Opinium

[timmullen1]

Apr 9, 2022 21:24:38 GMT bridgyoldboy said:

Apr 9, 2022 19:22:23 GMT timmullen1 said:

what is the lead on old methodology?

Pass; am on an iPhone and I can’t open the data tables.

[clyde1998]

Apr 5, 2022 18:12:32 GMT batman said:

It's hard to believe other than that in Hungary the pollsters just wanted to believe that Orban would lose & "adjusted" the figures accordingly. I'm quite sure that wishful thinking occurs within some British polling too, although polling in this country has become more reliable than it was a generation ago (e.g. the 1992 election)

I wonder if it was a weighting/sample issue, rather than any deliberate attempt to overinflate the opposition. Looking at the vote shares of those living in Hungary, about half of polls 'only' seem to be about three points out for the main parties - a six point swing in Orban's favour (this is roughly comparable to our 2015 general election). Unexpected differential turnout, an inaccurate sample (particularly in rural areas), a late swing, a heavily disproportional split among undecided voters or any combination of factors could explain a difference of that scale.

The other half may have had these issues to an even more extreme level, or had questions worded (or ordered) in a way that overinflated the opposition. Without access to any of the data tables, it's hard to give a clear view of what went wrong.

One additional point I'd note, the diaspora vote went 93%+ to Fidesz - moving the overall result further in their favour. Fidesz were 1.5pp higher and the United Opposition were 2.5pp lower once factoring in the diaspora - which would be very hard to poll accurately.

[batman]

Apr 10, 2022 7:51:59 GMT clyde1998 said:

Apr 5, 2022 18:12:32 GMT batman said:

It's hard to believe other than that in Hungary the pollsters just wanted to believe that Orban would lose & "adjusted" the figures accordingly. I'm quite sure that wishful thinking occurs within some British polling too, although polling in this country has become more reliable than it was a generation ago (e.g. the 1992 election)

I wonder if it was a weighting/sample issue, rather than any deliberate attempt to overinflate the opposition. Looking at the vote shares of those living in Hungary, about half of polls 'only' seem to be about three points out for the main parties - a six point swing in Orban's favour (this is roughly comparable to our 2015 general election). Unexpected differential turnout, an inaccurate sample (particularly in rural areas), a late swing, a heavily disproportional split among undecided voters or any combination of factors could explain a difference of that scale.

The other half may have had these issues to an even more extreme level, or had questions worded (or ordered) in a way that overinflated the opposition. Without access to any of the data tables, it's hard to give a clear view of what went wrong.

One additional point I'd note, the diaspora vote went 93%+ to Fidesz - moving the overall result further in their favour. Fidesz were 1.5pp higher and the United Opposition were 2.5pp lower once factoring in the diaspora - which would be very hard to poll accurately.

pardon the digression; until not long ago we had a Patisserie Valerie in Richmond, where I quite often had coffee with the late Prof Byron Criddle. They had a young Hungarian woman as the manager and once I mentioned to her that in certain areas of London where I'm sent to work I get quite a few Hungarian respondents. "Tell me which areas these are" she replied "and I'll make a point of avoiding them." I think I understand what she meant after reading the above.

[The Bishop]

The diaspora being so massively pro-government seems slightly unusual?

[John Chanin]

Apr 10, 2022 9:28:13 GMT The Bishop said:

The disapora being so massively pro-government seems slightly unusual?

If you read up thread you will see that it is mainly ethnic Hungarians who live in Slovakia, Serbia, Romania, and Croatia, who naturally are very keen on greater hungarian nationalism.  They have all been given the vote by Orban, while voting by emigrants in places like the UK has been made awkward and difficult.

[timmullen1]

[bridgyoldboy]

Apr 23, 2022 19:06:54 GMT timmullen1 said:

Bearing in mind the change in methodology adopted about 6 weeks ago which is broadly based on making an assumption that a certain percentage of don’t knows will return to the Tory fold it would be interesting to know what the Labour lead would be under the old methodology and what the likely implications will be for the local elections in 12 days time?

[batman]

I really think this is an outlier with an odd sample. All other polls have had Labour's lead at least 6% for a couple of weeks & some quite a bit higher than that. It ceases to be an outlier of course if another poll or two find similarly.

[andrew111]

Apr 23, 2022 20:33:48 GMT batman said:

I really think this is an outlier with an odd sample. All other polls have had Labour's lead at least 6% for a couple of weeks & some quite a bit higher than that. It ceases to be an outlier of course if another poll or two find similarly.

-2% for Labour is within the margin of error and doesn't need an odd sample.

As the difference between two variables, the Labour lead is bound to fluctuate quite a bit, statistically speaking.

[timmullen1]

It was an odd survey, assuming it was the one I did on Wednesday, as it spent seemingly forever on issue questions - arming Ukraine, Rwanda migrants, Brexit, energy prices, before coming to the VI section; I don’t know what the feeling here is but credible polling analysts in the US strongly argue that you should do VI before issue questions.

[bridgyoldboy]

Apr 23, 2022 23:23:17 GMT timmullen1 said:

It was an odd survey, assuming it was the one I did on Wednesday, as it spent seemingly forever on issue questions - arming Ukraine, Rwanda migrants, Brexit, energy prices, before coming to the VI section; I don’t know what the feeling here is but credible polling analysts in the US strongly argue that you should do VI before issue questions.

Could that approach not be considered as “push polling”?

[John Chanin]

Apr 23, 2022 23:23:17 GMT timmullen1 said:

It was an odd survey, assuming it was the one I did on Wednesday, as it spent seemingly forever on issue questions - arming Ukraine, Rwanda migrants, Brexit, energy prices, before coming to the VI section; I don’t know what the feeling here is but credible polling analysts in the US strongly argue that you should do VI before issue questions.

Yes that is correct survey practice.  You should always do your ‘general satisfaction’ at the beginning before drilling down into whatever detailed subjects you are interested in. Otherwise people’s views on the specific will unnaturally colour their general satisfaction responses for the 5 minutes it takes to complete the questionnaire.  Everyone who constructs surveys knows this, even those who deliberately ignore it for campaigning or partisan reasons.

And yes, for political polling VI counts as general satisfaction.

[The Bishop]

Indeed, whenever I have been asked about VI by YouGov it has been before the more "specific" questions.

[aargauer]

Apr 10, 2022 9:28:13 GMT The Bishop said:

The diaspora being so massively pro-government seems slightly unusual?

Isn't much of the Turkish diaspora in Germany pro Erdoghan?

[timmullen1]

[graham]

In terms of the previous methodology that probably equates to a Labour lead of 9% or 10%.

[jamie]

Interesting. Feels a bit obvious that your financial situation significantly impacts your voting behaviour, but the discussion around ‘is class politics dead?’ may lead you to think otherwise.

[mattbewilson]

May 28, 2022 18:57:51 GMT jamie said:

Interesting. Feels a bit obvious that your financial situation significantly impacts your voting behaviour, but the discussion around ‘is class politics dead?’ may lead you to think otherwise.

that would suggest class is determined by whether you're struggling or not

[jamie]

May 28, 2022 19:00:03 GMT mattbewilson said:

May 28, 2022 18:57:51 GMT jamie said:

Interesting. Feels a bit obvious that your financial situation significantly impacts your voting behaviour, but the discussion around ‘is class politics dead?’ may lead you to think otherwise.

that would suggest class is determined by whether you're struggling or not

It’s probably not entirely unrelated. 

[mattbewilson]

May 28, 2022 19:03:14 GMT jamie said:

May 28, 2022 19:00:03 GMT mattbewilson said:

that would suggest class is determined by whether you're struggling or not

It’s probably not entirely unrelated. 

I agree that if are working class you're probably more likely to struggling atm than if you're middle class but you're not working class because you're struggling