UPDATE: FOX 31’s Eli Stokols reports on this poll, and two other corroborating inside-MoE polls released today from CNN and Rasmussen. Also picked up by Political Wire and Alicia Caldwell of the Denver newspaper.
The first in a series of poll results for key Colorado races, released by Colorado Pols and conducted by Denver-based RBI Strategies & Research. Today, the U.S. Senate race:
Republican Ken Buck and Democrat Michael Bennet are locked in one of the most competitive US Senate races in the country. Currently, 43% of likely voters in Colorado are supporting or leaning towards supporting Bennet while 42% are supporting or leaning Buck. 3% say they will support the Libertarian, Maclyn Stringer, and 1% say they will support the Green Party candidate Bob Kinsey. Men and women are mirrored in their preferences, with men favoring Buck 45% to 41% while women favor Bennet 44% to 39%.
Both candidates benefit from strong support from their respective parties, with 81% of registered Democrats favoring Bennet and 80% of Republicans favoring Buck. Bennet holds a 41% to 34% advantage among Unaffiliated voters. Voter preferences vary widely by region with Bennet leading in the North Front Range and Denver while trailing in all other regions of the state. Voter preferences are also correlated to age, with voters under 45 favoring Bennet by 5 points and voters over 65 favoring Buck by 7. While Buck holds a 5 point lead among white voters, Bennet leads by double digits among Hispanic voters.
RBI Strategies & Research conducted a telephone survey of 501 Colorado voters who indicated it was likely that they would vote in the 2010 General Election. Interviews were conducted October 24 – October 26, 2010 by Standage Market Research of Denver, Colorado, a market research firm specializing in telephone survey interviewing. Respondents were randomly selected from a list of Colorado voters, purchased from Voter Contact Services, who voted in the 2008 General Election or registered to vote at any time following the 2008 General Election.
The margin of error for a survey of 500 interviews is +/- 4.4% at the 95% confidence level. The margin of error is higher for subsamples within the full sample. Other sources of error not accounted for by the stated statistical margin of error include, but are not limited to, question wording, question order, refusal to be interviewed, and demographic weighting.
Kevin Ingham of RBI Strategies will join us TODAY in this thread from 1-3PM to answer your questions about this poll. Tomorrow, we’ll release numbers on the gubernatorial race, and Friday on major statewide ballot initiatives. Kevin will join us for Q&A each day in comments.
Please be respectful in your comments and questions for Mr. Ingham. We appreciate the time that he is making for this Q&A session, and whether you agree or disagree with anything he says, there is NO reason you need to voice your opinion in a rude or disrespectful manner. We will not tolerate bad behavior from anyone during this Q&A session, so please be good Polsters.
I’ll be logging in around 1pm to answer questions you might have about the poll, the methodology, etc.
Ellie asked it yesterday.
How did you handle people who already voted?
And it looks like the distribution of early voting numbers reported by the SOS are very close to the reported early voting in the poll sample — slightly more Republicans than Democrats.
I hadn’t gotten to the crosstabs yet.
We ask two questions. First, we ask the voter if they are over the age of 18 and registered to vote in Colorado. We only called persons who were registered to vote (our sample came from a list of registered voters who either voted in 2008 or have registered since then) but this is standard practice just to make sure we have the right person on the phone.
Our next question asks how likely it is that the respondent will vote in the November election. We read them several posibilities: already voted, absolutely certain, very likely, probably, not very likely and not at all likely.
If anyone told us it was not very likely or not at all likely they will vote, we terminated the interview. If any of the other responses were given, we continued with the interview.
In the crosstabs, which are posted above, you can see those respondents who said they already voted broken out separately from the other responses.
Let me know if that is what you were asking.
Good to know you merely terminated the interview. I was reading the questions (Topline) and was concerned the surveyors were being directed to terminate citizens if they admitted they were under 18 and/or not registered to vote.
With record unemployment, levels not seen since 1948, this segment has learned an early lesson in life.
but let me try to address the question I think you are asking.
In 2006, voters under the age of 25 were 10.39% of those who had voted in 2004 or registered after the election (trying to compare apples to apples since our sample was pulled from voters who voted in 2008 or registered to vote after that election). However, they only made up 5.09% of those who voted in the 2006 election.
This year, under 25 voters are 7.4% of those that voted 2008 or registered after.
I imagine we will see some similar level of drop off with regards to their share of the electorate.
Does that answer your question?
The prevailing theory is Libby’s a bot with some type of rudimentary AI. His ‘question’ is undoubtedly what one might call ‘rhetorical.’
It’s a club. Welcome to it.
Your answer is interesting though. Thank you.
Why not more than 501?
Of the 501, virtually even R & D split, and only 22% Us.
Did you choose this weighting?
What counts as “metro”?
Why only 23% of respondents – I thought 50% of the voters were in the metro.
“Metro” is distinct from Denver, and also distinct from other counties normally considered metro.
There’s a breakout for Denver, and Metro, and Northern Front Range, all of which include metro area counties.
Denver is Denver. Metro is Arapahoe, Jeffco, Clear Creek and Gilpin. Northern Front Range lumps in Adams, Broomfield, Boulder, Larimer and Weld.
Douglas County is grouped with Eastern Plains and South Front Range — also including El Paso, Pueblo and all the plains counties.
That does seem like a strange way to sort the regions.
We can obviously quibble about how to break up the state. However, when we are trying to make useful crosstabs, there is only so small we can go in terms of the sample size before the margin of error explodes. We tried to do it in a way that helps us give a sense of the difference between the North Front Range and South Front Range. They are very different politically so we like to try and show that.
If you want me to make some different combination of counties, let me know which you want grouped. I can try to make that and tell you what we are looking at.
Libertad, bjwilson83, and H-man are all shilling for the GOP, so take any of their questions with a boxful of salt.
to note, I now see the crosstabs, so I’ll go there to review my questions.
So am I. Always have been.
I noticed th poll was “was weighted by age and registered party to more accurately reflect the likely makeup of the electorate as determined by turnout in past election cycles.”
First, this seems a little vague. Which election cycles? How many previous election cycles were used?
Second, are previous election cycles likely to predict turnout among specific demographics in this election? I would argue that the turnout in this midterm election will be quite different from the turnout in 08′.
1) We looked at 2004 to 2006 drop off and the make up of registered voters and the actual electorate that turned out in those elections. We also take into account the changes in registration between 2006 and 2008 and how that changed the electorate in 2008 and moving forward.
2) You are absolutely right. 2008 and 2010 will be very different. Two reasons for this – a) its always been that way! Pundits are acting as though it was reasonable to expect the same folks who turned out in 2008 to turn out again in 2010. That has never happened. What always happens in midterm elections is depressed young voter turnout (others can argue about why that is) and if you look at who registered to vote as a Democrat between 2006 and 2008, they were disproportionately younger voters. Therefore, young and Democratic drop off was always a given. However, Republican enthusiasm is undoubtedly high this year so that again throws another wrench in the gears.
Basically, we have to look at the past to predict the future but we also need to be reasonable in terms of what to expect. Under 35 voters will not be 23% of the electorate again like in 2008.
I really appreciate you taking the time to come on here.
I’m sure you’ll get a bunch of tough questions from us – in fact, I see some already! 🙂
Aggie already asked the mandatory “weighting” question, so I’ll add these to the list:
I see these are self-identified “likely voters”. Do most polls use this likely voter screen, or do they rely more on past voting history?
Did this survey include cell phones?
If no, did you weight your results to accommodate the ever-increasing number of cell phone users and their apparent disparities in voting preferences?
I’ll try not to dork out too much here but there are two types of sampling used by the pollsters you see releasing public polling.
RBS (Registration list based sample) – A sample of voters taken from a list of registered voters.
RDD (Random digit dial) – A sample of households based on the random generation of phone numbers within a given geography. (Its a bit more complex that than but I hope this will suffice.)
We use RBS and therefore we are able to pull samples based upon past voting history (as does PPP and many of the other private polling firms here in Colorado). For this survey, we drew our sample based on 2008 because it was the high water mark in terms of turnout so we pull those voters and anyone who registered after that election and then asked those voters if they planned on voting. If it was not at least “probable” that you would vote, we terminated the interview.
This is a pretty standard way of doing things but different pollsters use different screens.
On the cell phones question, this is not quite as straight forward as I would like. Yes we did. But we did not do a “cell phone sample.” Some of the phone numbers in our sample were cell phones because voters put down their cell phone as their contact number when they registered to vote. This is different from doing the standard methodology of calling cell phones (an RDD sample of cell phones which are then screened down to only those who are cell phone only). Therefore, we did call some cell phones. However, we did not quantify it.
Did that answer the question?
Sorry – this is what you get for coming to a politically geeky blog.. 🙂
Doesn’t this ignore people who were registered but didn’t vote in 2008 and will vote this election? There were a lot of conservatives who just didn’t vote based on the choice between McCain and Obama.
if people were registered in 2008, but did not vote, they weren’t sampled.
Given the usual fall-off between presidential election years and midterm elections that seems like a reasonable enough methodology: if you didn’t vote in 2008 why assume you’ll vote in 2010?
There may in fact have been conservatives who chose to sit out 2008, although I personally doubt it.
If you want to infer that this poll undercounts Buck supporters go for it! But it’s only an inference and no way to really know is there, BJ?
In an case, I’d say there’s plenty in this poll to give ammunition (or pause) to both the Buck and Bennet camps – and light a fire under their butts for the next 6 days.
So I’m just going to say State Line’s response is my response.
Adios y’all, see you next Tues night or Weds.
Off to WVA to work the Manchin campaign’s closing days.
Looking for 51-52 Dems in the new Senate – and “Speaker Boehner” (ouch!) too.
Biased, just like I said.
I didn’t read anything about only sampling people who voted for President. Or that that would even be possible. I’m pretty sure “active” is “active.”
Why would you have been so upset by McCain that you sat out all other races, but without benefit of a Presidential race will show up for nearly the same races? I would assume that if no ballot was cast purely out of McCain disgust, none would ever be outside of a Presidential year.
Much in the way some Republicans will not vote Maes or break ranks. That doesn’t mean they aren’t voting, just not for Governor. Lamborn, Coffman, not Markey, etc. all needed votes in 2008. Same as this year.
But of course, all of this and SL’s is related to an honest question. Not an “Ah ha!” bs post.
That year we didn’t. The conservatives that threw up their hands in disgust in 2008 will vote for tea party candidates this year.
All liberals to you?
who if you recall was at the top of the ticket that year.
2) Contested congressional races in a majority of the state’s 7 CDs, including several containing the lion’s share of the state’s conservatives: Markey vs Musgrave; Perlmutter’s & Salazar’s 1st re-election bids in very split districts; Coffman’s bid in an open seat; and Lamborn wasn’t that contested but it was his 1st re-election bid too. I suppose conservatives residing in the 1st & 2nd CDs could’ve rationally sat out those races – but again, why would they sit out the Senate race?
So I’m highly doubting many conservatives SO committed to the cause that they abandoned McCain would’ve sat out contested Senate and House races.
1) Don’t Independents slightly outnumber registered Ds and Rs in Colorado?
– If they are underrepresented in RBI’s sample;
– And IF they vote in #s roughly proportional to their % of the overall electorate;
– And if they favor Buck over Bennet;
wouldn’t that mean Buck’s real % is higher than your aggregate of 42%?
Appreciate your response and thanks again!
Not the place for me to post previous election stats, but not all unaffiliateds are “independent.” Some are unaffiliated simply because they don’t care. As a group, they historically turn out in lower numbers than voters who are affiliated with a party.
Most recent registration figures for active voters:
Dem – 799,981
Rep – 862,575
Unaff – 757,935
And even though someone might be registered as unaffiliated, a good portion of them will self-identify as one or the other party. Unaffiliated voters also vote at substantially lower rates than partisans in mid-term elections.
turnout in proportion to their registration, compared to affiliated voters.
The “active” thing is sort of a problem.
You’re “active” if you vote.
I like to look at turnout as a proportion of all voters, as opposed to active voters. An inactive voter can activate in any election just by voting.
the number AFTER the ‘do you lean to D or R’ question is asked – thus: ‘hard’ independents.
1st) Polsters take heed: party registration and party self identification are not the same thing. Think of it this way – party registration is a demographic which we can match objectively to how you are registered with the SOS. Party self ID is an attitude which changes often.
Allow me to demonstrate. Of those in our sample who identified as “just Independent” (i.e. they don’t lean towards either party), the party REGISTRATION breaks out as follows:
D – 13%
R – 20%
U – 66%
So about 2/3rd of those Independents we keep hearing about are actually registered with a party.
2nd) Lets look at what happened in 2006. In November of 2006, the breakdown of voters who had voted in the 2004 election or registered since was as follows:
D – 31%
R – 36%
U – 34%
Turnout in 2006 (the people who actually VOTED) was as follows:
D – 33% (+2 from registration)
R – 40% (+4 from registration)
U – 26% (-8 from registration)
(^^^theres some round error here)
Unaffilateds are much less likely to vote in a midterm while partisans are much more likely.
Apples to apples, the current make up of the voted 2008 or registered since universe is:
D – 33%
R – 35%
U – 32%
We tried to weight similar to how things turned out in 2006.
Does that mean RBI is comfortable that the 22% of ‘Independents’ in your sample – the ones who prefer Bennet over Buck 41%-34% (noting this still leaves ~25% ‘undecided’ at this late date) – accurately reflects likely I turnout?
For the reasons stated above.
But we predict Unaffiliated turnout to be 24% Notice the very last question in the toplines. This is taken from the voter file. We did not ask the person on the phone which party they were registered with.
The point of this whole exercise is to answer methodological questions, weighting, polling practices, etc.
1) To clarify, RBI used the term ‘Independent’ in their phone script, the reason I used it in my question.
2) If R/G’s registration #s are correct, unaffiliated voters are something like 31% of the electorate, not 22%.
3) If as stated in RBI’s Summary, Republicans don’t need to get a majority of unafiliated voters in order to win (b/c R turnout is expected to exceed D turnout), but Buck has a comfortable lead among unaffiliateds anyway……..how does Michael Bennet win?
Still curious about ‘Independents’ being only 22% of the N though…
But let’s see what Kevin says.
Look at the early voting numbers.
About Friday’s poll – just wondering if you surveyed Proposition 102? I have seen polling on the other measures but nothing about it. Thanks!
We did not.
Notice RBI states in the Summary that Bennet’s 1-point lead is due entirely to his 38%-31% advantage among ‘likely/probable’ voters (as opposed to ‘already voted’ or ‘will definitely vote’ respondents).
These of course are the least likely voters in your sample – and without them Bennet goes down.
As a Bennet supporter this makes me ‘twitchy’. And gets back to Voyageur’s emphasis on turnout.
Any comment Kevin?
Was just that: turnout is going to determine how this election goes.
And its not necessarily that Bennet goes down so much as a 1 point lead goes down a completely dead even race. That should make any Bennet supporter twitchy.
All it really means is since he couldn’t even put up a real lead in this obviously fake poll, he’s toast. This proves that Buck is leading and will win by 5% +.
Watch, tomorrow they’ll say Hick is leading- when we all know TT is going to pull out the surprise win.
If I didn’t already have the Tancredo one going I’d do it. But only if we could find someone to hold the money in escrow.
You mean you don’t have the money to pay another bet. No one turns down a sure thing.
But no one’s saying it isn’t a close race.
a great use of the blog!
1-there is obvious momentum for Bennet, which even shows in pro Republican polls like Rassy.
2-whether rassy, a 2-point buck lead, The Denver Can’t Be Named, dead even 47-47 or rbi,
1 point buck lead, these are all too close to call.
All within margin of error.
This race will be won at the doorbells and the phones. I’ve been making calls for Bennet and helping turn them out.
Keep up the good work, volunteers. Sometimes, the good folks win!
– Bennet had – or at least ‘had’ (since polls are only snapshots of the electorate at a particular moment in time) the momentum over the past 7-10 days.
Question is: Is it enough to merely pull him to breakeven – or can it drive him over the top?
Guess we’ll know in <7days’ time eh? (Fingers crossed.)
Few if any minds will change in the next few days. It’s a matter of getting our folks out to vote. Victory goes to the side that works the hardest. that’s rare in politics — but it’s just as it should be!
I’m not sure that’s right, V. Just yesterday, JO declared the church-state video clip of Ken Buck was what sealed the deal for him. And, anecdotally, I’ve talked with a LOT of voters who have a “pox on both their houses” view and are looking for a reason to vote for one or the other candidate. I really get the sense this is still up for grabs, and it isn’t just making sure supporters vote.
won’t vote at all. I mean, if they haven’t been moved to commit by the barrage of earned and paid media by now….
All the more important than, as Voyageur notes, to get your known supporters to actually cast a ballot if they haven’t done so already.
But they seem to be few and far between.
The Bennet campaign was very understanding with my limitations and my desire to help the campaign. They set it up so that I can make calls from my house. I have been making quite a few every single day and most of the people I’ve talked to have already made up their minds. Every once in awhile, I come across someone who wants to be convinced. This is when Buck’s social issues stances plus Bennet’s voting record come into play.
The conservatives on this site want everyone to believe that Colorado voters are upset about Bennet’s voting record. They aren’t. I’ve used his votes to convince many people that he has been fighting for CO many times.
How much is being spent (per vote) to change minds?
What’s the difference on paid advertising between mobilizing your base and swinging undecideds? I will say the GOTV effort is entirely devoted to turning out. If someone is genuinely undecided, of course we will talk to them — briefly. But we NEVER try to argue with them. the old rule applies:
With the only poll that really matters coming up so soon we may as well enjoy the opportunity to feed our poll addiction without arguing any more about which polls are the most accurate. We’ll all know soon enough. That goes double for H-man. But, polls aside, I hope all the Romanoff folks will get over it and call or knock on doors for Bennet. After all Bennet will vote exactly the same way Romanoff would in almost every case. Buck never ever will. At the very least you can help your county party out in your HD with GOTV that supports all the Dem candidates.
Craig Hughes works for RBI? Isn’t RBI essentially running Bennet’s Campaign?
so my assumption is these data are accurate.
(Plus, they track other polls out in the past 3-4 days.)
I don’t think I have seen him in our office since mid 2009.
Our official consultant role with the Bennet campaign ended June 1 and we have not had a formal role as a consultant since.
So no, we are not running Bennet’s campaign.
No bias there, just because your “official” role ended June 1. I think it’s obvious that you’re still probably on their side.
if you don’t want to trust the results, that’s fine. you can obviously do so.
However, the whole reason to come on here is so that people can ask questions about how we conducted the poll and I can answer them. It benefits us in absolutely no way to put out inaccurate polling.
Plus, we tried to be as open as possible about every aspect from where we bought our sample to the frim we used to do the interviews (a firm owned by a Republican, btw).
If you try to use logic with Beej your head will hurt worse in the morning than a night on the town with cheap tequila. Do what we do… make fun of him and then move on.
Thanks for the great info though, this has been very insightful
You need to fire up the base. Good job; mission accomplished. The other two polls that came out today say you are wrong though; Buck is winning.
And the world was created in six days. And gays are pansys. And Dan Maes, err, now Tom Tancredo, is a sure thing for governor.
And rape victims are just suffering from buyers remorse. And stomping on the heads of helpless women is okay if you’re a tea partier.
Have I forgotten anything from the Douchebag’s
to allay any suspicions but of course Davie Chest-nutz sees a conspiracy.
It would be nice for Beej 2 stfu and just read but your not capable or respectful discourse.
Why can’t ColPols institute an IGNORE function?
It’s called CoDA.
You may have read it (MAYBE), but you sure as fuck didn’t comprehend it. That’s why you’ll never learn.
So go on, support an idiot with only talking points and pray he wins. Then watch the disintegration again. Of course, who wants to lead when you can have a big soap box anyway?
to join the beej boycott. Occasional answers to his inane posts are fine, but he really needs to go without a playmate in the sandbox for a while.
You do nothing but comment on my posts.
And I spend my days plotting ways to undo the damage wreaked upon this state by wealthy special interests who have bought it off. 🙂
A math professor is lecturing to his students on some aspect of a theorem. He writes down a statement and says, “Now this is completely obvious.” Then he pauses, and says to himself, “Wait, is it obvious?”
He leaves the room to do some furious computations, and half an hour later he comes back into the class. “Yes, OK, it’s totally obvious. Here’s why…”
Kevin: What is correlation between those persons that self-select as probable voters vs. those person that actually vote at the end of the day? In particular, in your experience, does this poll’s number/percentage of “probable voters” seem normal, above or below average?
We used the exact same methodology for 25 years and have found this to be highly accurate. In fact, the “probably” vote number always goes down as you get closer to the election. We were at like 8%-9% “probably” vote back in July. Now its down to 2%.
about how polling firms are dealing with the high # of people who ONLY have cell phones – and what generalizations if any one can infer about such individuals’ political preferences – that would be great!
For a few reasons. First of all, its not cheap and calling cellphones is outside of many campaign/organizations budgets. A number of reasons factor into this: it uses minutes to talk on a cell phone and people refuse to take the interview on the cell more often than land lines. also, call backs are much more common because people are often out and about and not able to take the interview right there. Further, teens use cells and don’t qualify for the survey. Finally, and probably most importantly, it is illegal to call a cell phone using a predictive dialer (most pollsters now a days use this technology). So that means that interviewers have to hand dial the number. That increases cost substantially.
Second of all, cell phone users are different politically that land line only users. In fact, Pew Research just released a study on this showing the Republican bias to be as much 4-6% (which can’t be accounted for by weighting). This is a problem and some pollsters are dealing with it by just making it a standard part of their methodology. Many media pollsters now do this. However, if you are a candidate for, say, state legislature, your budget can’t match a big media firm and therefore you can’t afford cell phone dialing.
VERY interesting about the +4-6% R bias among cell phone users from Pew.
I would have figured it to be exactly the opposite: a bias toward Dems.
Interesting implications, not especially good ones either….
He said the Republican bias is as much as 4-6%, but he didn’t say in which sample. I assumed he meant the Republican bias among landline-only users, but that’s just because I figure most folks who only use landlines are older, and older folks tend to lean red. Kevin, if you check in again, can you clarify?
If you are only surveying land lines, according to Pew Research, the result will be 4-6% biased in favor of the Republican.
I’d also expect that number to decline in future cycles, and to vary 1-4% in some localities.
On the cell phones question, this is not quite as straight forward as I would like. Yes we did (call cells). But we did not do a “cell phone sample.” Some of the phone numbers in our sample were cell phones because voters put down their cell phone as their contact number when they registered to vote. This is different from doing the standard methodology of calling cell phones (an RDD sample of cell phones which are then screened down to only those who are cell phone only). Therefore, we did call some cell phones. However, we did not quantify it.
I understood your poll assumes turnout for independents to be 22%. What do you assume the breakdown to be between Republicans and Dems?
About the difference between party registration and party self ID so I’d first suggest reading that.
We are assuming a 4 point REGISTERED Republican edge based on past elections.
D – 36%
R – 40%
U – 24%
Some pollsters are using smaller and larger margins but the one thing everyone agrees on is that Rs will have the edge in turn out. I’d point you to the summary of findings we released. It talks a bit about this.
Don’t people just lie to pollsters?
but also people from both parties ticket split all the time or vote with the other party’s candidate all the time. Heck I know a couple “republicans” that haven’t voted for a republican in their life, except for the primary, but they are registered as such and will tell some people that they are. They obviously love the secret ballot.
Voted for Obama. Your registered party doesn’t require that you vote that way. However, it is very highly predictive of how you will vote. In this case, less than 1 in 10 registered Republicans are voting the the Democratic candidate. Its even lower among registered Democrats voting for Buck.
So if a pollster who formerly worked for Democrats says the numbers will be 40-36, then because of the WAVE™ we can just add 10 to the Republicans and get
which means Buck will win by 102%. Math!
but when you are screening voters and one indicates they are likely to vote, yet they haven’t in the past, does that drop them in the rankings from likely to possible? And vice versa, if someone indicates they are only possibly going to vote, but have voted in the last several elections, do you bump them up in the probability rankings?
Likelihood to vote is asked and reported in the toplines and crosstabs exactly as the respondent reported it to us. If you have no vote history, that means you registered sometime AFTER the 2008 election. Our sample only includes people who voted in 2008 or registered after that point. In other words, they had to show some propensity to vote if we called them.
So, even if they have no vote history, if they said they are certain to vote, then we put them as certain to vote in the crosstabs and toplines.
I think I got it. Basically if someone registers in, say 2005, but haven’t voted, they aren’t on the radar.
that of the total who have voted so far, 58% are Female and 42% are male.
55% of Females lean toward Bennet and think Colo. is on the right track.
Males are at 45% for both.
If this trend contnues, then Bennet should win.
As per the poll summary, men favor Buck 45% to 41% while women favor Bennet 44% to 39%.
Same argument, same conclusion. Women are outvoting men 58% to 42% in votes cast so far. Positive trend for Bennet.
(Women also lead men in the “absolutely certain” and “very likely” to vote categories).
So far, 40,000 more Pubs have voted than Dems, yet the race is still tied.
First, I could be reading the survey wrong, but I noticed that 38% of Bennet supporters are “less than certain” to vote, while only 31% of Buck supporters are “less than certain” to vote. Is this indicative of democrats staying home this year, as we’ve been hearing? Or to rephrase, more democrats staying home than republicans staying home?
Second, I noticed that your likely voter age breakdown is 31% 18-44. In the obama wave of 2008, that number was 47%. But you also mention that your sample has been weighted by age to reflect turnout in previous years. My question is, does your survey suggest the “young Obama voters” are staying home this year? If so, can you quantify how much they aren’t turning out? It’s hard to tell from the survey because of the weighting.
1) Yes and no. Hypothetically, the people in this sample are going to vote. At least that’s the operating assumption with this likely voter survey. So I don’t think it can be indicative of more people staying home. But, the fact that less Bennet voters are certain to vote is probably indicative of “the enthusiasm gap(R)” which we all keep hearing so much about. Republicans are REALLY fired up to vote and Democrats are much less so. So this can probably be a good proxy for enthusiasm.
2) I’m not sure where you are getting your figures from on under 45 turnout. Our voter file says 42% were under 45. But regardless, yes. Young voters are not going to turnout like they did in 2008. But that’s not necessarily indicative of anything because that happens EVERY midterm. Pasting from a comment above:
“2008 and 2010 will be very different. Two reasons for this – a) its always been that way! Pundits are acting as though it was reasonable to expect the same folks who turned out in 2008 to turn out again in 2010. That has never happened. What always happens in midterm elections is depressed young voter turnout (others can argue about why that is) and if you look at who registered to vote as a Democrat between 2006 and 2008, they were disproportionately younger voters. Therefore, young and Democratic drop off was always a given.”
and I got my 47% number from exit polling. I found it here: http://www.cnn.com/ELECTION/20…
But now that we have info from about 1/4 of the people that are voting and the split is closer to 6, would it not be a better predictor to use that information instead of your pre election estimate?
But past election history shows that Democrats manage to close that gap a bit as it gets closer to election day. Plus, our sample shows that lots of Democrats are sitting on their ballots despite already voting (based on comparing our sample to actual returns).
It’s a very legitimate question.
I’m back tomorrow from 1pm – 3pm.
Thanks for the great questions. I hope I was able to answer them all.
If not, we can try again tomorrow.
among likely voters six days before election day, particularly relative to the 1 percentage point gap and the MOE.
It is hard to believe that there aren’t some leans in there.
Unless they are exactly evenly split or don’t vote, this seems to deprive the poll of much predictive power.
split for these voters, however.
Kevin/Pols: it’s my guess that this is the first time many of the posters have seen full polling data including summary, crosstabs and toplines. It raises the level of awareness here, which is helpful as coloradopols has become the go-to blog for Colorado politics. In particular, one can better understand what difficult decisions campaigns have in interpreting this data. It puts each poster in the position of asking, “so what does the campaign do next?”
You’ve shown why RBI is a go-to pollster. I hope this trend continues, but we can all say RBI was the first.
You should get a medal or something!
I have seen lots of polling data over the last 20+ years, including RBI’s data, and RBI does a good job.
That’s why they get hired.
If you bothered to read the thread, RBI is not on anyone’s payroll.
but it’s hard to tell sarcasm on the internet.
My comment was not meant as faint praise. It was and is sincere. I can certainly understand why, when reading comments on a blog how true praise could be seen as snark, tongue-in-cheek, or feigning — it’s just not the case here.
Maybe I didn’t do a good job of getting my point across. I’ll try to re-state it.
On this blog, there used to be real political professionals that had great gossip and insider knowledge. The blog got popular, there were some “outings,” and it became less easy for real campaigners and political-pro’s to post. In their place, came a bunch of people who care about politics, but really have no knowledge of campaign work (there are, of course, major exceptions). The makeup of Coloradopols posters changed, but it is still the go-to place for those interested in politics, including political reporters and campaign hacks.
In this new landscape, I’ve come to the conclusion that very few posters on here have actually read a poll, like a campaign would. This conclusion is based upon voluminous comments about poll methodology, bias, etc., that no person that has been around polls would normally spout.
With this context, I think it is absolutely fantastic (truly, no sarcasm here) that RBI provided this exclusive poll and that it provided access to one of their pollsters in the waning days of campaign season. I really hope some of the posters here learned something.
As for RBI, it’s my understanding that campaigns on both sides of the aisle know and respect its work. They have a long history of hitting the mark and providing reliable data. That history includes work here in Colorado, nationally and internationally. Among the political professional class, that has been driven away from posting here, RBI is beyond reproach.
To that end, it surely cost RBI time, money and resources to do this poll and I wonder why they would do this. I don’t know for certain, but as I think about them and the rise and fall of print media, I laud them for trying something new and embracing the next wave of information sharing. So, while RBI is experimenting with this format, I hope this isn’t their only foray, we’re all better for having this information in the public sphere.
Hickenlooper by 8 points.
Rationale: the neck-and-neck polling in the Governor’s race seems to have been done by pollsters who also have Buck with a comfortable lead against Bennet.
I don’t see any Bennet-Tancredo ballot splits taking place. But I could see some Buck-Hickenlooper splits.
Still too close for comfort. The Democrats still need to get the word out about how unethical, incompetent and dangerous Tancredo is.
I’m being a little more conservative for Hickenlooper than fivethirtyeight.com, which has Hickenlooper with a composite 49.9%, Tancredo with 40.5%, and Maes with an astoundingly low 8.9%.
This may very well end up being the Total Disaster scenario for the Republicans: Tancredo can’t beat Hickenlooper, but he pulls Maes under 10%, thus screwing the GOP royally.
Can’t be trusted.
Tanc wins in a landslide. Mexicans leave Colorado in droves.
and Colorado will be renamed “Grand.”
gets the most widely divergent poll results? Are pollsters having a hard time figuring out how to weight it? From what I’ve seen out canvassing, I’m inclined to agree with polls that show a decisive Hick victory (though not by 14) than those that show Tanc within the margin of error.