit. The polls predicted wrong.
Also in the Netherlands polls were wrong about a Ukrain-treaty referendum.

I do see a common theme: the polls seem to reflect the opinion of the globalists of the deep states and NOT that of the common people.
Do they have wrong samples? Do voters give a honest answer? Do you admit you have a non-PC opinion to a stranger on the phone?

Conclusion: I also take the polls with a pinch of salt...

a reply to: neoholographic

Polls are used now to influence opinion and manipulate elections by skewing perceptions and reinforcing biases.

Political polls are useless because this country has never been a direct democracy.

There was obviously a lot of people that voted for him that didn't want anyone to know they were voting for him. They were also not really voting for him. They were voting with their wallet. He basically flipped a lot of Democrats and none of the conventional polling was relaying that data. That is what happened.

Polling Trump is basically a greasy stick of dynamite. The pollsters could take that into account this time that not much has changed since the last election. He's not losing those voters unless the economy tanks because that is why they voted for him in the first place; more money in their pocket. Do they have more disposable income? Of course they do. If anything, the polls will be more screwed up if they use the same methodology because he will get more votes this time.

I am hoping pollsters are part of Lizs' two million job cuts. Them and insurance people.

originally posted by: neo96
Political polls are useless because this country has never been a direct democracy.

The electoral college is not so complicated that polling can't be weighed accordingly. How hard would it be to factor in state by state results?

That isn't the issue.

originally posted by: ganjoa
Devil's Advocate.

Polls, especially political ones, are extremely useful - normal people just don't know what their real purpose is.

The purpose of the poll is not to accurately predict, but to influence behavior by spreading the idea of the certainty of an 'expected' outcome.

This is why different polls with sponsors of different candidates (or ideas) have very different results, known as "cherry picking". NOT cherry picking the results - but picking the universe of respondents, targeting specific groups for specific outcomes.

As long as people seek and adopt the opinions of others instead of their own, pollsters will have an honored place in bringing these PSYOPS to our attention. Polls are basically PSYOPS.

OPs point about AIs are well taken. Political bias and predetermined polling results can be programmed into the AI, but then it would be just as inaccurate as humans. In my decades with big data, I found that regardless of the endeavor, bias is strong with assumptions underlying predictive models. Most mathematical models, not just polls, reek of bias and assumptive manipulation in academia and throughoutbusiness.

SO my position is that Polls are not useless, because they achieve the results their patron$ are looking for.
That's usually not in line with what the normal people reading the polls are looking for and don't expect.

ganjoa

Polling is a business and just like any good business, they cater to their customers so I agree on that take. NPR did a good segment on it a while back. Some of the pollsters were saying they would totally throw out data if it wasn't helping their client.

The best polls though are the internal party polling which we will never see. They know what is going on. The stuff we see is what the use car salesman wants you to see. He's not going to tell you he poured two bottles of radiator stop leak in it and emptied two cans of fix a flat in the back left tire.

a reply to: neoholographic

Political Polls are useless in the age of Big Data

Really...?

Then I wonder why trump is obsessed with his poll numbers. And why he went off on Fox for broadcasting his unfavorable poll numbers. The WH had it's own pollsters but trump fired them for leaking unfavorable numbers.

www.cnbc.com...

www.politico.com...

If trump can't even persuade Fox to skew the numbers....what's that tell you?

originally posted by: olaru12
a reply to: neoholographic

Political Polls are useless in the age of Big Data

Really...?

Then I wonder why trump is obsessed with his poll numbers. And why he went off on Fox for broadcasting his unfavorable poll numbers. The WH had it's own pollsters but trump fired them for leaking unfavorable numbers.

www.cnbc.com...

www.politico.com...

If trump can't even persuade Fox to skew the numbers....what's that tell you?

It is to Trump's advantage to discredit ALL polls. It plays well to his base. He does it because it is the smart play.

In case you aren't aware, Fox, like everyone else, outsources it's polling.

Forrest for the trees detective. Forrest for the trees........

originally posted by: Mach2

originally posted by: olaru12
a reply to: neoholographic

Political Polls are useless in the age of Big Data

Really...?

Then I wonder why trump is obsessed with his poll numbers. And why he went off on Fox for broadcasting his unfavorable poll numbers. The WH had it's own pollsters but trump fired them for leaking unfavorable numbers.

www.cnbc.com...

www.politico.com...

If trump can't even persuade Fox to skew the numbers....what's that tell you?

It is to Trump's advantage to discredit ALL polls. It plays well to his base. He does it because it is the smart play.

In case you aren't aware, Fox, like everyone else, outsources it's polling.

Forrest for the trees detective. Forrest for the trees........

I'm sorry, I forgot trump is a "stable genius" and the "chosen one" my mistake....