Methodology - The Fairleigh Dickinson University poll was conducted by landline and cellular telephone September 28-October 2, 2016 among a random national sample of 862 registered voters, including 788 likely voters. Results have a margin of sampling error of +/- 4.4 percentage points for likely voters (the basis for the analysis presented in this press release), including the design effect.
Survey results are also subject to non-sampling error. This kind of error, which cannot be measured, arises from a number of factors including, but not limited to, non-response (eligible individuals refusing to be interviewed), question wording, the order in which questions are asked, and variations among interviewers.
Interviews were conducted by SSRS of Media, Pennsylvania, with professionally trained interviewers using a CATI (Computer Assisted Telephone Interviewing) system. Random selection is achieved by computerized random-digit dialing. This technique gives every person with a landline or cellular phone number (including those with unlisted numbers) an equal chance of being selected
Does not include the VP debate either.
Margin of error is nearly double what my stats professors said was considered acceptable. crappy poll =outlier.
CC
“This kind of error, which cannot be measured, arises from a number of factors including, but not limited to, non-response (eligible individuals refusing to be interviewed)”
So evidently a non-response was placed in the Hillary column.
One way the poll is misleading is that it calls itself a poll of likely voters.
In fact it’s a poll of of “registered voters, including likely voters.”
It’s not a likely voters poll. Of course any poll of registered voters will likely voters by definition.
September 28 - October 2 being released and reported today as new information...making it essentially misinformation. That data is too old by now with other events occurring to affect the race...namely a nationally televised debate.