Excluding DKs doesn't make them go away

It incorporates to the poll's observation an assumption they will split proportionally to the decided

The veracity of this assumption has nothing to do with accuracy of the poll - but is ultimately used to measure it

Hence: unscientific

Sunak is unpopular and I'd be shocked if Labour didn't win, whenever the general election is.

BUT

A bunch of Reform will end up voting Conservative, fair number of 'don't knows' will break Con.

Right now, these people are lying saying 'Labour by 21' and going to blame the

Not sure who cares about UK elections

But they're doing it again

'It' meaning 'lying about their polls'

Tell me how you get Labour at 44% out of this?

Better yet, tell me how an AVERAGE of polls gets to 44% from this data.

You can't, because it's lying.

Riddle:

How do you get an AVERAGE in the 40s from a bunch of polls in the low 30s?

(Hint: it requires lying)

Me: tons of data to show undecideds almost never split 50/50. So to characterize the error of 'the polls' based on this false assumption is really bad.

In cases where they go for the candidate polling lower, this assumption can cause more error than the poll!

Statisticians:

Stuff like this is why I'm a 'put up or shut up' guy.

He blocked me

Was it something I said?

Thanks to Kabir K. / kabirkhanna.bsky.social for reporting the numbers as they actually are, and not by 'spread'

This is true

I graduated Columbia in ‘68 & was heavily involved in that set of campus protests.

1) Outside agitators were a problem then & likely now

2) Antisemitism has no place anywhere, least of all at schools

3) Hamas is a terrorist organization

4) The Israeli govt has got to go

Young people:

We want less war, affordable housing and healthcare, jobs better than indentured servitude, and a planet we can live and retire on

Nate Silver:

Mostly just group allegiance and signaling imo

Gritty Get Litty 🚩 umichvoter 🏳️‍🌈 'winning' and 'ahead, based only on voters who are currently decided'

Are not the same

Me: margin is a junk metric

Experts: UGH you're so PEDANTIC we KNOW what margin measures

Poll data: 46-45

Experts: this data predic person with 46 will win by 1 🧐

Lmao

omG WhY BiDeN At 43

Please, shut up

Each of these dots represents a poll taken from a known population of 50%

Almost none of them returned exactly 50%, and many were far off

These are IDEAL polls taken from KNOWN population

Consider possibility that nonideal polls from unknown population could fluctuate more

When other people knowingly lie

'Illegals only allowed in to vote Dem'

It's tempting to show them how and why they're wrong

But it's a brilliant tactic to lean into it.

'Yep, they sure are, that's why we're doing it. I'm helping them do it and that's why you can't win'

Dear poll aggregators:

Taking a poll with a 'lean' question that produces '49-47'

And another with no lean question that produces '45-45'

Does not equal an average of 47-46.

Do with that what you will.

There's a disconnect between people who want to be informed by data, and people who want to produce it