Friday, October 26, 2012

Size Really DOES Matter

Some weeks ago I noticed that, as the days progressed, the more recent polls within the RCP average showed a far more positive result for Governor Romney than that shown by the RCP average itself.  All you had to do was bring forward the 'drop off' date to eight days from the present.   This indicated that Mr Romney was overtaking the Black Arsed JackAss whose only good news was old news.

PollDateSampleMoERomney (R)Obama (D)Spread
RCP Average10/15 - 10/24----47.847.1Romney +0.7
Associated Press/GfK10/19 - 10/23839 LV4.24745Romney +2
Rasmussen Reports10/22 - 10/241500 LV3.05047Romney +3
IBD/TIPP10/19 - 10/24948 LV3.54547Obama +2
Gallup10/18 - 10/242700 LV2.05047Romney +3
ABC News/Wash Post10/20 - 10/231394 LV3.04948Romney +1
Monmouth/SurveyUSA/Braun10/18 - 10/211402 LV2.64845Romney +3
NBC News/Wall St. Jrnl10/17 - 10/20816 LV3.44747Tie
CBS News10/17 - 10/20790 LV4.04648Obama +2
WashTimes/JZ Analytics*10/18 - 10/20800 LV3.54750Obama +3
Politico/GWU/Battleground10/15 - 10/181000 LV3.14947Romney +2

A far more observant observer than I (I'm damned if I can find where I saw the piece but when I do I'll put up the link.  Found it!) has noticed that there is a consistent variation depending on sample size.  All the polls in which more than 1,000 likely voters were sampled are strongly in favour of President Romney.  With one recent exception, posted since the piece was written, every one of the polls with less than 1,000 likely voters sampled favours Obama.

I haven't yet looked at the internals to see what the Democrat/Republican split is for the smaller polls but I can't help wonder whether those outfits which shill for Obama are too lousy to spend the large amounts of money it takes to phone up 1,000 people to talk to just thirty.  Yes, the last figure I saw for poll respondents was just three percent.

My guess is the second biggest loser after Obama on November 6th will be the US media associated public polls.


Lindsay Addie said...

The trends in these polls is getting most interesting. I still think the actual numbers need to taken with a grain of salt but there is no denying what the trend currently is.

The ABC\WaPo poll has just been updated and is now showing Romney is 3% ahead. It also reveals compared to 2008 the GOP enthusiasm levels are much higher.

Anonymous said...

I think you'll find the pollsters have 1,000 respondents, not just 30. That is, they are calling about 33,000 people, and getting 1,000 who complete the survey.

Also, the difference in poll sizes you give isn't that great, so the difference in margins of error of the smallest to largest polls is not that great. That is, the larger polls have smaller MoE, so tell us the Romney/Obama support levels a bit more accurately, but not by that much...

Just saying.

The Veteran said...

Like it or not the outcome is NOT determined on the popular vote but on the Electoral College vote and, from what I see and read, that still looks good for Obama.