Friday, March 12, 2010

When Did Statisitcs Become Political Science?

It seems fairly logical that when you have a question on what a poll says you ask someone who knows about statistics; and when you have a question about politics, you ask someone who knows about political science. However the press in Canada is increasingly saving time by combining sources; and in a recent example the habit is becoming ridiculous.

Jane Taber in her recent post combines statistics and political science by relying on Frank Graves, president of the polling company EKOS, to offer an opinion of the budget. Jane Taber gives the following quote by Mr.Graves concerning the budget:

“The impact was overwhelmingly neutral. For the [Conservative Party] this is probably a good thing as this was a risky budget. No great dollops of cash/goodies for voters but the public seemed fine with the message of fairly painless restraint.”
The interest is of course the suggestion by Frank Graves that this budget was "risky" for the Conservatives; the reason why this is ridiculous is because this budget clearly wasn't. No commentator, no economist even came close to calling this budget risky, instead it was the opposite, everyone saw this as a "status quo", "stay-the-course", mediocre, safe budget.

The budget was clearly not risky, and though Mr. Graves' suggestion that it was is absurd, the larger abuse of rationality is being increasingly committed by the media. Where the results of polling surveys are not enough to incite a readers' interest, the supposed objective journalist jots down the commentary of polling executives to report the current state of political affairs in Canada. This, however, is erroneous because statisticians aren't political scientists, they aren't political experts of any kind.

And as I have written in a previous post, asking polling experts for political commentary can be dangerous because when they stop calculating statistics, just like anyone, they have a potential for bias.


pollster said...

I may have been unclear in my commentary and I apologize for any confusion. I quite agree that there was nothing particularly risky about the budget. In fact, it was a very cautious and safe budget. My reference to risk referred not to the substance of the budget but to the poliical context in which it occurred.
The risk was heightened for the government because of the linkage of prorogation to the rationale of needing to "consult" and "recalibarate". Moreover, the government's standing in the polls was significantly lower than it had been before prorogation and it could ill afford to suffer any further damage with a poorly received budget. So it was a "safe" budget in response to a risky politcal environment.
On the broader point of whether the pollster should stray from the numbers to offer interpretation , this is clearly a controversial question . Although I have been trained in statistics and probablity theory , I am a sociologist , not a statistician . Furthermore , I don't believe that numbers speak for themselves and it is the responsibility of the researcher to offer his interpretation based on his anaylsis and expertise. This should be rooted in logic and empirical evidence, not ideology or partisan views.
Sometimes I fail badly at that task but I do believe that my responsibilities don't end with submitting the raw data.
Fortunately we have an able audience of critics to point out alternative interpretations or errors. I do hope that empirical evidence , analysis and debate about public opinion has a constructive impact on the democratic process
Frank Graves said...

Mr.Graves, not to exaggerate, but you and your company do provide a necessary service to democracy, and in calling a comment you made absurd I showed undue disrespect and apologize.

If I understand correctly you had meant that the budget was risky not because of its content but because there was so much riding on it. However not to debate the issue but to discuss it, I think perhaps then there was a misplaced adjective or that possibly another adjective could of been used that better described the situation.

I recognize the controversy you speak of, on whether a pollster should offer commentary, but to me in regards to the above post, it was not the pollster I was holding accountable, it was the press. It is the press that I mainly rebuke, as they are supposed to write news, not back-door opinion.

As a polling company president it would be foolish of you not to extend influence and to articulate your work, not to mention offer insight into possible historic trends. I would do the samething. However I do see problems in predictions of voter intent and analyses of subjects outside the purview of surveys.

I will add that I recognize my posts appeared quite confrontational to you, and by your writing of a well thought out reply, you have humbled me.

Scott Ross

pollster said...

Thanks for your reply . I wouldn't have commented if I didn't think you raised valid points. Keep up the good work;
and it's important to hold those with a podium accountable for their statements.