Beware of those citing polls – Ukrainian elections 2012July 18, 2012
The Ukrainian parliamentary election campaigning begins officially in about 10 days time as the 90 day period prior to election day rapidly approaches.
If you live in Ukraine, you would naturally be forgiven for thinking it had started prior to Euro 2012 as far as some candidates are concerned, as I have mentioned before. The face of Natalia Korolevska for example, is now so familiar to me from local television stations, posters, leaflets, small tents containing Korolevska people handing out her political glossary on almost every main street in Odessa, I see her face more than that of my wife.
Anyway, once the official electioneering is underway, polls from various sources will be used, abused, misunderstood or misrepresented by vested interests and media alike either deliberately or through complete ignorance.
As tempted as I am to delay this entry until the official electioneering begins and polls start to be banded about, it is probably wise to write this entry now and have it available to refer back to during the campaigning once it gets into full swing.
The aim of this post is to put some perspective on polls and just how accurate they really are in the event they prima facie pass an academic litmus test – or not.
The first and most obvious issue is that of using selective polling results both by the media supporting side A or B and also those from within parties A or B. Here we must firstly acknowledge that a professional polling company (or poll professionally compiled by any organisation) generates a veritable mountain of information the vast majority of which does not make the media or is highlighted by spokesman for party A or B.
They will naturally highlight the parts of the poll which support their declared position and normally ignore, or at best gloss over, parts that undermine their position.
No doubt a very frustrating thing for those who compile a poll to see snippets being passed off as the whole result. That said, we as the public who maybe influenced or have some form of belief in these snippets have to shoulder some blame. In the vast majority of cases if we are asked a question of a political nature in abstract, we may answer it quite differently as to how we will actually vote after a series of political questions that provide a picture of us as a whole. I may prefer party A on the issue of X, but in general my political leanings are towards party B over a much larger range of issues.
Therefore if a media outlet of a known certain bias towards any particular party is attempting to draw you attention to a specific poll result, it is because others are either contradictory or other parts of that same poll are not so flattering to its declared position.
My point, such as it is, is to look at any particular poll not only in comparison with others, but also in and of itself, as the whole rather than an abstracted and highlighted part.
This brings me to another point. If a certain poll widely touted holds very little comparison to a number of others, then one has to consider it with a degree of caution. The exception is hardly ever the rule and therefore a poll that seems to be the exception may not be a true representation of opinion through either a faulty academic model or a deliberate manipulation of the model upon which it is based or simply a fluke set of results.
Such manipulation can be deliberately caused by polling in cities known to favour party A or B and despite the poll then stating it was conducted in numerous cities around the nation, it was in fact deliberately skewed. Another way is to include a disproportionate number of men or women, people of certain age ranges, a high number of employed verses unemployed etc.
Another reason a poll may seem beyond the normative results of all the others is the manner in which it was conducted. People may react differently to a telephone poll than to a poll in the street or via the Internet. It may also be down to the questions themselves. A slightly different wording or different emphasis on certain words when the question is asked can bring quite different results. The nuance of language can and does effect the objectivity of a poll and the statistics they produce. Something all too often overlooked.
Thus we can unwittingly be trying to compare apples with oranges, a fact normally hidden by party spokespeople or the media.
There is also the issue of the actual size and make-up of the poll. A poll of 1000 normally is interpreted to be plus or minus 3 points. What that actually means when comparing polls is that there could, at the extremes, be an academically sound 6 point gap between two entirely legitimate polls. Something not to be forgotten.
The composition of these polls also matters when claims are made relating to “every region”. As an example, polling company X carries out a poll across Ukraine of 1000 people. Of those 1000 people across the regions, only 30 were from Crimea. That immediately makes any results attributed to Crimea as a region a nonsense as the number of Crimeans polled is so small, the margin for error is so immense it holds no academic or statistical value whatsoever as an indicator to Crimean regional voting. It would become even more worthless if they all come from the same town or the same age group or the same age range or the same ethnic group.
In short, for a poll to have any legitimate standing, at a minimum there should be 1000 people involved and the model upon which the results are based must be correctly weighted. In its most fundamental form, it should have the right number of respondents relating to age, ethnicity, region, gender etc etc in proportion to the country to have a country wide relevance. The numbers involved though cannot be used as a realistic guide to regional results as I have explained above. Regional results would require a survey of 1000 people in that region, also weighted to take account of social composition.
None of this will be brought to the attention of the public by the spokespeople of party A or B and neither will it be explained or brought to the attention of the public by the media who will be too busy backing their horse and trying to hobble the other, to let a small matter of transparency or accuracy get in the way – even if they have to resort to quoting polls that simply have no real worth.
All of this applies to any elections in any nation naturally, but now I have written this, I hope that we can look at whatever polls and political and media lines taken from them, in a much more subjective fashion.