05 September 2008

Shaping political attitudes

The quarterly CRA promotional poll hit the streets today and it's interesting to see, among other things, one rather curious difference between the numbers as reported and the corrected figures.

Search Bond Papers and you'll find plenty of commentary on these quarterly surveys and they way they are misused and misinterpreted by all sorts of commentators in the province.  Last spring, for example, we looked at the discrepancy between the polling numbers and the actual election result a year ago.

As for the misused and misinterpreted, one need only look at the Great Oracle of the Valley, a.k.a. voice of the cabinet minister, which headlines its online story Williams and His Government as Popular as Ever: Survey.

As popular as ever?

Not even close.

Corporate Research Associates likes to report its results as a percentage of decided respondents.  That is, when figuring out the numbers they report, they drop out the undecided people and those who gave no response and then recalculate the percentages using only the rest.

CRAAugust08 The most obvious effect of this approach is that it inflates the number, as you can plainly see in the chart at left. Over the last year, the difference between the reported result for the Provincial Conservatives and the corrected result (as percentage of total respondents reported) has been between 13 to 15 percentage points. 

In the latest poll, the difference is 17 points.

The effect of this inflation is no where near as dramatic for the opposition parties where the difference between one number and the other is only a couple of percentage points.

Bear in mind, of course, that distortion from reporting as percentage of "decideds" is on top of the distortion evident from comparing the poll results and election results last fall.  That variance was potentially upwards of 20 percentage points on its own.

The other distortion can only be seen when you actually take the time to correct the figures. Rather than seeing a political party which is every bit as popular today as it was six months ago - as the Oracle reported -  there's apparently been a fairly steady decline in support for the Provincial Conservative party.

Now the decline is not any sign of impending collapse but stop and think about it for a second.  News reports which state that the government retains its high popularity are strictly accurate:  government remains extremely popular.  But such reports miss the actual result.

The fault here lies not with reporters and editors in the handful of newsrooms covering the province.  How many of them have the time to flick their calculators on and make the adjustments?  How many of them would be able to report adjusted figures - as opposed to simply reporting the expert's results - without running the risk of accusations that they were biased or unqualified to change what they had been handed?  How many have the time in the course of a busy day to find a polling expert who could legitimately provide a different interpretation, again with the risk that such action would be criticized as "biased"?  Since there is no polling readily available to contradict the CRA results, on what basis would they ethically pursue an alternate interpretation in the first place?

What we are left with is  a situation in which reporters relay the information they have.

Consider the impact that this situation may have on public opinion. With everyone reporting huge popularity for the governing party that doesn't seem to vary over time, it's not to hard to imagine people who disagree with the government feeling isolated.  After all, anyone not feeling too favourably disposed to the government party generally or on a single issue and just catching these numbers quickly would think that he or she is merely one in 18 or 20.  In reality, they are almost one in 40, if we allow that the so-called undecideds are not favourably disposed either but, for the most part, they would not know that.

There is a fairly large body of social psychological research since the end of the Second World War on notion of conformity.  That is the idea that individuals will tend to adjust their stated opinions and their behaviour to conform with a real or perceived standard.

And we don't really need to get too deeply into the phenomenon of conformity to pose the notion that CRA polls aren't really measuring public opinion accurately any more.  In fact, there is a good reason to suggest that these polls have become - in effect  - part of an effort to shape public opinion. 

While it was once dismissed, the notion of poll goosing now appears to be generally accepted.  Recent revelations suggest the government employs a tightly managed system of information release both to coincide with polling periods and, in some instances, to bury unfavourable news. Those who have critical opinions report being advised by politicians and senior government officials to keep silent. Some, such members of as the offshore industry association, have been subjected to public attacks.  Others have been threatened with legal action.

The Premier himself comments regularly on his own concern with counter-acting what he calls "counter-spinning negativity".  That is, by his own accounts, he spends considerable energy coping with opinions that differ from the official government position.

We saw a classic example this week of the effort to enforce conformity. An accusation that some members of his own caucus may not support the Anybody But Conservative campaign is met with an e-mail that, while it claims there will be no repercussions for dissent, then demands a statement in writing as to whether "you support the government's position against the Harper government or if you support the Harper government". The e-mail uses the word "team" which suggests, in itself, the notion of suppressing individual views and actions in favour of a conformal position. The e-mail, incidentally, is not the first time, the Premier has emphasized conformity from his caucus.

There are indications of the conformity phenomenon outside the government caucus as well.  Consider the number of people calling open line programs or leaving comments on local news web sites who feel the need to preface their criticism of government with some variation of the phrase "Now I support the Premier as much as anyone, but...". On the face of it, that sort of phrasing suggests a perception of a social norm which must be acknowledged first in order to make acceptable the expression of an opinion. 

None of this is conclusive and the language is deliberately conditional.  That is because the notion of conformity and public opinion in Newfoundland and Labrador would require a far more detailed study than can be offered in this space. Nonetheless, there are  reasons to believe that the only public opinion polls in the province available to the public are inaccurate and that, in a larger sense, they actually serve inadvertently as part of a wider effort to shape public opinion in a sophisticated and integrated fashion.

In the context of the current federal election campaign, an accurate interpretation of events may depend as much as anything else on appreciating the difference between the only poll known to the public and the others available only to some of the political parties.