Thread by @DamianSurvation

Some notes on our Hartlepool polling and a new phenomenon, *Brexit Party* Amnesia - a feature, not a bug.

We interviewed 500+ Hartlepool residents last week & as can be seen from the data tables the number of people stating "The Brexit Party" to the question "Thinking back to the general election in December 2019, which party did you vote for, or did you not vote? was really low.

As per our house style, this is an unprompted question - , but still, only 4% of those that voted remembering voting BXP vs the 26% figure we know from GE 2019 is interesting.

Brexit Party 2019 voters are in the sample (e.g. the proportion of leave 2016 voters was recalled at 66% vs the 69.6% actual) - however many are incorrectly recalling they voted LAB or CON. There are logical reasons for not remembering voting for BXP 2 years ago however...

We would normally use recalled past vote as a weighting factor, but given this "Brexit Party Amnesia" that did not make sense here, and weighted by age, sex, ward group (weighting efficiency 1.21) = seven point CON lead.

Aha! Shout the polling commentators. Non-standard weighting! (true) Missing BXP voters! (hidden not missing) Labour could be ahead! (not true in this poll).
Why so?

Remember our weighting efficiency of 1.21 using age/sex/ward group and the 7 point CON lead?(A weighting efficiency figure of *1* here would represent no change created by the weights used to the top line from the unweighted data - the optimal situation).

If we had used past vote 2019 as an additional weighting target, the CON lead would have been *higher* - 10 points but the efficiency a much worse 1.77. Adding EU Ref 2016 to GE 2019? A 10 point CON lead and weighting efficiency of 2.3 (even worse).

So there you have it. Without reminding people, The Brexit Party of 2019 in the post Brexit world is not being remembered. We might have to jog people's memories through prompting and see what difference it makes. Brexit Party Amnesia is *not* a reason to be cheerful i'm afraid.