How does political polling work?

Next week's Republican presidential debate requires candidates to hit at least 3% in two national polls -- or 3% in one national poll and two early state polls. But have you ever wondered how those polls are conducted? CBS News executive director of elections and surveys Anthony Salvanto explains.

#news #politics #polling

[StephanieStults]

I have never seen people being polled and I have never been polled, nor do I know anyone that has been polled EVER

[brothermine]

Although pollsters do a reasonably good job of sampling the population, they do a terrible job of choosing the questions to ask.
For example, "approval" polls lump together people who give the same answer despite having opposite preferences. Someone who says he "disapproves" Biden's performance could mean he prefers Trump (or other Republicans) over Biden, but it could mean he prefers Bernie Sanders over Biden (and would vote for Biden over Trump and other Republicans). It could mean he's disappointed at how slow the DoJ investigation of Trump's Jan6 role has been. Approval/disapproval is a false dichotomy, meaningless and misleading because it's really just a linguistic shorthand that refers to a comparison to an unstated alternative that's different for each person responding.
Two other examples of false dichotomy polls are "support or oppose" and "right track or wrong track." Suppose 35% of the people prefer a track further to the left and 40% prefer a track further to the right. Then 75% would say "wrong track" and the poll would make the current track appear very unpopular. But in fact the current track is very popular: a 60% majority (35%+25%) prefer it over a track further to the right and a 65% majority (40%+25%) prefer it over a track further to the left.
The best polls elicit head-to-head preferences, which have a clear meaning. All of each person's head-to-head preferences can be obtained by letting each person express his/her order of preference. That would allow the counting of all the head-to-head majorities, which is what we should be paying attention to. In the example above, the 60% and 65% majorities are head-to-head majorities, but the 75% majority is nonsensical because it lumps together people who have opposite preferences.

[turtlefeet]

terrible description, the spaghetti looks like vomit too.

[leighagnello]

To me the system can easily be bias and maniplate polls in areas they want to demotivate or motivate voters.

[reverbscherzo]

I just came here to see if polls are still phone-based, and what alternative to phone polls there are out there. Sounds like we still aren't actually getting a representative sample, since it sounds like we're still using mostly phones, which they even said people don't answer anymore. Who does? Old people and very old people. I'm an old people, but I got rid of my landline ages ago, and haven't participated in a single poll since. Maybe we should do a poll to see how many left-leaning old people still have landlines as opposed to right-leaning old people.

[JT.common_sense]

Are most polls done in NYC, Chicago and San Francisco?

[whiskeybrown]

Poor explanation, he didn't say how accurately the polls were compared to the real election.

[kennethmorrison]

I'm 77 & have never been polled & niether has anyone I've encountered in my lifetime. They poll the bourgeoisie in the the suburbs & that's why the right wing coservatives maintain their strangle hold on society

[johnmesser]

Let us call 1000 people and that is a poll of the whole 300 million Americans.
Fn ridiculous

[LLLL]

One thing I notice is that Trump is leading not Biden. We know that CBS is woke!!

[graciegrae]

why do the numbers change every day? are they taking new samples/surveys every day, and then reporting/compiling these lists to...the internet?? who has actually been polled?

[reading]

Thank you 🙏 I was wondering how they worked!

[grafixnetz]

My problem with polls is the number of participants. Compare results from a poll of 25, 000 to 10, 000 to 1000 and chose a minimum that best matches reality.
A national poll should have at least 50, 000 IMHO.
And *ALWAYS* show the number of people polled please!

[UserName-sjfg]

So you don;t phone people. What do you do? What are some sample questions? How do you contact people?

[jacobmacaree]

Again. Its exactly like the phone system. Nobody wants to fill out your surveys or questionnaires.

[pedropenalver]

Stupid system. The real definition of fallacy.

[meta]

Grandma's spaghetti looks like the inside of the average democrats' head.

[zxeonub]

Of course you’ll give creepy job I higher score 🙄🖕

[TheMarquescalwile]

Only reason for this segment is because Biden is losing and the sheep need an elementary lesson.

[bradleybruvva]

People should be able to opt in to polling just like voting. We get no real say in our candidates until it comes down to the last 2 candidates. Then we have to choose from two crappy candidates.