Deleted
Deleted Member
Posts: 0
|
Post by Deleted on Feb 3, 2014 12:25:31 GMT
Today 41/32/11/9 but data looks more than a bit weird , Labour ahead of Con in every region of England including SE but Conservatives ahead of Labour in Scotland . :S |
|
Sibboleth
Labour
'Sit on my finger, sing in my ear, O littleblood.'
Posts: 16,044
|
Post by Sibboleth on Feb 3, 2014 15:20:47 GMT
Poll internals are generally best ignored. Even aggregated ones. I regret that I'm in a minority on this matter.
|
|
|
|
Post by marksenior on Feb 3, 2014 16:11:13 GMT
Poll internals are generally best ignored. Even aggregated ones. I regret that I'm in a minority on this matter. I agree that internal sub samples are not meaningful however a bad internal sub sample can in some circumstances make the whole sample give false results . |
|
|
|
Post by marksenior on Feb 7, 2014 11:56:11 GMT
Today's Populus Lab 36 Con 33 LD 9 UKIP 15
Massive changes to Populus weightings make the results not comparable to previous Populus polls
|
|
|
|
Post by woollyliberal on Feb 7, 2014 12:51:39 GMT
What changes have they made?
|
|
|
|
Post by marksenior on Feb 7, 2014 13:00:34 GMT
What changes have they made? The changes are to weighting by Party ID . My calculations are that the effects are Lab -2 LD -2 and UKIP + 4 Anthony Wells has Con -1 Lab -1 LD -2 UKIP +3 but I cannot see why he should have Con -1 as their weightings have not changed at all . The Lab and UKIP changes are more correctly -1.5 and + 3.5 so rounding effects will give Lab -1 or -2 and UKIP +3 or +4 |
|
|
|
Post by woollyliberal on Feb 7, 2014 13:06:53 GMT
Thanks Mark. Populus were alone at rating UKIP support so low, so it's probably about right. Interesting that they reduce Lab and LD to increase UKIP but not Con.
|
|
Andrew_S
Top Poster
Posts: 28,241
Member is Online
|
Post by Andrew_S on Feb 7, 2014 13:14:41 GMT
|
|
|
|
Post by marksenior on Feb 7, 2014 13:34:06 GMT
Thanks Mark. Populus were alone at rating UKIP support so low, so it's probably about right. Interesting that they reduce Lab and LD to increase UKIP but not Con. Whilst true , that puts Populus in line with the other internet panel pollsters but does nothing to clarify whether the online pollsters or the telephone pollsters are correct . |
|
|
|
Post by anthonyjwells on Feb 7, 2014 13:56:42 GMT
What changes have they made? The changes are to weighting by Party ID . My calculations are that the effects are Lab -2 LD -2 and UKIP + 4 Anthony Wells has Con -1 Lab -1 LD -2 UKIP +3 but I cannot see why he should have Con -1 as their weightings have not changed at all . The Lab and UKIP changes are more correctly -1.5 and + 3.5 so rounding effects will give Lab -1 or -2 and UKIP +3 or +4 The figures I've quoted are what Populus told me the results would have been if they'd weighted the data using their old regime. |
|
|
|
Post by marksenior on Feb 7, 2014 14:29:04 GMT
The changes are to weighting by Party ID . My calculations are that the effects are Lab -2 LD -2 and UKIP + 4 Anthony Wells has Con -1 Lab -1 LD -2 UKIP +3 but I cannot see why he should have Con -1 as their weightings have not changed at all . The Lab and UKIP changes are more correctly -1.5 and + 3.5 so rounding effects will give Lab -1 or -2 and UKIP +3 or +4 The figures I've quoted are what Populus told me the results would have been if they'd weighted the data using their old regime. OK , thanks for that . As I said their Conservative weighting is to 28% before and after the adjustments so I fail to see how they can be correct in that the Conservative figure should be down by 1 . |
|
Sibboleth
Labour
'Sit on my finger, sing in my ear, O littleblood.'
Posts: 16,044
|
Post by Sibboleth on Feb 7, 2014 17:27:41 GMT
Why do polling companies expect to be taken at all seriously when they do this sort of thing?
|
|
|
|
Post by marksenior on Feb 7, 2014 17:36:03 GMT
Why do polling companies expect to be taken at all seriously when they do this sort of thing? Well all polling companies have their own individual methodological adjustments which they believe will give their polls greater accuracy . This , though , is I think the first time a pollster has changed their methodology just so that its results are closer to certain other pollsters but farther away from others ( the telephone pollsters ) . |
|
Sibboleth
Labour
'Sit on my finger, sing in my ear, O littleblood.'
Posts: 16,044
|
Post by Sibboleth on Feb 7, 2014 18:10:49 GMT
Half of them probably just make their numbers up. There have been a number of scandals of that sort in the U.S in recent years.
|
|
|
|
Post by marksenior on Feb 10, 2014 14:47:54 GMT
Today's Populus same weightings as Friday's poll
Lab 36 N/C Con 34 ( +1) LD 11 ( +2 ) UKIP 12 ( -3 )
|
|
Deleted
Deleted Member
Posts: 0
|
Post by Deleted on Feb 10, 2014 15:25:03 GMT
Seems that this new methodology is very tory friendly.
|
|
|
|
Post by marksenior on Feb 10, 2014 15:40:06 GMT
Seems that this new methodology is very tory friendly. The Conservative weightings are unchanged so the changes should not have any effect on their VI . They do knock 2% off the Labour VI and therefore the Labour lead and 2% off the LD VI adding 4% to UKIP . Thinking about this logically , as more UKIP support has come from the Conservatives and least from Labour a more logical change would have been to reduce Con VI by 1.5% and Labour by 0.5% and increased the Labour lead by about 1% . |
|
The Bishop
Labour
Down With Factionalism!
Posts: 39,015
Member is Online
|
Post by The Bishop on Feb 10, 2014 16:23:39 GMT
Populus surely cannot be unaware that most UKIP support has come from the Tories - or at any rate much more than Labour? Curious.
|
|
|
|
Post by Davıd Boothroyd on Feb 10, 2014 16:42:40 GMT
|
|
Deleted
Deleted Member
Posts: 0
|
Post by Deleted on Feb 14, 2014 10:01:27 GMT
electionista @electionista
UK - Populus poll:
CON 32%
LAB 38%
LDEM 9%
UKIP 14%
A more 'normal' result
|
|
