Deleted
Deleted Member
Posts: 0
|
Post by Deleted on Sept 23, 2014 7:43:54 GMT
Support for Ed Miliband is....
|
|
|
|
Post by Devonian on Oct 15, 2014 15:07:14 GMT
|
|
Sibboleth
Labour
'Sit on my finger, sing in my ear, O littleblood.'
Posts: 16,044
|
Post by Sibboleth on Oct 15, 2014 15:12:20 GMT
Though not that plus one percentage point is not statistically significant movement.
|
|
|
|
Post by Devonian on Oct 15, 2014 15:38:27 GMT
Though not that plus one percentage point is not statistically significant movement. However I think the fact that three separate polling companies have given UKIP a record high score in the past few days is statistically significant. |
|
|
|
Post by thirdchill on Oct 15, 2014 17:35:01 GMT
Though not that plus one percentage point is not statistically significant movement. However I think the fact that three separate polling companies have given UKIP a record high score in the past few days is statistically significant. It will only be significant if it stays that way. If it drops back to the usual in 3 weeks it'll not mean very much. |
|
|
|
Post by carlton43 on Oct 15, 2014 21:47:46 GMT
However I think the fact that three separate polling companies have given UKIP a record high score in the past few days is statistically significant. It will only be significant if it stays that way. If it drops back to the usual in 3 weeks it'll not mean very much. No. It is significant whether it subsequently goes up, down or remains the same. Every one is a certain sort of sample by snapshot and is significant in its own way for that moment. When a series of such snapshots all break to a new high that makes the significance just a little more significant.......for that moment. Nothing is static or forever. Significance is a relativity to a moment because all is ultimately in flux and all ultimately equally meaningless and inconsequential. |
|
|
|
Post by Devonian on Oct 15, 2014 22:17:19 GMT
However I think the fact that three separate polling companies have given UKIP a record high score in the past few days is statistically significant. It will only be significant if it stays that way. If it drops back to the usual in 3 weeks it'll not mean very much. What is 'the usual' for UKIP? This is the Wikipedia graph of polling average of the four parties  Perhaps you can point out exactly what percentage is 'the usual' for UKIP? 'The usual' it fell back to after the May 2013 elections was higher that 'the usual' before. 'The usual' it fell back to after the Euro elections was higher than 'the usual' a year earlier. The fact that Ipsos-MORI, Survation and and now two nights in a row YouGov have all shown record high scores for UKIP in the past few days suggests that 'the usual' level for UKIP is rising. |
|
john07
Labour & Co-operative
Posts: 15,818 
|
Post by john07 on Oct 15, 2014 22:23:04 GMT
It will only be significant if it stays that way. If it drops back to the usual in 3 weeks it'll not mean very much. No. It is significant whether it subsequently goes up, down or remains the same. Every one is a certain sort of sample by snapshot and is significant in its own way for that moment. When a series of such snapshots all break to a new high that makes the significance just a little more significant.......for that moment. Nothing is static or forever. Significance is a relativity to a moment because all is ultimately in flux and all ultimately equally meaningless and inconsequential. Wrong. While polls certainly are snapshots in time, so are election results. Significance is something different. A poll is statistically significant (at say 5%) if the result could have arisen through random chance more than 5 times in a hundred. Put another way if we could reject the hypothesis that the UKIP vote was no higher than previously recorded with 5% significance (95% confidence). Having three companies giving similar results certainly strengthens the likelihood that the results are indeed significant. However since poll companies do do use random samples are use manipulated stratified samples makes it harder to draw inferences. |
|
|
|
Post by jonarny on Oct 16, 2014 1:01:59 GMT
carlton43 is only wrong if he intended the word 'significance' in statistical terms; I suspect the suggestion that every poll 'is significant in its own way' indicates that it was not intended as such. There is also anon-statistical significance of the Survation poll which showed UKIP on 25%; even if it turns out to be a rogue (and statistically of little significance) its media impact was significant.
I suspect that taken across the three polls concerned, a null hypothesis "that the UKIP vote was no higher than previously recorded" would be rejected at least at the 5% level and probably also at 2.5%. .
|
|
|
|
Post by East Anglian Lefty on Oct 16, 2014 7:55:14 GMT
It will only be significant if it stays that way. If it drops back to the usual in 3 weeks it'll not mean very much. What is 'the usual' for UKIP? This is the Wikipedia graph of polling average of the four parties Perhaps you can point out exactly what percentage is 'the usual' for UKIP? 'The usual' it fell back to after the May 2013 elections was higher that 'the usual' before. 'The usual' it fell back to after the Euro elections was higher than 'the usual' a year earlier. The fact that Ipsos-MORI, Survation and and now two nights in a row YouGov have all shown record high scores for UKIP in the past few days suggests that 'the usual' level for UKIP is rising. The high for this May and the high for last May are not radically different. The level in between those two points was consistent enough and lasted for long enough that it can be treated as 'the usual'. |
|
|
|
Post by carlton43 on Oct 16, 2014 9:33:25 GMT
No. It is significant whether it subsequently goes up, down or remains the same. Every one is a certain sort of sample by snapshot and is significant in its own way for that moment. When a series of such snapshots all break to a new high that makes the significance just a little more significant.......for that moment. Nothing is static or forever. Significance is a relativity to a moment because all is ultimately in flux and all ultimately equally meaningless and inconsequential. Wrong. While polls certainly are snapshots in time, so are election results. Significance is something different. A poll is statistically significant (at say 5%) if the result could have arisen through random chance more than 5 times in a hundred. Put another way if we could reject the hypothesis that the UKIP vote was no higher than previously recorded with 5% significance (95% confidence). Having three companies giving similar results certainly strengthens the likelihood that the results are indeed significant. However since poll companies do do use random samples are use manipulated stratified samples makes it harder to draw inferences. You do not have a copyright on the definition or the usage of the word significant. There may be a specific understood meaning entirely within the narrow study of statistics? I neither know nor care about that. The word has a general and much earlier ordinary meaning and it is used in that manner by me. That use is perfectly valid and IMO makes more sense to an ordinary person. You scholastic types imbued with arcane 'how many angels can dance on the head' of your statistical pin, may see it otherwise. The rest of us are simpler souls. |
|
john07
Labour & Co-operative
Posts: 15,818
|
Post by john07 on Oct 16, 2014 9:47:23 GMT
Wrong. While polls certainly are snapshots in time, so are election results. Significance is something different. A poll is statistically significant (at say 5%) if the result could have arisen through random chance more than 5 times in a hundred. Put another way if we could reject the hypothesis that the UKIP vote was no higher than previously recorded with 5% significance (95% confidence). Having three companies giving similar results certainly strengthens the likelihood that the results are indeed significant. However since poll companies do do use random samples are use manipulated stratified samples makes it harder to draw inferences. You do not have a copyright on the definition or the usage of the word significant. There may be a specific understood meaning entirely within the narrow study of statistics? I neither know nor care about that. The word has a general and much earlier ordinary meaning and it is used in that manner by me. That use is perfectly valid and IMO makes more sense to an ordinary person. You scholastic types imbued with arcane 'how many angels can dance on the head' of your statistical pin, may see it otherwise. The rest of us are simpler souls. But the post you replied to made reference to statistical significance. That has one accepted meaning. Especially as this is a thread about opinion polling. |
|
|
|
Post by carlton43 on Oct 16, 2014 10:46:10 GMT
You do not have a copyright on the definition or the usage of the word significant. There may be a specific understood meaning entirely within the narrow study of statistics? I neither know nor care about that. The word has a general and much earlier ordinary meaning and it is used in that manner by me. That use is perfectly valid and IMO makes more sense to an ordinary person. You scholastic types imbued with arcane 'how many angels can dance on the head' of your statistical pin, may see it otherwise. The rest of us are simpler souls. But the post you replied to made reference to statistical significance. That has one accepted meaning. Especially as this is a thread about opinion polling. I have looked back on the threads John and you are right about the initial use of 'statistical' which I had skated over in my reading. So I apologize to you. I was commenting quite outside the narrow area of statistics but on the generality of observed trends as seen by an 'ordinary man'. Also the wider point of long term ephemerality of it all! |
|
Deleted
Deleted Member
Posts: 0
|
Post by Deleted on Nov 12, 2014 12:45:57 GMT
Evening Standard poll Con 32% Labour 29% UKIP 14% Lib Dem 9% SNP 8% Green 7%,  |
|
The Bishop
Labour
Down With Factionalism!
Posts: 39,009
|
Post by The Bishop on Nov 12, 2014 13:41:24 GMT
SNP 8% and set to win all 59 seats there - even they probably don't believe that.
|
|
Deleted
Deleted Member
Posts: 0
|
Post by Deleted on Nov 12, 2014 13:53:12 GMT
SNP 8% and set to win all 59 seats there - even they probably don't believe that. well, some of them ...... |
|
Tony Otim
Green
Suffering from Brexistential Despair
Posts: 11,913
Member is Online
|
Post by Tony Otim on Nov 12, 2014 13:54:20 GMT
Unless they're picking up a large number of voters south of the Border. SNP gain in Corby  |
|
Sibboleth
Labour
'Sit on my finger, sing in my ear, O littleblood.'
Posts: 16,044
|
Post by Sibboleth on Nov 12, 2014 18:32:03 GMT
Some slightly... mad... figures there. Scotland only makes up about 8% of the GB population!
|
|
Deleted
Deleted Member
Posts: 0
|
Post by Deleted on Nov 12, 2014 18:45:06 GMT
Some slightly... mad... figures there. Scotland only makes up about 8% of the GB population! It's because us pesky Scots are all riled up and more likely to say we'll vote. |
|
The Bishop
Labour
Down With Factionalism!
Posts: 39,009
|
Post by The Bishop on Nov 12, 2014 18:49:10 GMT
Even so the figures are highly improbable (just like MORI's recent Scotland-only poll)
|
|
