2016 Postmortem

aaaaaa5a

(4,667 posts)

10. With smaller sub-groups within a poll there is a greater margin of error.

Thu Oct 25, 2012, 12:56 AM

Oct 2012

This is why the polling for hispanics and African Americans is frequently off. In the Survey USA poll of Ohio (O+3) they had Romney winning 22% of the black vote.

What happens is, say you have a poll of 100 people. In a group this size, maybe only 3 of them might be hispanic. If you happen to get one person who likes Romney..... BANG..... you have hispanics giving 33% of their vote to the GOP.

You have to be very careful when pulling out minority data from any poll.

Edit history

Please sign in to view edit histories.

10 replies

= new reply since forum marked as read

Highlight:

PPP: Barack Obama now leads 49-48 in our national tracking. 3 point gain since he was down by 2 on M [View all] smorkingapple Oct 2012 OP

Great News! n/t courseofhistory Oct 2012 #1

They're catching up to the better pollsters Maximumnegro Oct 2012 #2

Could be mere noise, but it's welcome noise. nt geek tragedy Oct 2012 #3

Wow IrishMidget Oct 2012 #4

MITTMENTUUUM!!!! Maximumnegro Oct 2012 #5

If nothing else, better poll results for Obama primaryblue Oct 2012 #6

58/37 with Hispanics seems low. NT Bleacher Creature Oct 2012 #7

That's what I was thinking ncav53 Oct 2012 #8

With smaller sub-groups within a poll there is a greater margin of error. aaaaaa5a Oct 2012 #10

Looks like Romney threw $70 million at an insoluble problem alcibiades_mystery Oct 2012 #9