2016 Postmortem
Related: About this forumNBC Online Survey: Hillary Clinton's Lead Narrows (Hill down 7; Bern up 4 in a month)
Last edited Tue Sep 22, 2015, 02:06 PM - Edit history (1)
http://www.nbcnews.com/meet-the-press/nbc-online-survey-hillary-clintons-lead-narrows-n430996Hillary Clinton's lead over Bernie Sanders has dwindled from 24 points to just 13 points over the last month.
According to the latest NBC News online survey conducted by SurveyMonkey from Wednesday night through Friday, 42% of Democrats and independents who lean Democrat nationwide support Clinton, compared to 29% who support Sanders. Just five months ago, Clinton led Sanders by 45 points in the April NBC News-SurveyMonkey poll. In spite of this weakening in support, a majority of Democratic voters continue to say they expect Hillary to win the Democratic nomination for president.
Clinton's standing has fallen among a number of key voting groups, including those over 65 (-11 points and between 45 and 64 years (-9), college graduates (-10), and both whites (-10) and blacks (-11). Both Sanders and Joe Biden, who has not officially entered the race, have made small gains in each of these groups.
still_one
(92,394 posts)methodology has differences from the other pollsters.
Here is a good article from 538 discussing it:
http://fivethirtyeight.com/features/surveymonkey-was-the-other-winner-of-the-u-k-election/
Wankle Ronnie
(66 posts)But I have never seen them take political polls up until now.
I've been surveyed via Surveymonkey by local clubs, schools, businesses. They are very good.
magical thyme
(14,881 posts)they've been doing online surveys for business forever...or since the beginning of the internet anyway
I've had to take a bunch of surveymonkey survey's over the years in various jobs...
still_one
(92,394 posts)magical thyme
(14,881 posts)Two changes in the 16 years since the company was founded have eased that challenge. Just about everyone in the U.S. and U.K. is online, and fewer people are answering when telephone pollsters come calling.
Telephone pollsters who dial numbers selected randomly sometimes say they have a probability sample, meaning that respondents are selected randomly. But the lower response rates go, the greater the risk that random-digit dialers arent getting a random group of respondents, but instead are collecting information from a self-selected subgroup of people who are willing to answer a strangers questions. There is no such thing as probability sampling in 2015, Jay Leve, who runs telephone-polling company SurveyUSA, said in an email. Its an archaic concept.
artislife
(9,497 posts)I have a SurveyMonkey account. I have used it for clients etc.
This is the first I knew they did the polling. Very interesting article, thanks for bringing it over. Now to reply to the posts I made in another thread.
Doh!
magical thyme
(14,881 posts)more good news for insurgent candidates!
Biden is also up 4 -- looks like he and Bernie mostly split Hillary's loss.
Historic NY
(37,453 posts)thats where the rubber meets the road.
magical thyme
(14,881 posts)sabrina 1
(62,325 posts)which is coming from people that are not being polled in these traditional polls.
It's great to see Bernie getting support from the Dem base, but that was never his real goal. He was aiming at the Indy vote, which he has, and at the that huge demographic of people who gave up on politics, and opted out. I know for a fact he is succeeding with that, having signed up some of those non voters myself.
magical thyme
(14,881 posts)that doesn't give even a clue as to what's really going on.
sabrina 1
(62,325 posts)are now trying to use cell phones, since so few people have landlines anymore, particularly the young. While people with cell phones were never polled.
This came up when statistics showed a few years ago that fewer and fewer people were using landlines anymore. Polling companies haven't caught up to that yet.
Godhumor
(6,437 posts)First of all, you have to be a Survey Monkey user. Secondly, you have to self identify that you would like to be considered for polling. Right in the methodology it states that this poll had no MoE because it is not a randomized sample. Instead they calculate a bootstrap equivalent (Quite literally they use assumptions to guess what the MoE would be if it were a random sample).
The sample bias in this poll, regardless of results, is staggering.
magical thyme
(14,881 posts)they explain that right in the 538 article; I posted the paragraphs about it above.
Again, for your convenience:
Two changes in the 16 years since the company was founded have eased that challenge. Just about everyone in the U.S. and U.K. is online, and fewer people are answering when telephone pollsters come calling.
Telephone pollsters who dial numbers selected randomly sometimes say they have a probability sample, meaning that respondents are selected randomly. But the lower response rates go, the greater the risk that random-digit dialers arent getting a random group of respondents, but instead are collecting information from a self-selected subgroup of people who are willing to answer a strangers questions. There is no such thing as probability sampling in 2015, Jay Leve, who runs telephone-polling company SurveyUSA, said in an email. Its an archaic concept.
Godhumor
(6,437 posts)Telephone callers still call from a truly random list. They might complain about the samples they receive, but they are starting from a general population methodology. This is why they have MoEs.
Survey Monkey doesn't even bother with a general population. Only participants who are on Survey Monkey and have self selected willingness to participate are polled, which makes it literally impossible to compare to general population (hence the no MoE).
All polls have issues, no doubt--especially with teaching younger voters. But internet polls notoriously start poorly to begin with due to inherent sample bias.
magical thyme
(14,881 posts)selected.
One thing the NBC/Survey Monkey poll doesn't do that phone polls do is 'weight' responses as to how likely they think somebody is to vote given the response levels in their demographic group. They believe that just the fact that somebody bothers to respond at all indicates a likelihood of voting.
And while their track record is short, it is also quite good:
SurveyMonkey has polled (US) elections before. It correctly identified the winner in 48 of 50 states and Washington, D.C., in the 2012 U.S. presidential election, both gubernatorial races in 2013, all 36 Senate races in 2014, 33 of 36 gubernatorial races that year, plus the Washington, D.C., mayoral race.
They've also been very successful with UK elections.
Godhumor
(6,437 posts)Clinton got good news from the CNN poll and Sanders got good news from this one. If I come off as biased here, I apologize. I am not trying to be snotty; I just professionally don't hold internet polls in high regard. I am certainly not trying to dismiss a good result for Bernie out of hand. Apologies if it came off that way.
magical thyme
(14,881 posts)it also has a very good track record so far.
Godhumor
(6,437 posts)To use the phrasing from the article, I see a sample size of 1 not of 5000 individuals because there is no population for comparison.
But I am really not trying to be an asshole here, so I will disengage from the thread. Otherwise I will become one of those people that shouts, "Don't listen to this poll because of REASONS!" And I really don't want to be that poster.
magical thyme
(14,881 posts)"SurveyMonkey has polled (US) elections before. It correctly identified the winner in 48 of 50 states and Washington, D.C., in the 2012 U.S. presidential election, both gubernatorial races in 2013, all 36 Senate races in 2014, 33 of 36 gubernatorial races that year, plus the Washington, D.C., mayoral race."
They've also been very successful with UK elections.
DCBob
(24,689 posts)That's about right.
Wankle Ronnie
(66 posts)Especially if the Biden voters are actually Anyone but Clinton voters..
DCBob
(24,689 posts)DCBob
(24,689 posts)from the PPP Iowa poll..
"Among Biden's voters 43% say Clinton would be their second choice to only 15% for Sanders. Reallocate them to their second choice and Clinton would lead Sanders 50/25, almost identical to the 52/25 lead we found for her last month when we didn't include Biden"
Historic NY
(37,453 posts)neither is Hillary she benefits the most.
sabrina 1
(62,325 posts)Bernie, and he is way ahead.
DCBob
(24,689 posts)upaloopa
(11,417 posts)not a random poll
Means nothing
magical thyme
(14,881 posts)Survey Monkey starting doing political polling a couple years ago. They're just now going public with it, due to their good results. They don't use typical online poll methodologies, nor do they use telephone poll methodologies.
"SurveyMonkey has polled (US) elections before. It correctly identified the winner in 48 of 50 states and Washington, D.C., in the 2012 U.S. presidential election, both gubernatorial races in 2013, all 36 Senate races in 2014, 33 of 36 gubernatorial races that year, plus the Washington, D.C., mayoral race."
They've also been very successful with UK elections.
upaloopa
(11,417 posts)Study statistics
magical thyme
(14,881 posts)Telephone polls no longer have random selection. They are self-selection due to the high percentage of people who refuse to answer them.
Try reading the article. The telephone poll people say this, not me.
"Two changes in the 16 years since the company was founded have eased that challenge. Just about everyone in the U.S. and U.K. is online, and fewer people are answering when telephone pollsters come calling.
Telephone pollsters who dial numbers selected randomly sometimes say they have a probability sample, meaning that respondents are selected randomly. But the lower response rates go, the greater the risk that random-digit dialers arent getting a random group of respondents, but instead are collecting information from a self-selected subgroup of people who are willing to answer a strangers questions. There is no such thing as probability sampling in 2015, Jay Leve, who runs telephone-polling company SurveyUSA, said in an email. Its an archaic concept.
magical thyme
(14,881 posts)"SurveyMonkey has polled (US) elections before. It correctly identified the winner in 48 of 50 states and Washington, D.C., in the 2012 U.S. presidential election, both gubernatorial races in 2013, all 36 Senate races in 2014, 33 of 36 gubernatorial races that year, plus the Washington, D.C., mayoral race."
They've also been very successful with UK elections.
aspirant
(3,533 posts)how many times you have to repeat these facts. The power of Brainwashing is amazing.
All polls must be verifiable.