Talk:Opinion polling for the January 2015 Greek legislative election

26–28 Dec Marc polling results

I don't understand how the results for the 26–28 December poll from Marc came about. The source shows SYRIZA 28.1% and ND 25.1%, but the table currently shows 31.6% and 28.2% respectively. Rescaling to remove the 9.6% undecided gives me e.g. 31.1% and 27.8%. There should probably be a note on how results are rescaled in order to be comparable across polls. Filipvanlaenen (talk) 10:49, 31 December 2014 (UTC)

There IS already a note next to each pollster's name indicating how results are rescaled. PD: About the specific Marc poll, you are right. In the previous link the data for undecided and abstentionists was not given, so a rough approximation based on the data given in previous Marc polls was done. I'll fix this right now, now that the data is available. Cheers. Impru20 (talk) 11:31, 31 December 2014 (UTC)

Using the rule of three is not very reliable. Opinion polls reflect what people indicate they will vote. Taking undecideds and distributing them as the other people are distributed is not sound practice. It has been found that in the past undecideds tend to vote with unpopular parties (and as there are several this time around) distributing undecideds is giving false information. — Preceding unsigned comment added by 173.52.21.99 (talk) 16:57, 24 January 2015 (UTC)

It may not be "sound practice", but it's what opinion pollsters actually do. Except for Palmos Analysis or Public Issue, which actually do use their own, specific statistical treatments, all other pollsters have been shown to use this rule of three at some point (some of them frequently do it; see Metron Analysis or MRB). And it actually seems pretty consistent with actual election results (such as those of 2012). Btw, it is needed to do that since, otherwise, opinion polls may not be reasonably compared with each other (since most of them do not make the same attributions of the "Undecided/Will not vote/Blank/Invalid" section), meaning that it would be impossible to make a Graphical summary, as a consequence. Cheers. Impru20 (talk) 17:38, 24 January 2015 (UTC)

Opinion pollings

I'm Greek and I'm pretty sure that most of your opinion pollings are not true. — Preceding unsigned comment added by 91.140.72.195 (talk) 16:31, 9 January 2015 (UTC)