Tuesday, May 29, 2012

Lies, damned lies and Greek opinion polls

Repeat after me: You have no choice

Yesterday the world's financial markets rose like a helium balloon on the news that weekend opinion polls in Greece showed that the conservative, pro-memorandum party, New Democracy was ahead. Even the Athens Stock Exchange got a respite from a nosedive that has seen its index fall to figures last seen in the 1990s.

There are many possible explanations for New Democracy's reversal of fortunes. Antonis Samaras has been building a conservative coalition and perhaps many voters who would have voted for smaller right wing parties are now transferring their votes to him as the only viable centre right alternative to SYRIZA.

On the other hand the constant barrage of predictions on TV about possible exit from the Eurozone has frightened older voters who fear their life savings will evaporate should the drachma be reintroduced and so perhaps signals a swing back to more mainstream politics.

Yet, according to an insightful analysis there may be another more concrete reason for the sudden challenge to the popularity of Radical Left Coalition (SYRIZA). In the opinion of Greek blogger, Alepouda there is a systematic bias in polls conducted in Greece with the upshot that means that estimates of SYRIZA's share of the vote are wildly off. The most inaccurate being a GPO poll carried out for the current affairs programme, Anatropes on MEGA TV on the 25th April, just weeks before the 6th May elections. According to the poll SYRIZA was predicted to get 6.2% whilst the final election result was 16.78%. Strangely, the results for both PASOK and New Democracy were less than 2% off.

Nor is this anomaly limited to GPO, other pollsters have revealed similar deviations in their estimates of SYRIZA's electoral strength. Indeed, according to Alepouda's infographics the average result opinion poll researchers claimed for SYRIZA was 9.1%, more than 7 percentage points lower than their final figure. In the case of all other political parties the difference was less than 1%.

How to explain such  amazing accuracy in the case of every other party apart from SYRIZA? Statistically, there seems to be no logical explanation for such a large error. On the other hand, more sceptically minded commentators might be left with the impression that the media war being waged upon Greek parties opposed to austerity measures is not limited to the mainstream press but has extended to those companies that rely on such media outlets for their very survival.


What Should I Do said...
This comment has been removed by a blog administrator.
Anonymous said...

You know what? Shortly after the last elections, when SYRIZA appeared to be topping all polls, I predicted that, other things being equal, the trend would be reversed as we neared the next elections, and New Democracy would lead polls. I'm neither psychic, nor a pollster, nor an analyst.

There's a simpler explanation than the complex scenarios of conspiracy theories against SYRIZA: so far, that party's been all talk and no substance. Its former platform was childishly naive, giving ready fodder to critics and sceptics. Its current platform is still reminiscent of the 'and there'll be peace on earth' kind of hopelessly wishful thinking.

Of course, no party has control over freak events and statements (see Lagarde's recent ones on Greece vs Nigeria and Merkel's earlier remarks on a Euro referendum), which can turn a tide and change a trend in unforeseen ways. But every party has control over its own platform. If SYRIZA's final version of its political programme, to be announced at the end of this week, appears to be based on realistic solutions and measures, the party will be much better equipped to win debates and votes. Remember that, 'traditional' SYRIZA voters have a reasonably high degree of leftist education: by and large, they're neither religious nor blinkered in the KKE manner. It's harder to sell empty promises to people who won't buy anything on faith, are not indoctrinated to accept unquestioningly the party line, and possibly judge those promises not exclusively on the basis of party allegiance but also of rationality.

To expose fallacious attacks, fend off critics, and win voters, SYRIZA, of all parties, needs to fill those promises with convincing arguments. The silliness of the 'Yia Ola Ftaiei O SYRIZA' campaign and the supposed allure of Tsipras's looks are of very limited value when you're trying to win voters who are not party disciples - at least that's how I see it.

-- AG

teacher dude said...

Interesting, but none of that explains the strange ability of pollsters to predict every party's vote, except SYRIZA's.

Anthony Z said...

So presumably if all the other party shares were accurate and SYRIZA's 10 points low, the uplift in SYRIZA's vote must have come from undecided voters (mathematically - there's nowhere else for that 10% to appear from if other shares are constant).

Is it plausible that people making last-minute decisions skewed heavily towards SYRIZA as a "sod the lot of you" vote to the traditional parties. Seems plausible to me. In which case, there might well be some unwinding of that vote in the second election if people feel like they've had their protest and now it's time to go back to the mainstream. We'll find out on 18 June.

keepquestioning said...

I don't know what will happen on 17 June. But I know that public opinion polls are generally skewed, both at European level (see Eurobarometers phrasing) and at national level. Regarding Greece, it is a common practice among the polling companies to try to form the questions in a way that the authority/ company that delegates the poll gets the results they want. In this respect, and from the safe assumption that Ned Democracy and PASOK have closer ties to the media and business world, it would surprise me to see even one poll showing SYRIZA on the lead.

Anonymous said...

what would they gain from under estimation o sirza rate? e.g. it may lead voters who didn't consider sirza, to vote for her, in order to keep an healthy opposition. under valuation is a well known pool tactics. my assumption is that the pooling models are not up for the tectonic class movements. while they built there polling model based on income and potions, the sharp class division destroyed that. large swat of the middle class joined the working class etc. the old sampling changed therefore the marginal to major error.

Anonymous said...

nice posting.. thanks for sharing.