HomeLatest ThreadsGreatest ThreadsForums & GroupsMy SubscriptionsMy Posts
DU Home » Latest Threads » Forums & Groups » Retired » Retired Forums » 2016 Postmortem (Forum) » Stop with the "CNN didn't...

Wed Sep 14, 2016, 08:46 PM

Stop with the "CNN didn't poll anyone under 50"

They do not report data for subgroups with margins of error above a certain threshold.

17 replies, 3411 views

Reply to this thread

Back to top Alert abuse

Always highlight: 10 newest replies | Replies posted after I mark a forum
Replies to this discussion thread

Response to Loki Liesmith (Original post)

Wed Sep 14, 2016, 08:51 PM

1. It seems harder than it used to be to find data on the demographics of respondents.

ORC used to report those numbers with their polls but I didn't see any this time. And I've noticed this with other polls, too.

It's almost like they don't want us to know.

Reply to this post

Back to top Alert abuse Link here Permalink


Response to pnwmom (Reply #1)

Wed Sep 14, 2016, 09:03 PM

3. I posted links

to the CNN/ORC polls on another thread. BTW, if N/A is listed in one of the crosstalk it means that there were fewer than about 140 likely voters in that group. I figure that for the FL poll no more than 1/4 of the likely voter respondents were under 50. The LV screen probably kicks a lot out of the original sample.

Reply to this post

Back to top Alert abuse Link here Permalink


Response to machomaas (Reply #3)

Wed Sep 14, 2016, 09:07 PM

5. Yes, MOE would be > 8%

they won't report that.

It may still be a flawed sample, but not at all clear.

Reply to this post

Back to top Alert abuse Link here Permalink


Response to machomaas (Reply #3)

Thu Sep 15, 2016, 01:14 AM

15. I saw that but I didn't see how many respondents they had in each category. They used to include

answers to those questions (age, race, income, etc.) in the report of the survey itself.

Reply to this post

Back to top Alert abuse Link here Permalink


Response to Loki Liesmith (Original post)

Wed Sep 14, 2016, 09:03 PM

2. Your concern concerning this concern is concerning. eom

 

Reply to this post

Back to top Alert abuse Link here Permalink


Response to MohRokTah (Reply #2)

Wed Sep 14, 2016, 09:05 PM

4. Who's concerned?

HRC is likely to win.

But I do like data based critiques to actually be data-based.

Reply to this post

Back to top Alert abuse Link here Permalink


Response to Loki Liesmith (Original post)

Wed Sep 14, 2016, 09:08 PM

6. Fair to ask if they interviewed ENOUGH people under 50

I get maybe not having enough for 18-29, but 35-49 too? Don't buy that

Reply to this post

Back to top Alert abuse Link here Permalink


Response to MadBadger (Reply #6)

Wed Sep 14, 2016, 09:10 PM

7. sure. always fair to ask

Who knows? This week is fucked pollingwise. We were lucky to not get hammered in QPac too.

I'm hoping they do not Poll NH or PA till next week. Worried we'd get a bad result by accident.

I may just tune out for a week, this place gets too hype.

Reply to this post

Back to top Alert abuse Link here Permalink


Response to Loki Liesmith (Original post)

Wed Sep 14, 2016, 09:37 PM

8. MoE is calculated from sample size and they are inversely proportional.

So if the 50+ demographic has a large MoE then it is likely that the sample of 50+ was small.

Reply to this post

Back to top Alert abuse Link here Permalink


Response to Persondem (Reply #8)

Wed Sep 14, 2016, 09:57 PM

9. aware

but the sample size would be small anyway. It is not *nonexistent*.

Reply to this post

Back to top Alert abuse Link here Permalink


Response to Loki Liesmith (Original post)

Wed Sep 14, 2016, 10:01 PM

10. Agreed. It's a bad poll for us, but a legit poll. Poll truthering doesn't accomplish anything.

Reply to this post

Back to top Alert abuse Link here Permalink


Response to DanTex (Reply #10)

Wed Sep 14, 2016, 10:11 PM

11. I expected a point or two movement in the national polls to HRC this week

that seems to have happened.

That pushes close states, like FL and OH, over to Trump.

If his share of the national vote ebbs again, we will see polls we like better in the close states.

Reply to this post

Back to top Alert abuse Link here Permalink


Response to Loki Liesmith (Original post)

Wed Sep 14, 2016, 10:18 PM

12. CNN only polled 18% of voters under 50 (less than half)

Politicususa has the story. I'm on my phone & can't paste the link. The updated CNN Poll story is at the end of the story.

Reply to this post

Back to top Alert abuse Link here Permalink


Response to displacedtexan (Reply #12)

Wed Sep 14, 2016, 11:01 PM

13. 18%

under 50? Sounds bizarre. Why was that?

Reply to this post

Back to top Alert abuse Link here Permalink


Response to Loki Liesmith (Original post)

Thu Sep 15, 2016, 12:10 AM

14. Some subgroups represent too small a share of the pop of Ohio to produce an accaptable sample error

 

Crosstabs on the following pages only include results for subgroups with enough unweighted cases to produce a sampling error of
+/- 8.5 percentage points or less. Some subgroups represent too small a share of the population of Ohio 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.

Reply to this post

Back to top Alert abuse Link here Permalink


Response to Loki Liesmith (Original post)

Thu Sep 15, 2016, 02:06 AM

16. If you're going to get into the weeds of the cross tabs on this one

You also have to do the same with the polls that show us ahead, so we know that similar undercounts aren't happening with, for example, whites without a college degree or voters 65+.

I prefer not to do either, just take the good with the shitty and plow ahead. I have faith in our GOTV and (somewhat) in the American voter (I hope).

Reply to this post

Back to top Alert abuse Link here Permalink


Response to Loki Liesmith (Original post)

Thu Sep 15, 2016, 03:02 AM

17. Only 18% of their sample was under 50. Less than half of what it should have been.

http://www.democraticunderground.com/12512418223

Even if they weighted the final results to account for age, their initial small sample could have been skewed.

Reply to this post

Back to top Alert abuse Link here Permalink

Reply to this thread