Latest Post | Last 10 Posts | Archives
Previous Post: SUBSCRIBERS ONLY - Today’s edition of Capitol Fax (use all CAPS in password)
Next Post: Rate Londrigan’s new TV ad
Posted in:
* The margin of error in both polls is 9 percentage points, however…
If sampling error were the only type of error in a poll, we would expect candidates who trail by one point in a poll of 512 people [533 in the Bost poll] to win about two out of every five races. But this probably understates the total error by a factor of two.
…Adding… I think this is true. I thought that claimed MoE was too high for polls of 500+ people. From comments…
More precisely, the MOE is +/- 4.7 for IL-6 and +/- 4.6 for IL-12. That actually is about right for a sample of ~500 voters.
* On to the polls. Republican US Rep. Peter Roskam leads Democrat Sean Casten 45-44 with 11 percent undecided. Republican US Rep. Mike Bost leads Democrat Brendan Kelly 44-43 with 13 percent undecided. Two-thirds of respondents in both surveys were reached on their mobile phones.
In Roskam’s suburban district, President Trump’s job approval rating is just 36 percent while 57 percent disapprove. The president is slightly above water in Bost’s rural/blue collar district 48-46.
51 percent in the Roskam poll said they preferred Democrats to take control of the US House, compared to 44 percent who wanted the GOP to stay in power. Those numbers went the other way in Bost’s district, with 46 picking the GOP and 43 choosing the Democrats.
* The Roskam poll is here. The Bost poll is here. Tell us what you think.
posted by Rich Miller
Friday, Sep 7, 18 @ 9:17 am
Sorry, comments are closed at this time.
Previous Post: SUBSCRIBERS ONLY - Today’s edition of Capitol Fax (use all CAPS in password)
Next Post: Rate Londrigan’s new TV ad
WordPress Mobile Edition available at alexking.org.
powered by WordPress.
This poll seems useless. I mean, I love polls. But a 9 or 10 point margin of error does us no good in races this tight.
I’m more interested in the incredibly low number of people willing to take the poll. Troubling — and response levels that low start to cast doubt on the poll’s representation of the electorate, IMHO.
Comment by Soccermom Friday, Sep 7, 18 @ 9:25 am
Not good news for either incumbent in a change year as undecideds won’t be breaking for more of the same. Kelly’s done everything right in the tougher of the two districts. Casten’s campaign hasn’t been perfect but he’s in more favorable territory. Narrow win for both challengers?
Comment by Unreliable Sources Friday, Sep 7, 18 @ 9:26 am
It was an interesting exercise and I am surprised the results are so close. Still, it’s a wee sample and the y got too many undecideds for me to think this will remain an accurate representation of the districts even next month and certainly not in November.
Comment by Cheryl44 Friday, Sep 7, 18 @ 9:27 am
We knew that, given what is expected to be a good year for Democrats, these races would be close. Now we have data that confirm what we already knew. The advantage remains with the incumbents but these will both be close races.
Comment by 47th Ward Friday, Sep 7, 18 @ 9:27 am
+ or - 9 points. That was the most fun useless poll ever.
Comment by Ducky LaMoore Friday, Sep 7, 18 @ 9:29 am
A 9 point MOE is quite a bit to swallow. But in the interest of discussion, the Trump approvals and MOE in IL-06 give Casten a lot of room to grow; the same in IL-12 give Kelly not so much.
Comment by Arsenal Friday, Sep 7, 18 @ 9:30 am
It confirms what we already knew, previously safe Republicans are locked in tight races and vulnerable in a way that they haven’t been since they took office.
Comment by slow down Friday, Sep 7, 18 @ 9:31 am
Surprised amid the current environment, with a deeply unpopular president and governor as the faces of the party, in a deep blue state, the Democratic challengers aren’t significantly ahead.
Comment by JB13 Friday, Sep 7, 18 @ 9:32 am
@Soccermom is absolutely right. The margin of error too is high and the polling numbers are too low.
This is just a feeling but it seems like people are messing with the pollsters more and more.
Comment by JS Mill Friday, Sep 7, 18 @ 9:36 am
Although the methodology may be suspect, this basically confirms and reflects other polling that shows that these are 2 very competitive races. And we are still 2 months out.
Comment by illini Friday, Sep 7, 18 @ 9:37 am
This poll is not very precise given the MOE, so we can’t draw too much from it. Still, for someone like Roskam to be only 1 point ahead this early in the race (many people only are thinking about Congressional races if/when they see TV ads) does not bode well for him.
Comment by Techie Friday, Sep 7, 18 @ 9:38 am
After Trump’s win, I’ve come to realize that polls are totally meaningless. They have no meaning. They’re just a snapshot of the polling instant where the answer given has an equal chance of being true or false.
All that matters — and all that’s real — is the moment when you stand in the voting booth and check the box. That’s it.
*shrug*
Comment by Macbeth Friday, Sep 7, 18 @ 9:39 am
Today a pollster’s chance of actually connecting with someone who wants to talk aren’t good. And I would have some suspicion regarding those that actually do express a willingness to express an opinion. I suppose the margin of error reflects that to an extent but it leaves me wondering if polls of this type have outlived their usefulness.
Comment by Pundent Friday, Sep 7, 18 @ 9:39 am
The preconceived narrative by the NY Times and others is that there will be a blue wave and these races are tight.
A poll with a 9% margin of error does not confirm or deny that premise.
More interesting is if the poll sample of over 36,000 calls were likely voters or just the general population
.
I guess I am not the only one who does not answer calls from unfamiliar numbers. With a 1.4% answer rate, polling seems like a tough business.
Comment by Lucky Pierre Friday, Sep 7, 18 @ 9:41 am
Yeah, plenty to pick on here regarding the polling science, but does anyone doubt these races are close?
Comment by Roman Friday, Sep 7, 18 @ 9:42 am
==Surprised amid the current environment, with a deeply unpopular president and governor as the faces of the party, in a deep blue state, the Democratic challengers aren’t significantly ahead.==
These are both strong Republican districts. IL-06 was designed to be a Republican vote sink, and IL-12 swung hard to Trump. These are only competitive at all because of the environment.
Comment by Arsenal Friday, Sep 7, 18 @ 9:47 am
Three big trends happening to reduce polling accuracy:
1. Few landlines left anymore
2. People with cellphones not answering unknown numbers, or immediately ending the call when it is a computerized caller.
3. Hyper political environment, makes it more sensitive for voters to express their opinion to a stranger.
Comment by Downstate Friday, Sep 7, 18 @ 9:47 am
Roman,
I don’t think Bost’s will be close at all. I think he’ll win by 6% or more. As I’ve said before, the judicial races in his district are a “tell” on the strength of one party.
Republicans won the appellate seats handily, even though they were outspent by more than 5-1.
Comment by Downstate Friday, Sep 7, 18 @ 9:50 am
It’s hard and expensive to poll Cds and even more so to poll state districts that is why you see so few. It’s nice to have some data at this level.Mist leaks from internal poll to people like Cook.
Comment by Not a Billionaire Friday, Sep 7, 18 @ 10:08 am
The poll doesn’t bother me. The +/- 9 is terribly troubling.
The side that takes the most from this poll is probably the side wishing more than knowing.
A +/- 9 isn’t know all that much.
Comment by Oswego Willy Friday, Sep 7, 18 @ 10:12 am
I think you looked at the poll wrong. The MOE is actually 4.7 for the Roskam and 4.6 for the Bost one.
Comment by Anonymous Friday, Sep 7, 18 @ 10:21 am
More precisely, the MOE is +/- 4.7 for IL-6 and +/- 4.6 for IL-12. That actually is about right for a sample of ~500 voters. I suppose I can understand why somebody would say that’s a 9-point MOE, but to call it +/- 9, as was done at least twice in this thread, is wrong.
Comment by Cromulent M. Biggens Friday, Sep 7, 18 @ 10:33 am
==I think you looked at the poll wrong. The MOE is actually 4.7 for the Roskam and 4.6 for the Bost one.==
Uh, This is the explanation on the NYT poll itself: “each candidate’s total could easily be five points different if we polled everyone in the district. And having a small sample is only one possible source of error.”
So if each candidate’s result *could* be off in either direction by as much as 4.5% that’s how you could get to a possible total of a whopping 9% MOE.
Comment by Responsa Friday, Sep 7, 18 @ 10:41 am
Lucky Pierre: Only you would figure out a way to bring the NYTimes into this. Made me laugh. Thanks.
Comment by walker Friday, Sep 7, 18 @ 10:42 am
===but to call it +/- 9, as was done at least twice in this thread, is wrong.===
I’ll be more precise and less lazy in my thought.
There is a 9 point swing possible from one side… to another side.
The nine point swing isn’t a great snapshot for either side to take any comfort.
Comment by Oswego Willy Friday, Sep 7, 18 @ 10:44 am
It’s actually more than a 9% possible MOE. And plus/minus is the way the NYT had to describe it in order to be honest.
Comment by Responsa Friday, Sep 7, 18 @ 10:44 am
Sorry, didn’t see the update when I typed that.
Also, they’re doing something a little technical but for most polls the margin of error is dependent on the sample size. You can usually calculate a margin of error yourself, enter the sample size on your calculator, hit the square root button and then hit the “1/x” button. Multiply that by 100, that’s your +/- value for your margin of error. Do that for 600 respondents and you’ll get a margin of +/- 4. Some pollsters will add in a bit more technical corrections to the calculation but often times not.
Comment by The Captain Friday, Sep 7, 18 @ 10:50 am
OW, if by “a 9 point swing possible from one side … to another side” you mean either or even both of these candidates might be anywhere between 40 and 50 percent, then you’re right. But, as you note, one shouldn’t infer much beyond that.
And, again, a +/- of 4.7 for a survey of ~500 respondents is not unusual. In fact, it’s close to typical for that sample size.
Comment by Cromulent M. Biggens Friday, Sep 7, 18 @ 10:54 am
===But, as you note, one shouldn’t infer much beyond that.===
Yep.
Comment by Oswego Willy Friday, Sep 7, 18 @ 10:58 am
I know that dumping on polls and pollsters is all the rage and obviously there are some pollsters worthy of scorn (ahem Victory Research). But a narrative has taken hold about 2016 that is just flat out wrong.
In reality, the national polls in 2016 were remarkably accurate. The final Real Clear Politics Average predicted Clinton would win the popular vote by about 3%. She won it by a bit more than 2%. What was problematic were the quality of some of the state polls. But even there, in the final months Hillary was never up by more than single digits.
Closer to home, many of the internal pollsters nailed the governor’s race despite the public narrative of a super-tight race.
Finally, a MOE of 4.7 on a poll like this is pretty standard. What we can learn from these polls is simple. These races are absolutely a tossup that we won’t know the outcome of until election day.
Comment by Chicago Cynic Friday, Sep 7, 18 @ 11:21 am
If you read through the material about sampling error, there is a link to a study done with academic statisticians about polling results versus election results (that’s where the ‘MoE’=9 confusion is coming from):
“In a new paper with Andrew Gelman and Houshmand Shirani-Mehr, we examined 4,221 late-campaign polls — every public poll we could find — for 608 state-level presidential, Senate and governor’s races between 1998 and 2014. Comparing those polls’ results with actual electoral results, we find the historical margin of error is plus or minus six to seven percentage points. (Yes, that’s an error range of 12 to 14 points, not the typically reported 6 or 7.)”
The point being that sampling error is not 9, but if this race/s end up matching study results we could be in the 9 range for error because of sampling, non-response, etc. Think of this as something like total polling error.
Comment by Precinct Captain Friday, Sep 7, 18 @ 11:26 am
That 11% undecided is more red than blue. From the poll..
“About 11 percent of voters said that they were undecided or refused to tell us whom they would vote for. On questions about issues, these voters most closely resembled Republicans.”
Comment by Texas Red Friday, Sep 7, 18 @ 11:34 am
==What we can learn from these polls is simple. These races are absolutely a tossup that we won’t know the outcome of until election day.==
Well then that wasn’t a very good use of polling time and money was it?
Comment by Responsa Friday, Sep 7, 18 @ 11:44 am
I reside in Williamson County, in IL12. Couple very interesting items from the poll. One, the MOE is not 9% but that’s been discussed. Trump approval took a huge hit in the district. Down to 48%, that’s a big deal. 52% of the district supports a ban on assault weapons and high capacity magazines. Again, huge. This is deep south Illinois remember. The immigration and opioid questions appear to be non-issues for southern voters, which is odd considering the district is very rural. I’d say hitting Bost on his failure to standup for working people and families is one the strongest issue Kelly can go on.
Comment by Southern_Dawg Friday, Sep 7, 18 @ 11:55 am
It seems to me that left leaning voters are eager to participate in polls and right leaning, not so much.
Although the left is energized, I believe less than 4% of youthful voters voted in the last IL primary. It’s nice to protest but you have to vote.
When all is said and done, there may be far stronger Republican support than projected.
Comment by Adjusted Friday, Sep 7, 18 @ 11:55 am
. . . The Roskam poll is here. The Bost poll is here. Tell us what you think. . . .
I think Yogi Berra was a wise man:
* It ain’t over ’til it’s over.
and
* That’s why they play nine.
Comment by Hamlet's Ghost Friday, Sep 7, 18 @ 12:10 pm
Very cool by the Times, very transparent, very well done.
Comment by Molly Maguire Friday, Sep 7, 18 @ 1:00 pm
These polls are useful as a baseline. Am curious to see how many times NY Times will repeat the exercise and how noisy the results are over the next few weeks.
Comment by Anonymous Friday, Sep 7, 18 @ 1:28 pm
==It seems to me that left leaning voters are eager to participate in polls and right leaning, not so much.==
If the Republicans can’t be arsed to answer the phone, going to vote is going to be a real burden for them.
Comment by Arsenal Friday, Sep 7, 18 @ 1:52 pm
The challengers are good, but the economy is better. Incumbents do well with a booming recovery, regarless of party.
Comment by VanillaMan Friday, Sep 7, 18 @ 2:50 pm
no comment other than I had a lot of fun watching this real-time polling come in on the NYTimes site. And it seemed like each one I was watching showed voters divided about evenly. Ready for election day to get here and just find out the rest of the story.
Comment by it'smyopinion Friday, Sep 7, 18 @ 3:49 pm
When will Roskam dust off his mentor - Denny Hastert?
https://roskam.house.gov/media-center/press-releases/roskam-receives-historic-desk-from-former-speaker-hastert
Comment by Duopoly Friday, Sep 7, 18 @ 4:25 pm