Monday, October 12, 2009

Today the WAP is Far From a Serious Threat to Stelmach, but Tomorrow?

I have been following the media on the class project poll on Albertan's voter intentions now, over two years in advance of a practical reality of an actual election.  Some perspective needs to be put on all this.  First, the results are meaningless in any practical sense of measuing voter intention when voters don't intend to vote when the question is asked.  What the polls does say, is the obvious, Albertans are tiring of what they are seeing as the ineptness of the Stelmach governance approach.

So what!  The media coverage of the Glenmore by election, the pending Wildrose leadership vote and speculation on the outcome add a publicity presence that amplifies voter attention.  But that is a far cry from settling voter intention when there is no election in the offing.  Take the results of such polls as good gossip fodder but not much more. 

This poll, and its results, could use a bit more scrutiny than the MSM has had the time, space or inclination to give it.  I found the results very interesting, especially if you go beyond the sound bite level of analysis.  The methodology in the study says they  took the sample of 1,201.  The admittedly deemed some sample numbers by "statistically weighted where necessary to even better reflect the demographic distribuion of the Alberta population."  This means where the sample size for gender, for example, was not reflective of the population in a region the researchers took the small unrepresentative sample size they had and apparently boosted its relative impact to make it look like it was reflective of the actual gender distribution of 50/50.  Why not do the job right and stay sampling in the region until you got a real gender balance outcome?

Here is what I mean.  If in the north region they had 200 participants but 150 of them were male, they would take the data from the 50 females and enhance the impact of that number say 3 times to equal the male data.  Hardly an accurate take on what may be happening in hte minds of females in the region.   They also don't tell us exactly where they did the weighting, how much they "weight in" and what was the base data results they artificially amplified.  Nor to they tell us the sample size in all the regional results nor the margin or error to be attributed to those results.

I would also like to know  the wording of all the questions, and the order they asked them in and how many people were called but refused to answers.  If they had to make 10,000 calls to get 1200 to participate we have a very high degree of self-selection happening and the randomness is weaker.

I don't offer this criticism only on this poll but on all polls.  The professionals make the same mistakes and the results are misleading and meaningless in the same ways.

Now lets look at those results.   The media reports are focused on "decided voters" but you have to wonder how one becomes a "decided voter" when there is no practical possibility of a pending election. Also the results said 957 of 1115 participants had already made up their minds about the next election.  A stretch at best.  With 14% saying they were "Undecided" and we know 60% of Albertan did not even vote last time it is hard to reconcile the high percentage of "decided" voters to these realities.

Therefore I think the more interesting and reliable numbers are in the All Respondents findings.  The second most popular option there is a combination of Undecideds (14.2%) and Other Parties (7.3), which is larger than the Wildrose support.  The "decided" results show that, for now, the WAP is a parking lot for disgruntled PC supporters.

The regional breakdown is even more interesting, but I have suspicions about its reliability for reasons already stated and it is not clear if the results reported are on All Participants or just "Decided" voters.  I think it is only on the "Decided" voter because the overall bar graph is based on the "Decided" voters but it is far from clear. I also wonder where the "statistacal weighting" was applied, why, how much and on what base data. 

That aside the rural north and south PC support is holding rather well in the 45% range.  WAP is the "threat" but it is not much of a threat at the 24% level.  Stelmach has not much to worry about in his country support if these results are accurate.  But the mainly rural caucus tht leads the Stelmach inner circle are spooked by these results for sure.

The real political story from this poll is in the big cities. We are told constantly in the media and in conversations that Calgary does not like Premier Stelmach. Well compared to Edmonton, they sure do.  He has 38% support in Calgary but only 31% support in Edmonton.  Calgary has decided the political threat to Stelmach should be the WAP at 27%.  In Edmonton disgruntled citizens feel the Liberals should be the threat at 27.5%.  The PCs and Liberals are essentially tied in Edmonton with these numbers being within the margin of error.  The WAP trails the NDP in Edmonton and that is the place in Alberta where change and flux of political fortunes are being played out.  Edmonton, not Calgary, is the political caldrun of a yearning for change based on these numbers.

The WAP is a political force but this poll is precious little proof of its power.  Premier Stelmach is in trouble with Albertans but the drama has just begun and the story is far from being told.  Read these opinion polls like you would poetry.  It is more about the imgination and imagery they induce than the facts they prove.