@SimonMcMahon i put another limit up @ 18% if you want

Note this market with a similar condition, but with a difference of up to 3% instead of 2%: https://manifold.markets/ShreyasArathrigqgN/the-iowa-selzer-poll-will-prove-to

The poll reports 5% split between "Not Sure" and "Don't want to say". Even if Selzer had the relative proportion of Trump and Harris voters exactly correct, shouldn't expect at least one of them to get another 2% from this group?

@Nick6d8e Will that majority come from people who are ashamed to be supporting the shameless racist or women who want to make sure they have access to healthcare without angering their husbands or the members of their congregation? Will it be an even split? Will some of those support 3rd party candidates?

@becauseyoudo A perfectly even split AND a full 20% of the Not Sure/Don't Want to Say going third party still puts both Trump and Harris' numbers 2% off. If the split is anything but perfect, or the Not Sures go any less than 20% third party, one candidate is outside of the 2% bound. And that's assuming that Selzer's numbers for Harris and Trump are exactly correct.

@Nick6d8e So it only really works if the % of 3rd party support is higher than 20% and it's not an even split.

So there will be a minimum of 5% voting for neither?

seems like a very easy no “?

@cryptog Good assumption but Iowa votes for 3rd party candidates at a higher rate than other states. This year there is support for a few independent and minor party candidates.

This should be strictly lower than the main market on who will win Iowa.

This relies on a large number of blank/spoiled/third party votes, right? People can't vote "undecided".

@BrunoParga the poll suggests there is a lot of 3rd party support including support for Kennedy which is no longer running officially but is still a registered candidate.

@BrunoParga Even the upper value for both candidates implies at least 5% of the ballots have to be for 3rd party candidates. I find that pretty unlikely.

@TimothyJohnson5c16 The independent candidate won 18.71% of the vote in Iowa in 1992 and 8.52% in 1996. Kennedy was the strongest independent candidate in decades and his vaccine messaging resonated with a lot of people post COVID. The two major party candidates are also not incredibly popular.

@becauseyoudo The Selzer poll also put RFK at 3%. That's very good for an independent, and he might have been even higher if he hadn't sort of dropped out.

But that doesn't seem like enough for this market.

@becauseyoudo I'm not sure the criteria as stated capture the question title very well. I understand this is not a straightforward matter. I don't know how I'd make it better; maybe focus only on the two-party vote. The poll says 47-44, that's a total of 91. Multiply both those numbers by 100/91, then consider the poll "accurate" if the final two-party percentage falls within some range of those numbers. But anyway, it's not my question.

@BrunoParga The poll suggests a 3 point margin so another question could test the accuracy of the poll based on the suggested margin between the major party candidates but I think it's useful to focus on the absolute accuracy of the poll and questions like the likelihood of seeing high support for 3rd party candidates in such a close race.

@becauseyoudo how much third-party support does the poll predict? I imagine it's less than the 9% that said they'll vote for neither viable candidate, because even today there are still people who claim they're undecided. At the very least those should be factored out.

@TimothyJohnson5c16 I checked the numbers and it's 3% not sure, 2% won't say. My point is that at least these two numbers can't possibly be "accurate" in comparison with the final result, that doesn't make sense. And I believe since everything has to add up to 100%, that makes a simplistic calculation problematic.

Ultimately, people spend their mana however they want. I just don't think this comparison is very meaningful.

@TimothyJohnson5c16 lol I should have tagged @becauseyoudo in that comment.

@BrunoParga Selzer claims the poll is limited to likely voters and is based on a weighted representative sample of the state's population. I really don't think it's going to add up to 100% for these two candidates in Iowa and there's no requirement that it does.

@becauseyoudo I'd like to know what you mean in terms of actually numbers. If you'd humor me for a moment, let's assume for the sake of argument that the poll is perfectly, miraculously accurate. Can you please write here what percentages of votes you think the Iowa Secretary of State will report in that scenario? Assume they're faithfully reporting the votes actually cast, no shenanigans.

To recap, the candidates' numbers in the poll are:

• Harris 47

• Trump 44

• Kennedy 3

• the libertarian guy <1

• other candidates 1

And then there's the non-candidate answers "not sure" (3%) and "won't say" (2%); we know those poll answers won't be included in the official result.

The SoS will report total vote numbers which can always be converted to percentages, so your answer should add up to 100%, with a small margin for rounding.

Can you please tell me those numbers?

@BrunoParga are you asking me to change the parameters of the market? I think it's too late for that.

If you're asking where I think the 5% that "don't know" or "won't say" but have either already voted or are planning on voting tomorrow will be tabulated, I have no idea. Cohn seems to think the majority of those nonresponsive voters are pro Trump but the Harris campaign has been specifically promoting nonresponsive or duplicitous voting behavior in the last few weeks.

@becauseyoudo no, I am asking you to write me a comment with the numbers you predict would be reported by the Iowa SoS given two assumptions: the poll is magically perfect and the SoS does their job. Just that. Nothing else.

@becauseyoudo like, I am requesting you to do you exactly what I wrote, if you'd humor me.

@BrunoParga Just to humor you, here are some random numbers that fit the criteria and add up to 100%.

Harris 48%

Trump 46%

Kennedy 3.5%

Libertarian 0.9%

Other 1.6%

To get to these numbers, in most states, you would have to assume some level of candidate fatigue which isn't a stretch with any candidate running for president for the 3rd time, let alone a candidate with narcissistic tendencies and an outlandish personality. If the candidate spent the last few weeks melting down and talking about firing squads and persecuting his enemies he could risk alienating some of his previous support.

@becauseyoudo okay, thank you for the numbers.