FOX Ohio poll is also a pice of horse manure: +- 5,5% - 9 % error margin among subgroups

[factfinder_77]

The Fox Ohio poll is pure crap.

the error margin: (registed, likely voters) among:

independents: +- 8 % and 9 %
voters under 45 : +- 6 % and 7 %
dem/rep : +- 5%
college degree: +- 5 %
income under 50K: +- 5,5 and 6 %

[Dawson Leery]

1. Trump TV had to do something to damper HRC's good days.

Any poll with an MOE over 4 points is weak, over 5 should be dismissed, over 6 be banned from consideration.

[LenaBaby61]

9. Any poll with an MOE over 4 points is weak, over 5 should be dismissed, over 6...

IS a joke.

My political science teacher told/taught me the very same thing you're talking about.

ANY poll with a MOE over 3.5, I generally don't pay attention to.

5.5 though?

[vadermike]

2. And

How the hell could she be leading by 5 nationally on the NBC 4 way but trailing in these three swing states Makes no sense I hope these fox polls are wonky

[factfinder_77]

4. error of plus-minus 3.1 percentage points in the nbc/wsj poll

The NBC/WSJ poll was conducted Sept. 16-19 of 1,000 registered voters - by both landline and cell phone interviews - and it has an overall margin of error of plus-minus 3.1 percentage points. Among the 922 likely voters the survey interviewed, the margin of error is plus-minus 3.2 percentage points.

[democrattotheend]

8. What is the MOE for the subgroups in the NBC poll?

I would be they are higher than the MOE for the poll overall. That was certainly the case with the Monmouth poll that showed Hillary up by 9 in New Hampshire.

Also, it's an apples to oranges comparison because national polls usually have a larger sample and thus a smaller MOE. The larger the sample size, the smaller the MOE.

[democrattotheend]

3. Subgroup MOE's are always high

In any poll, because the number is small. If 800 people are polled for a state poll, each subgroup will only have a few hundred people and thus will have a higher margin of error.

I don't see any glaring deficiencies in this poll. They did about an even number of landline and cell phones, and they had bilingual callers for Nevada.

[factfinder_77]

5. your ignorance and logic are duly noted

lol

[democrattotheend]

6. I used to work with professional pollsters

I am sorry that you think I am ignorant because I don't bury my head in the sand whenever I don't like a poll.

By comparison, here are the subgroup margins of error for that great Monmouth New Hampshire poll that came out today:

MARGIN OF ERROR
unweighted
sample
moe
(+/-)
LIKELY VOTER Total 400 4.9%
SELF-REPORTED
PARTY ID
Republican 98 9.9%
Independent 201 6.9%
Democrat 98 9.9%
IDEOLOGY
Conservative 114 9.2%
Moderate 178 7.4%
Liberal 99 9.9%
GENDER
Male 193 7.1%
Female 207 6.8%
AGE 18-49 181 7.3%
50+ 218 6.6%
COLLEGE DEGREE
No 182 7.3%
Yes 212 6.7%
INCOME
<$50K 112 9.3%
$50-100K 108 9.4%
$100K+ 134 8.5%

[factfinder_77]

10. The Square Root Law and Subgroups

http://www.stat.umn.edu/geyer/old01/3011/examp/poll.html

http://www.pewresearch.org/fact-tank/2016/09/08/understanding-the-margin-of-error-in-election-polls/

Generally, the reported margin of error for a poll applies to estimates that use the whole sample (e.g., all adults, all registered voters or all likely voters who were surveyed). But polls often report on subgroups, such as young people, white men or Hispanics. Because survey estimates on subgroups of the population have fewer cases, their margins of error are larger – in some cases much larger.

A simple random sample of 1,067 cases has a margin of error of plus or minus 3 percentage points for estimates of overall support for individual candidates. For a subgroup such as Hispanics, who make up about 15% of the U.S. adult population, the sample size would be about 160 cases if represented proportionately. This would mean a margin of error of plus or minus 8 percentage points for individual candidates and a margin of error of plus or minus 16 percentage points for the difference between two candidates. In practice, some demographic subgroups such as minorities and young people are less likely to respond to surveys and need to be “weighted up,” meaning that estimates for these groups often rely on even smaller sample sizes. Some polling organizations, including Pew Research Center, report margins of error for subgroups or make them available upon request.

[triron]

14. Demographics

by age aren't available from what I gathered looking at internals. Do you continue to feel polls thes Fox polls are junk?

[factfinder_77]

15. Fox deliberately composed a poll sample that disfranchise Clintons electorate and fit a narrative

Hence giving T a spike in the polls and the “race” seems to be close or T leading.
CNN did the same thing in Ohio and Florida.


why do you think they didn’t include a bigger sample among young voters, or any other paticular voting group ?

[triron]

16. Yes

CNN did that even more so I judge. The Fox polls seem somewhat more obscure though. Seems to be an incongruity with the national poll out today and even the other North Carolina poll. Nevada was a little worse than the others out today (T +2).

[karynnj]

19. That explanation is completely on target

The formula for the error is a function of the variance and sample size. As said, the sub groups have much smaller sample sizes than the full sample.

Many kids learn this in high school.

[geek tragedy]

7. Ohio is probably closer than that but consensus seems to be that Trump is ahead in Ohio.

The Clinton campaign is privately expressing pessimism that they can win the state.

[factfinder_77]

11. It depends on the young turnout

In 2012, the Ohio voting rate among 18-29 was 50,2 ( in thousands) 66,6 among 30 and older.

Hillary needs a high turnout to cancel the white vote that go nearly + 20 in favor T

https://www.census.gov/prod/2014pubs/p20-573.pdf

[BzaDem]

12. All polls have huge margins of error among subgroups.

That is just how statistics work. The larger poll has a lower margin of error than subgroups, because subgroups have fewer people in them (by definition of subgroup).

[factfinder_77]

17. but then the particular subgroup would be weighted to compensate for the small sample

For example, if you polled 100 people, 88 women and 12 men, you would obviously not weight the results for a population that was distributed 50/50, even if the sample was perfectly random

But if the sample were 54/46, and the sampling was good, you would have a good subgroup estimates.

[karynnj]

20. You are confusing the estimate with the moe

The moe of the sub group will be based on the sample size of the group and it would be huge. The moe of the total would be computed based on a formula that is based on the moes of the groups and the population weights.

No pollster were intentionally undersample a group. It makes their poll less accurate.

[factfinder_77]

21. Trump get 6 % of dem voters, Clinton only get 4 % of rep. Trump get 39 % women, Clinton only 41 %

The poll sample does not make any sense.

[Doctor Jack]

13. The gains we are seeing this week for clinton are mostly nationally

State polls almost always lag behind national polls, so expect to see a jump in the swing states maybe early next week.

[whistler162]

18. No if done on Wednesday and a odd day rule!