2016 Postmortem
Related: About this forumSo, According to CNN, There Are No Minorities in Colorado
Bullshit Poll![url=https://flic.kr/p/LEgsG1][img][/img][/url][url=https://flic.kr/p/LEgsG1][/url] [url=https://www.flickr.com/photos/vdogg/][/url]
99% white? And we're supposed to take this seriously?
vdogg
(1,384 posts)These are Colorado's
[url=https://flic.kr/p/LEm2Nc][img][/img][/url][url=https://flic.kr/p/LEm2Nc][/url] [url=https://www.flickr.com/photos/vdogg/][/url]
Links to Both
http://i2.cdn.turner.com/cnn/2016/images/09/26/relpa1.pdf
http://i2.cdn.turner.com/cnn/2016/images/09/26/relco1.pdf
remaineruk
(156 posts)Bernardo de La Paz
(49,035 posts)Bernardo de La Paz
(49,035 posts)and if you had seen this you would have NOT written what you did.
pling error of +/- 8.5 percentage points or less. Some subgroups represent too small a share of the population of
Colorado to produce crosstabs with an acceptable sampling error. Interviews were conducted among these subgroups but results for groups with a sampling error larger than +/- 8.5 percentage points are not displayed.
Coolest Ranger
(2,034 posts)We're never polled on anything
Response to vdogg (Original post)
Post removed
remaineruk
(156 posts)Across each area? It means it isn't reflective of the polling area per say. And therefore is not reflective of the state. And therefore is as the OP has said is not valid
klook
(12,165 posts)Let's work together to get out the youth vote for Hillary.
Ilsa
(61,697 posts)I have a hard time believing that.
pangaia
(24,324 posts)remaineruk
(156 posts)And instead are concentrating on other polls. I do find it hard to believe that CNN are trying to create valid polls if they can not find one under 35 and or a non white in around 1000 people. It's just a click bait for twitter
Bernardo de La Paz
(49,035 posts)It's not that "they cannot find one". It's that the number they found was not statistically significant enough to support a meaningful breakdown for that category.
The minorities and younger voters are included in the main totals.
It's not CNN's fault and it is not a nefarious conspiracy inside CNN. It is simply how things work out in a state like Colorado or Pennsylvania when you have a relatively small poll (about a thousand).
The overall numbers do have significance so don't try to read too much into the breakdowns and take ALL polls with a grain of salt because there is only one poll that counts: Election Day.
remaineruk
(156 posts)In a population containing 23% non white residents I think it is fair enough to question a poll that couldn't get enough of a sample. ...not even 1% of those polled to show up in the demographics. Just saying
Bernardo de La Paz
(49,035 posts)It's way more than 1% (way more than 10 people), but by time you break that down into the categories you end up with small enough groupings that the whole column is invalidated due to sampling error. Even if minorities are under-sampled some (which is likely), it is not going to miss 90% of them.
Look at the sampling error for the other columns. When the sampling error is over something like +/- 8 %, then they can't include them in the breakdown. It would be nonsensical.
It seems that you have very little acquaintance with statistical analysis, perhaps none beyond high school. I took a couple of courses in university, but pollsters have degrees in the subject and I respect their judgement. Also, it is difficult and expensive to run polls, especially if you are running them in 50 states with 1,000 or more in each state. That's at least 50,000 people.
And this "just saying" crap is like "whatever": a dismissive meaningless motion of air molecules. It is less than unconvincing. Best to leave it off.
remaineruk
(156 posts)I know that excluding the views of 23% of the population...oh and that doesn't include the % of under 35s btw probably doesn't give the most reflective poll of how a state will vote. Some polls are outliers they just are.
Bernardo de La Paz
(49,035 posts)Let me guess, your highest math course was high school and you learned about average and median and the bell curve and not much more?
You have reached your conclusion and you don't read my points or think about them because you aren't responding to them in any kind of meaningful way.
Your repetition of your single only "point", which has been shown to be false, does not make it stronger and convinces nobody who hadn't already closed their mind after reading the Original Post.
BlueInPhilly
(870 posts)"They cannot find one" - please don't take it too literally. The truth is they couldn't find enough sample to provide a meaningful result for certain population segments. Once you take this into consideration, then you render your entire sampling false, and your results invalid.
It may not be nefarious, but it is biased and misleading and should never have been released.
Bernardo de La Paz
(49,035 posts)When people write something and use a number, we do take it literally.
Bombast and hyperbole are ultimately not very convincing and may be counter-productive.
The sample IS meaningful for the whole population, including the minorities. But the subsamples are not meaningful for breakdown analysis. Big difference. Look at the huge margins of error (8 %) on some of the other breakdowns. It's fundamentally a small poll, too small to encompass all the desired breakdowns, but that does NOT negate the results reported for the questions.
BlueInPhilly
(870 posts)You seem to be intent on arguing with everyone who disagrees with you.
I stand with my assessment: a sample that has immaterial segments should not be published. It is misleading. Not in a election where population segments are crucial and should be sampled accordingly.
Bernardo de La Paz
(49,035 posts)BlueInPhilly
(870 posts)That's the fallacy of polls. Any polls.
You did say... they did not have any data to meaningfully slice and dice by segment. That there is a tell.
I stand by my assessment.
uponit7771
(90,359 posts)Bernardo de La Paz
(49,035 posts)vdogg
(1,384 posts)Polling no minorities or having so low a response rate of minorities that the results cannot be tabulated has the same effect. In the end it is not a poll that accurately reflects the population at large.
Bernardo de La Paz
(49,035 posts)Minorities are INCLUDED in the overall counts, but they can't be meaningfully broken down by age or minority, statistically.
They are accurately reflected in the overall totals (within sampling error and subject to proper statistical caveats).
It is false to suggest that they did not poll minorities or that they did not tabulate them.
They DID tabulate them.
But though they affect the overall outcome, it is not possible to analyze the DEGREE of their effect.
remaineruk
(156 posts)But it doesn't mean their conclusions are incorrect
Bernardo de La Paz
(49,035 posts)remaineruk
(156 posts)vdogg
(1,384 posts)If they poll 1000 people, and they ask each person their race, they damn well know exactly how many Blacks/Hispanics/etc. participated in their poll. This is not an unknowable quantity. Even if the sampling error is great due to the small sample size, the raw totals should be included for transparency. Anything less is suspect.
Bernardo de La Paz
(49,035 posts)vdogg
(1,384 posts)Excluding that I truly don't believe that Trump is anywhere close to 40% minority support, their likely voter screen seems to be very aggressive. It's hard for me to believe that they're using either 2008 or 2012 numbers for that screen. It seems that they're expecting an electorate closer to 1980.
Bernardo de La Paz
(49,035 posts)kennetha
(3,666 posts)Then Trump would STILL lose.
But how much are you willing to bet me that minority turnout in Colorado will be insignificant. If CNN had any integrity as a news agency, it would explain that its likely voted screen, together with its methodology, eliminated ALL minorities from this and many other cross tabs.
It would also explain that its sample if minorities was so small that even in the registered voter cross tabs, its same of minority voters was so small as to have an 8% error rate.
Then they would help the listener understand how to weight a poll with such problems.
But that would undercut the point of their polling -- which is merely to generate headlines about the horse race and to give the lazy and somewhat brain-challenged anchors something to blather on about without having really to do any serious intellectual labor.
NoGoodNamesLeft
(2,056 posts)I live in a battleground state and have been polled several times. They have attempted to poll my adult daughter too, but she won't do it. Each time they call they ask your race and age. They called for her over the weekend and she would not do it. I said I would and they asked how likely I am to vote, my race and my age. I was too old to be polled for this one. If no minorities are being polled then Hillary is doing WAY better than any of these polls imply. That poll shows ONLY white and maybe Native Americans. It doesn't mean that there are no minorities in Colorado. It just means that they were not included in this poll.
Bernardo de La Paz
(49,035 posts)You are accusing them of unbalancing the poll by exclusion when in fact they are working hard (making extra calls) to balance it by INCLUSION.
NoGoodNamesLeft
(2,056 posts)When I gave my age they said thank you for your time. I asked if they were only polling younger people and the person said yes.
In order to find out what younger voters are thinking and care about they need to poll ONLY certain age groups. If you poll all ages and groups together then it's more difficult to determine what issues matter and be able to get the most out of the research.
I'm certainly not offended by it. It makes sense to actually poll only younger voters before the media talks about how young voters are likely to vote. I was polled several times already when they were targeting my age range. You would think that since they called the same number and asked for me by name and then my daughter by name and that we have the same last name and registered on the same day that they would have called her back the same day they called me when they were polling a week ago, but they didn't.
Here's something everyone should be aware of...polls, just like surveys, statistics and any other research can and often is targeted. Anyone who has worked in organizations or agencies depending on grant writing will tell you that.
but data seems to imply that 18-34 broke for Trump. Very hard to believe. How did I arrive at this?
In the 3 age groups with data Trump wins by 1% but overall he is up 2%. So the 18-34 age group definitely broke for him??