Confirmational Bias versus the latest Mommouth Poll

[bluewater]

So, a poll that showed Biden getting only 19% support is an "irrational finding", as one poster called it, or an outlier to be discounted?

Or are these responses examples of confirmational bias, the tendency to interpret new evidence to conform to one's existing beliefs or theories?

I think the latter. People's expectation that Biden must continue to have at least 30% support is so strong, that they ignore any evidence to the contrary.

Is this Monmouth poll really THAT much of an outlier, having Biden at 19% support? No, not really -- unless one chooses to ignore several other recent polls showing Biden's support in the low 20% range.

Consider these recent polls showing Biden's support:

Aug 24-27, 2019 B rated YouGov 570 LV 24%

Aug 17-20, 2019 B rated YouGov 559 LV 22%

Aug 10-13, 2019 B rated YouGov 592 LV 21%

Aug 4-7, 2019 C+ rated HarrisX 1,319 RV 25%

Aug 3-6, 2019 B rated YouGov 573 LV 22%

Aug 1-5, 2019 B+ rated Ipsos 1,023 RV 25%

Aug 1-5, 2019 B+ rated Ipsos 1,258 A 22%

Aug 2-4, 2019 C+ Change Research 1,450 LV 23%

That's EIGHT polls from FOUR different Pollsters just in AUGUST that show Biden's support not that different than the 19% he got in the Monmouth poll.

So are we to ignore 8 polls this month, 9 polls counting Monmouth, because they are ALL outliers? Or are we to think that the polling landscape is not as stable as some people would like to think?

We will have to wait until the actual voting to see how accurate the various polls really are, but we should at least stop saying "that's only one poll! it's an outlier!" when we choose, perhaps subconsciously, to ignore evidence to the contrary.

[OilemFirchen]

1. OMG!

LOL!

[bluewater]

3. lol An excellent example of Confirmational Bias?

Seems so.

[OilemFirchen]

6. OMG!

No.

[Hortensis]

46. :) Maybe it's just too much OP verbiage? Sure you're disappointed,

but past time to cut emotional line and move on?

We're in the poll blizzard season, and it'll be continuing for over a year. When a poll smells and spins like a bad egg, step past and see what comes next? I don't crack and sniff, much less swallow it whole or serve it to my friends. There'll always be others along, but never to become emotionally invested in them because they're, you know, only polls and many months out at that.

[brooklynite]

2. Outlier (noun):

a statistical observation that is markedly different in value from the others of the sample

Monmouth has acknowledged that the poll was an outlier. I think that resolves the question.

As for your polling sample, I know that there isn't an A-rated poll in the group.

[bluewater]

5. They said it "appears to be an outlier."

Perhaps it is. Time will tell.

They also said they chose NOT to re-weight their results just to conform to what other polls were showing, they kept their usual methodology and stood by the results.

But then, it wasn't the ONLY poll that showed Biden with a similarly low level of support in August.

Perhaps those other EIGHT polls from FOUR different pollsters were outliers too. But perhaps not.

[George II]

43. Do those other polls show Warren and BS with similarly high level of support?

[Thekaspervote]

4. Here's what the director at Monmouth u said about their poll

Putting it very nicely, they admit they messed up

https://www.democraticunderground.com/1287257211

The latest HarrisX poll looks very similar to the Suffolk and Quinnipiac polls out today

https://www.democraticunderground.com/1287255855

[empedocles]

7. To the knock down Biden folk - what an autoritative bust

[Thekaspervote]

8. Indeed... go Joe go...we've got your back!!

[TCJ70]

22. Where did they admit they messed up? n/t

[still_one]

40. Been Waiting for this one. The only conformational bias is coming from one poster in this thread

Enough people are becoming wise to what flame baiters try to do,

[George II]

44. The statement, the last two sentences of the first paragraph are key:

[lapucelle]

51. The director of the Monmouth poll admits its "bias confirming" result was an outlier.

The new data, which shows Mr. Biden earning the support of about one in three Democratic voters, stands in contrast to a poll published Monday that showed a virtual three-way tie between Mr. Biden, Ms. Warren and Mr. Sanders at about 20 percent support each. On Wednesday, the director of that poll called those results an outlier.
snip===================================================================

Those numbers contradict a Monmouth University Poll that drew attention from the news media, including The New York Times, earlier in the week. It showed Mr. Biden plummeting 13 percentage points from June, putting him at 19 percent, with Ms. Warren and Mr. Sanders at 20 percent each. On Wednesday, Patrick Murray, the director of that poll, said in a statement that it was clear the results had diverged from the polling consensus.

“It occurs very infrequently, but every pollster who has been in this business a while recognizes that outliers happen,” Mr. Murray said. “This appears to be one of those instances.”

https://nyti.ms/2ZFXfLm

[TwilightZone]

9. Give it up.

Even Monmouth says it's an outlier.

[LincolnRossiter]

10. Whatever helps you sleep at night, my friend.

[bluewater]

12. Eight polls from four different pollsters... that surprised even me.

I knew that several polls showed similar results, having Biden in the low 20% range, but I had no idea there were that many this month alone.

It made for a nice compelling chart.

Notice how respondents in the thread keep the blinders on and don't discuss the other 8 polls? That's telling confirmation that people, perhaps subconsciously, are choosing to ignore evidence that runs contrary to their own strongly held beliefs.

[DrToast]

11. Those polls you listed?

They’re all pretty much in line with previous polls by the same pollsters with respect to Biden.

Pollsters have different methodologies, so it’s a good idea to compare several polls by the same pollster to detect actual signs of people changing their minds.

It’s not just the percentages of the Monmouth poll that seemed unusual. It’s that it was wildly different then previous Monmouth polls. And if the results were because of people changing their minds, you would see similar movements in other polls.

Then poll was an outlier. It happens. No, it shouldn’t be excluded from polling averages because outliers can happen in the other direction too. That’s why polling averages are useful; they can help smooth out the outliers.

In summary, no, the fact that other polls show Biden at similar levels as the Monmouth poll doesn’t mean Monmouth wasn’t an outlier.

[bluewater]

13. How many similar polls does it take before they might not be outliers?

If 9 polls in one month isn't enough, how many are?

Time will tell how accurate any poll really is once voting starts.

But again, we should all stop saying "but that's only one poll! it's an outlier!" when it isn't only one poll and might not be an outlier at all.

[DrToast]

14. Confirmation bias indeed.

Why haven’t those other polls you listed shown similar movement?

Why are you completely ignoring what I said about how it wasn’t only the absolutely levels that suggest Monmouth is an outlier?

You can’t believe both polls reflect the current landscape. The Monmouth poll says, “This race has drastically changed!” The other polls say, “This race is remarkably stable!”

See the problem? It’s not just the levels. It’s the (lack of) movement.

[OilemFirchen]

16. Remarkably,

some weeks ago your interlocutor finally came to understand that the bulk of these polls (YouGov and HarrisX) sample the same individuals for all of their tracking polls... until he/she didn't. That happened, quite coincidentally, when those polls began to show a slight Biden dip and Warren upswing.

Confirmational (sic) Bias? Never heard of it.

[DrToast]

19. Thanks for the heads up

It’s seems pretty pointless.

[bluewater]

17. Seems like confirmation bias from where I sit.

Which polls? There are polls that have consistently shown Biden in the low 30% range.

Other polls have shown Biden consistently in the low 20% range.

Obviously pollsters are in disagreement, 10% is a significant difference.

Why is it so drastic that the latest Monmouth poll had results similar to those polls showing Biden in the low 20% range?

Monmouth is a live caller poll. Isn't it possible that maybe, just maybe, their results can capture the volatility of the polling landscape?

So, the latest Monmouth poll is NOT saying “This race has drastically changed!”

It is saying that it's latest poll is more in line with the polling by other firms showing Biden in the low 20% range of support.

Why is that such a drastic finding?

Honestly, the only thing making the latest Monmouth poll results seem at all drastic is a confirmation bias against the possibility Biden's support might be in the low 20% range.

[Thekaspervote]

18. With great conviction you have told us that the polls are the polls and you are right, they are

Saying this with kindness.... maybe you should consider your own best advice.

The old saying of “keeping one’s own counsel” it’s worthy advice for anyone

[bluewater]

20. The polls are what the polls are.

Why ignore the ones showing Biden in the low 20% range?

[Thekaspervote]

24. My point was lost

[bluewater]

25. Mine too, apparently.

Such is life.

[George II]

52. But you only listed the results for only one candidate, what about all the others? Did they....

....show Warren and BS at around 20% like the Monmouth poll? I don't think so.

The organization that conducted the poll itself calls it an outlier. Why is it so difficult to agree with the pollster?

[Gothmog]

15. From my twitter feed

Link to tweet

[BannonsLiver]

21. What do you think of Elizabeth Warren's healthcare ideas?

Last edited Wed Aug 28, 2019, 07:29 PM - Edit history (1)

What about her plan to curb gun violence, climate change and banks?

[bluewater]

23. Check out my journal!

I have a lot of OP's on both Warren and Harris in there, both of whom I like a lot.

[BannonsLiver]

26. So nothing off the top of your head.

“Check out my journal.”

[bluewater]

27. Tangential discussions aren't my thing. ;)

Start a thread about those and I will be glad to join in if you really wish to discuss those policies.

[BannonsLiver]

31. I actually did check it out.

And 18 of the first 20 posts i encountered were related to Biden (and the belief that he is a terrible candidate) and polling in general.

From looking at the journal, as was suggested, it would seem the purpose here has very little to do with advocating for Warren.

[bluewater]

36. Seems like you really didn't check it out closely...

Out of the last 21 journal entries:

7 were about recent poll results and related to all the candidates.

3 were about the Primary debates and who were qualifying.

5 were about Warren

5 were about Biden

1 was about Trump.

So, since 5 were just about Warren, plus 7 were about polls that included Warren, and 3 were about the debates that Warren would participate in... it would seem that at least 15 of the RECENT journal articles were related to Warren.

True, there were 5 posts just about Biden, but gee whiz, he is the front runner and gets a lot of media coverage.

So it would seem that you didn't actually take a close enough look at my journal.

[BannonsLiver]

37. So of the 21, "only" 16 weren't related to Warren.

[bluewater]

38. lol "only" if polls and the debates are not related to Warren.

Good god man! She IS running for President!



Ok, I think I am allowed one of these now...



Thanks for the discussion!

[OilemFirchen]

41. OMG!

ROFLMAO!

[bluewater]

42. lol

[qazplm135]

28. sure but

to your own point, where are the other candidates in those polls?
The outlier wasn't merely "Biden at 19 percent."

Analyzing this without bias as you say would involve looking at the polls in their entirety.

This poll had Biden behind, and two of the folks he was behind with some of their highest numbers. There were also big differences in three of the candidate's numbers. Of course also the small sample size.

It may very well be that Biden's support is somewhere around the low 20s, but that's not the only thing this poll said and not the only reason why it's probably at this point, an outlier.

Of course, we are still six months out so all the polls right now could end up being "outliers" once we get to game time, one way or the other. Maybe it ends up Biden falls apart. Maybe it ends up Sanders blows into the lead. Maybe it ends up Warren takes the first two. Maybe it ends up Biden takes the first two and SC and Arizona. Maybe Harris surprises in the last two with Biden winning one and Warren winning one and we are in a bit of a mess going into Super Tuesday. Maybe Buttigieg surprises in Iowa.

Lot of possibilities.

[bluewater]

29. The main point was Biden dropping 13% in that poll.

It really was the focus of the coverage about that poll.

The Monmouth poll and those I listed showed Sanders and Warren about where they usually are in almost all the polls, namely between 15-20%.

The Monmouth poll only had Sanders and Warren tied at 20%, it did NOT show any big drastic change in their polling support.

The only drastic change was Biden's support falling more than twice the +/-5.7% margin of error of the poll.

[qazplm135]

30. it also had

Warren and Sanders I believe +6 or +7 did it not?

Big rises for them, big falls for Biden.

A whole lot of movement for multiple candidates when not much newsworthy has happened is always a basis for skepticism.
As was the N. Ask Rachel Bitecofer, she tweeted the same thing.

[Vegas Roller]

33. The correct statistical term is

Confirmation Bias

But .... I must say...

OMG OMG OMG OMG OMG !!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!

[bobGandolf]

34. I agree with you, and then some.

I listened to them over and overemphasizing the Mammouth Poll as an outlier.
Morning Joe has pretty much anointed him. It boggles my mind how blatant they are about it.

[LanternWaste]

35. Ah, so you guys are dusting off the "anointed one" moniker again, eh?

I get it... bumper stickers are the best a lot of people have.

[bluewater]

49. Assuming gender?

Not a good look.

bumpitty bump.

[OilemFirchen]

50. OMG!

OMFG!

[bluewater]

53. lol

bumpitty bump.



They say you can't tell the players without a scorecard. Thanks for making things clearer.

[Clash City Rocker]

39. Can we just not bloviate over every poll?

Regardless of your favorite candidate, it’s too early for any poll to mean very much. Talking about these polls leads to a lot of inter-party squabbling, when the real enemy is the Republicans. Let’s not turn into these guys.

https://m.

[emmaverybo]

45. Curb your anxiety. As you purport to show, Biden isn't an authentic front runner, polls mean nothing

at this stage, and his support is unenthusiastic and shallow.

Why, he is so not a threat that you can afford to make a case several times a day for your preferred candidate.

[Thekaspervote]

47. This!

[still_one]

48. Excellent observation