What kind of a poll did NBC/WS Journal do?

[louis c]

A poll that had a 46% approval for Trump seems absurd to me. Then, I looked at the numbers. I didn't have to look too far for the answer Here's the demographic breakdown; White 76%: Black 12%: Hispanic 7%: Asian 2%: Other 3%. Seems a little skewed to me.

Link to Poll:
http://media1.s-nbcnews.com/i/today/z_creative/18852NBCWSJAugustN=600FlashPoll.pdf

[tonyt53]

1. Yep. Polls always have other numbers that people ignore, as in "who was polled".

[unblock]

2. In theory you can adjust for that

The Hispanic population should be more like 17%, but you can weight the results to compensate. Don't know if they did that.

[DemocratSinceBirth]

8. I sent the polling organization an e-mail.

It's - non-standard way of polling. They don't ask respondents if they're Hispanics. They have to volunteer the fact and the pollster goes from there.

The question should be are you

-white
-African American
-Hispanic
-Other

[ProfessorGAC]

3. 42% Of Those Polled Were 55 and Over

50% were aged 50 or over.

In 2017, the median age was 38.1 years. So, this poll is not reflective of the age demographic in the country. 50% of the respondents were 12 years or more above the median age of the US.

The age detail shows 28.54% of the population being aged 55 or higher.

So, this poll over-represents that segment by 75.2%.

That's egregious.

[NewJeffCT]

5. And, isn't the Hispanic population

somewhere around the same level percentage wise as African Americans - 12-13%? The poll only had 7%

[NewJeffCT]

6. replying to myself

Asian population is also closer to 5% as well, but only 2% in the poll.

[ProfessorGAC]

10. The Numbers I Was Using Are A Bit Unclear

It shows 17.1% in 2017, but that is "Hispanic of Any Race". So, there are black latinos counted in that number.

The numbers suggest that you might be onto something there., though.

[lagomorph777]

7. WSJ is a Rupert Murderer property.

So of course it's skewed.

[ProfessorGAC]

11. But, NBC Isn't

So, i'm disappointed that they would have used such an obviously flawed set of stats.

[lagomorph777]

12. +1

[2naSalit]

4. that and how the questions are formulated/asked...nt

[lapucelle]

9. They polled a 100 member non-representative sample.

76 of the respondents were white.

[fallout87]

13. 538 rates this particular poll at an A-

https://projects.fivethirtyeight.com/polls/

[Cicada]

14. 538 rates Wall Street journal nbc poll a-

Poll averages are more accurate than single polls. By chance alone single polls will bounce around several percentage points.

[jpljr77]

15. They're trying to mirror the electorate rather than the country's demographics.

I think it's a bit misleading, since the headlines make it out to be overall approval. It really should be "Presidential Approval Among Voters."

In 2016, the demographic breakdown of the electorate was:

70% white
12% black
11% hispanic
4% Asian
3% other

[louis c]

17. That's still not representative of the poll

and you can expect the 2018 electorate to be more minority, female, younger and less white.

[pecosbob]

16. More corporate media normalization of criminal Republican behavior

Whether you can see it or not (if you're blind for example), they're all carrying water for the Rs.