New CNN poll faulty as usual

originally posted by: dfnj2015
a reply to: face23785

They are bogus because you don't like them. Got it.

The Electoral College Poll a year ago was the most accurate.

a reply to: dfnj2015

So the point of it all is just scoring the cheap points now? Good to know.

a reply to: burdman30ott6

Yes it does. It shows that Trump is right. Mental illness is a major issue. Sadly, they are all in his base!

originally posted by: dfnj2015

originally posted by: face23785

originally posted by: dfnj2015
a reply to: face23785

They are bogus because you don't like them. Got it.

The biggest number in my OP was 1,021. The fact you can't understand it is very, very sad.

There is a science to polling. You don't have to poll every single person in the country. You just have to poll a small number and interpolate the results with a factor of error. Of course, how you frame a polling question is very important. Many people engineer polling questions to be elicit desired results. The number of people polled is not as big a deal if it is done right.

Nowhere did I state they had to poll every single person in the country. My entire issue was with the breakdown of the sample. Did you even read the OP? It would really help your understanding. That's 2 posts full of a complete lack of understanding of what I said. You can't possibly be this dumb, I have to assume you just didn't read it. Everyone else responding understands what I said, even if they don't agree. You just have no idea what I said. Please read the OP.

originally posted by: tadaman
a reply to: face23785

Land line polls are absolutely absurd.

Oh, that was my other favorite part I forgot to highlight in the OP:

The study was conducted for CNN via telephone by SSRS, an independent research company. Interviews were conducted from Nov ember 2, 2017 – November 5 2017 among a sample of 1,021 respondents. The landline total respondents were 411 and there were 610 of cell phone respondents.

40% of the sample was landline. I don't think I even need to comment further.

I don't think that there should be cherry picking vis-à-vis the people interviewed. In that, there seems to be a lot of people who are not so happy with either of the parties anymore. Not withstanding that, coming to an election could be a total different kettle of fish in what way they would vote if there is still only the two parties to vote for.
A new kid on the block party though could see those disaffected voting for them.

a reply to: face23785

If it was an election poll, then I think you might have a point.

But it's an approval-rating poll. So, it makes sense to just randomly ask a certain number of Americans if they approve of the job Trump's doing or not -- regardless of whether they vote, or not...or are registered, or not, etc.

originally posted by: IAMTAT
a reply to: face23785

A new poll from the same Fake News Network that, a day earlier, went out of it's way to accuse Trump of overfeeding fish in Japan by editing film?

CNN didn't edit. Their video is on the CNN website in full, not something posted from a cell phone, but some people don't bother about the 'little details'...when it suits.

The Electoral College Poll a year ago was the most accurate.

Well, it's only accurate if you're polling the electors. If you're polling the voters, it's not. *ZING* lol

originally posted by: MotherMayEye
a reply to: face23785

If it was an election poll, then I think you might have a point.

But it's an approval-rating poll. So, it makes sense to just randomly ask a certain number of Americans if they approve of the job Trump's doing or not -- regardless of whether they vote, or not...or are registered, or not, etc.

I agree. That's not what they did though. In my OP, I stated that I think for opinion polls they should survey a sample of the entire adult population, not just registered voters. They didn't though, they specifically went for only registered voters. If you're gonna do that, you should use the actual voter makeup to get an accurate sample. They didn't do that. Also, I did state that they even do this in election polls, which is a big problem.

Did anyone read the entire OP before responding? Is it really that long or complicated?

This is not the Mud Pit, people.

Please remember that, and post accordingly.

Don't respond to this post.

originally posted by: face23785

They didn't though, they specifically went for only registered voters....

...Did anyone read the entire OP before responding? Is it really that long or complicated?

YOU may have written that, but the poll doesn't say that and neither does the portion you quoted. So I ignored it assuming you are incorrect:

Among the entire sample, 32% described themselves as Democrats, 24% described themselves as Republicans, and 44% described themselves as independents or members of another party.

"Describing yourself" as being partisan or an independent doesn't mean you are a registered voter.

If you have a quote saying these were all registered voters -- or that the pollster "went for" registered voters -- please do share it.

Looking at the amalgamation sites, RCP has him at 38.8. 538 has him at 37.6.

Even Rasmussen, who have been consistently pro-Trump, have him in the low 40s.

You can keep ignoring the facts all you want. The simple truth is that he's a dud. He got elected not for his policies but because the Dems nominated an even more unlikable candidate.

a reply to: face23785

What percentage of these are Democrats/Republican vs independents leaning Democrat/Republican?

Overall, 48% of all registered voters identify as Democrats or lean Democratic compared with 44% who identify as Republican or lean toward the GOP.

It very well maybe that if you take away the option of "lean" you'll find that the numbers from Pew Research might line up more closely.

However, they chose to only poll registered voters.

That statement is inaccurate. From the SRSS document:

A total of 1,021 adults were interviewed by telephone nationwide by live interviewers calling both landline and cell phones. Interviews were conducted in English and Spanish. Among the entire sample, 32% described themselves
as Democrats, 24% described themselves as Republicans, and 44 % described themselves as independents or members of another party.

All respondents were asked questions concerning basic demographics, and the entire sample was weighted to
reflect national Census figures for gender, race, age, education, region of country, and telephone usage.

Results for the full sample have a margin of sampling error of +/-3.6 percentage points. For the sample of 910 registered voters, it is +/-3.8 percentage points.

910 out of 1,021 identified themselves as registered voters. (I'm assuming they didn't check voter registration, confirm identities of participants and crosscheck)

What they did do though was use the demographic data to weight the responses vs the US Census. It's not clear to me from skimming the document how that weighting was applied. That could be problematic depending on the raw numbers.

An extreme example to illustrate would be say having 1 white respondent and 99 black respondents and then going by demographic data from the census, count that 1 response from a white respondent as 63% and the 99 from black respondents as 19%.

No polling is perfect but it tends to get better with larger data sets. Maybe they should question 10,000 and then randomly choose from among those X amount of responses to perfectly reflect the adult population. Then again, different demographics vote at different rates from election to election. So then maybe you try to weight the results by the voting breakdown from the last election or two.

I don't know that SRSS/CNN did anything inappropriate here. I usually just follow Rasmussen as it's been shown to be the most right-leaning over and over again. That way I'm giving Trump the biggest benefit of the doubt possible.

The most recent Rasmussen presidential approval poll has Trump at 43% approve with a -21 spread on strongly approve/disapprove (27% strongly approve, 48% strongly disapprove).

Do you think those are good numbers?

By comparison, Rasmussen had Obama at this on Nov 7th, 2009:

Approval index: -7 (vs Trump's -21)
Strongly Approve: 31% (vs Trump's 27%)
Strongly Disapprove: 38% (v Trump's 48%)
Total Approve: 49% (vs Trump's 44%)
Total Disapprove: 50% (vs Trump's 56%)

Anyway, polls have shortcomings but they're always more accurate than Donald Trump's personal opinion which seems good enough for the average Trump supporter.

originally posted by: Eshel

The Electoral College Poll a year ago was the most accurate.

Well, it's only accurate if you're polling the electors. If you're polling the voters, it's not. *ZING* lol

Always a good idea to follow the U.S. Constitution.

originally posted by: MotherMayEye

originally posted by: face23785

They didn't though, they specifically went for only registered voters....

...Did anyone read the entire OP before responding? Is it really that long or complicated?

YOU may have written that, but the poll doesn't say that and neither does the portion you quoted. So I ignored it assuming you are incorrect:

Among the entire sample, 32% described themselves as Democrats, 24% described themselves as Republicans, and 44% described themselves as independents or members of another party.

"Describing yourself" as being partisan or an independent doesn't mean you are a registered voter.

If you have a quote saying these were all registered voters -- or that the pollster "went for" registered voters -- please do share it.

Edit: Addressed better in the post below

originally posted by: theantediluvian
a reply to: face23785

What percentage of these are Democrats/Republican vs independents leaning Democrat/Republican?

Overall, 48% of all registered voters identify as Democrats or lean Democratic compared with 44% who identify as Republican or lean toward the GOP.

It very well maybe that if you take away the option of "lean" you'll find that the numbers from Pew Research might line up more closely.

However, they chose to only poll registered voters.

That statement is inaccurate. From the SRSS document:

A total of 1,021 adults were interviewed by telephone nationwide by live interviewers calling both landline and cell phones. Interviews were conducted in English and Spanish. Among the entire sample, 32% described themselves
as Democrats, 24% described themselves as Republicans, and 44 % described themselves as independents or members of another party.

All respondents were asked questions concerning basic demographics, and the entire sample was weighted to
reflect national Census figures for gender, race, age, education, region of country, and telephone usage.

Results for the full sample have a margin of sampling error of +/-3.6 percentage points. For the sample of 910 registered voters, it is +/-3.8 percentage points.

910 out of 1,021 identified themselves as registered voters. (I'm assuming they didn't check voter registration, confirm identities of participants and crosscheck)

What they did do though was use the demographic data to weight the responses vs the US Census. It's not clear to me from skimming the document how that weighting was applied. That could be problematic depending on the raw numbers.

An extreme example to illustrate would be say having 1 white respondent and 99 black respondents and then going by demographic data from the census, count that 1 response from a white respondent as 63% and the 99 from black respondents as 19%.

No polling is perfect but it tends to get better with larger data sets. Maybe they should question 10,000 and then randomly choose from among those X amount of responses to perfectly reflect the adult population. Then again, different demographics vote at different rates from election to election. So then maybe you try to weight the results by the voting breakdown from the last election or two.

I don't know that SRSS/CNN did anything inappropriate here. I usually just follow Rasmussen as it's been shown to be the most right-leaning over and over again. That way I'm giving Trump the biggest benefit of the doubt possible.

The most recent Rasmussen presidential approval poll has Trump at 43% approve with a -21 spread on strongly approve/disapprove (27% strongly approve, 48% strongly disapprove).

Do you think those are good numbers?

By comparison, Rasmussen had Obama at this on Nov 7th, 2009:

Approval index: -7 (vs Trump's -21)
Strongly Approve: 31% (vs Trump's 27%)
Strongly Disapprove: 38% (v Trump's 48%)
Total Approve: 49% (vs Trump's 44%)
Total Disapprove: 50% (vs Trump's 56%)

Anyway, polls have shortcomings but they're always more accurate than Donald Trump's personal opinion which seems good enough for the average Trump supporter.

You've got me on the 910 vs 1,021 people. My mistake, good catch. Breaking those numbers out only slightly improves the sample, they still grossly misrepresented the population though:

218 Republicans when it should've been 400
291 Democrats when it should've been 436
400 Independent/Other when it should've been 73

Thanks for your help. The poll remains poorly done, even with that adjustment.

originally posted by: face23785

originally posted by: MotherMayEye

originally posted by: face23785

They didn't though, they specifically went for only registered voters....

...Did anyone read the entire OP before responding? Is it really that long or complicated?

YOU may have written that, but the poll doesn't say that and neither does the portion you quoted. So I ignored it assuming you are incorrect:

Among the entire sample, 32% described themselves as Democrats, 24% described themselves as Republicans, and 44% described themselves as independents or members of another party.

"Describing yourself" as being partisan or an independent doesn't mean you are a registered voter.

If you have a quote saying these were all registered voters -- or that the pollster "went for" registered voters -- please do share it.

Those are cute semantics, but if they weren't interested in your registration they wouldn't even have asked that question. They may have asked whether you're liberal or conservative, but Republican and Democrat are party affiliations. They're not political ideologies. This is a professional survey, it wasn't some college kids who don't know what these terms mean.

Maybe the pollster thought the public might be interested in having that info when contemplating the results. Personally, I think it's crucial information to understanding them.

I'm just not seeing anything all that faulty about this. Among that sampling of people who described their political leanings as (fill in the blanks), Trump's approval rating was 36%. And apparently about 90% of them said they were registered voters.

originally posted by: Xcalibur254
Looking at the amalgamation sites, RCP has him at 38.8. 538 has him at 37.6.

Even Rasmussen, who have been consistently pro-Trump, have him in the low 40s.

You can keep ignoring the facts all you want. The simple truth is that he's a dud. He got elected not for his policies but because the Dems nominated an even more unlikable candidate.

This is a cute deflection. Nowhere in the OP did I argue Trump was wildly popular. I merely presented how the polling methodology is flawed and therefore the results are likely inaccurate. Who knows what the real numbers are. Maybe they should do a real poll.

It's been reported that Trump is obsessed with his poll numbers. I'll bet he is crying in his sushi about now.