Free Republic
Browse · Search
News/Activism
Topics · Post Article

To: usafa92
The reason RCP doesn’t include them is because they are favorable to Trump.

No, RCP does not include them because they have no controls and are not statistically valid. Anyone could take the poll as many times as they like, could describe themselves any way they want, etc. Even the people who did the "poll" admit they cannot even calculate an MOE because they cannot validate any of the data...

43 posted on 01/05/2016 6:49:34 AM PST by CA Conservative (Texan by birth, Californian by circumstance)
[ Post Reply | Private Reply | To 39 | View Replies ]


To: CA Conservative

You really need to educate yourself on how online polling works. You are the lone voice on here constantly carping about it. You make the ridiculous assertion that this is a quasi Drudge poll which is not true. Sorry Cruz continues to get smoked nationally, but this poll is in line with every other national poll released in the last few weeks.


46 posted on 01/05/2016 6:56:16 AM PST by usafa92 (Conservative in Jersey)
[ Post Reply | Private Reply | To 43 | View Replies ]

To: CA Conservative

They most certainly do have controls. They are taken from 3 million daily respondents and as shown on page 7 of 7 are sampled to develop a confidence measure. All polls come with disclaimers. Some have been historically more accurate than others.

http://www.scribd.com/doc/294643292/NBC-News-SurveyMonkey-Weekly-Election-Tracking-Poll

The analysis of how accurate survey monkey has been over the years suggests that in 2014 it was slightly below 95% accurate.

**For a comparison, the absolute average accuracy for the US Presidential race was 0.055 in 2012 with three out of twenty-five polls falling outside of the 95% confidence intervals and 0.037 in 2008 with none of the nineteen polls falling outside of the 95% confidence intervals.

Accuracy problems with the SurveyMonkey polls are found across sample sizes. Seven out of the ten polls with sample sizes of less than 1,100 have accuracy measures outside of the 95% confidence intervals and ten out of the fifteen polls with sample sizes of 4,000 or more have accuracy measures outside the 95% confidence intervals.

Survey Monkey has stepped up their game in 2016 Pollsters routinely use statistical adjustments to make their poll results better predict election outcomes, because the small percentage of adults who actually respond to a survey usually aren’t representative of the whole electorate. Blumenthal says. SurveyMonkey pollsters may need to do fewer of these mathematical corrections because it has access to such a broad range of Americans. “They’re incredibly diverse,” he says of the people who respond to polls through the site.

http://www.xconomy.com/san-francisco/2015/10/22/surveymonkey-ups-cred-as-pollster-hires-election-czar/2/


51 posted on 01/05/2016 7:04:17 AM PST by JayGalt
[ Post Reply | Private Reply | To 43 | View Replies ]

To: CA Conservative
Anyone could take the poll as many times as they like, could describe themselves any way they want, etc.

I think most of these polls do not allow multiple votes from the same IP address. Some even require registration or at least an authentic email address.

76 posted on 01/05/2016 1:48:37 PM PST by itsahoot (Anyone receiving a Woo! Woo! for President has never won anything after the award.)
[ Post Reply | Private Reply | To 43 | View Replies ]

Free Republic
Browse · Search
News/Activism
Topics · Post Article


FreeRepublic, LLC, PO BOX 9771, FRESNO, CA 93794
FreeRepublic.com is powered by software copyright 2000-2008 John Robinson