Why polls are like horoscopes.

I’m start to think that poll results are a lot like newspaper horoscopes. My Globe and Mail horoscope for this week:

VIRGO (Aug. 24 – Sept. 23):

“You must be even-handed this weekend, both in your personal life and at work. If you favour one side over the other you will create a lot of bad feeling, and at some stage in the future it is sure to come back at you.”

I wish all pollsters were Virgos because the polls are wildly apart and that’s giving me a headache, which is a ‘bad feeling’ too.  Ipsos Reid reports that Stephen Harper’s Conservatives has 43% support and are cruising to victory with a solid majority government. Nanos has the Tories at 39%, which would put them in a minority government situation.  And Ekos  says the Tory support has dropped from 37% in the last election to 34.4% which would result in a significantly reduced minority:

What is clear is that this level of Conservative support would not secure a majority and would steer the results perilously close to a parliament where not only the opposition would control more seats than the ruling party, but the NDP and the Liberals would control more seats than the Conservatives.

ThreeHundredEight, a blog that projects seats based on polling results, points out that there is little in the other polls to support the Ipsos findings of a Conservative majority:

“The Ipsos-Reid poll for Postmedia grabbed the most attention yesterday, as it put the Liberals at an incredibly low 21%, behind the NDP at 24% and the Conservatives at 43%. There is little in other polls to argue that the Conservatives really are that far ahead, and when you compare it to Ipsos’s last poll taken April 5-7 you see that the variation has been within the MOE (margine of error).”

Why the disparity? Polling is a science, isn’t it? Well, apparently not. The good folks at the Simon Fraser University Elections Project explain how, even without any bias, different polling questions can influence the results:

“… variations are the result of quite different methodologies adopted by the polling companies. One difference lies in what specific questions are asked. For example, the Nanos polls simply asks an open-ended question about which party a person is likely to vote for. Ekos asks that question, but then provides a list of parties, including the Green Party. As a result, Ekos usually reports much higher Green support than Nanos.

“A third way of putting the questions is to attach the party leaders’ names to the list of parties: i.e. ‘Stephen Harper’s Conservative Party.’ But being reminded of the leader can evoke both heightened positive and negative feelings, and alter the response rates for a party.

“None of these three approaches are ‘tricks’ – they are valid attempts to probe in different ways how people think when it comes time to cast their ballot. But it is an open question which of these methods produces the most accurate results.

“A further difference among the companies is how aggressively they try to uncover the preferences of voters who initially say that they haven’t made up their minds. If a person answers ‘don’t know’ to the first question, most companies will follow up with a question asking about the party a person may be leaning towards. Different variations of this question may be posed repeatedly. A few companies also use a person’s answers to other questions to statistically predict how they would vote, based on a model of how other voters with those share traits are known to be voting.”

I’m sure the polling companies are working hard and using what they each consider to be sound methodology. For that matter, so do the astrologers I’ve met.

It’s just that — well, they can’t all be right in their predictions, and clearly, they aren’t.

(I’d love to know which of these companies use a person’s answers to ‘predict’ how they will vote and then puts that prediction forward as fact. We don’t see that qualifier in the media very often, do we?)

And since we usually aren’t told what questions were asked, don’t know who is included in the questioning (landlines vs. cellphones), frequently have no idea what the regional breakdown is for the polls that are reported, and are never told how the polling companies interpret the results they get, I think we should read them the same way we read horoscopes: for entertainment value.

I just wish they were presented in the same manner as horoscopes, instead of as  game-changing headlines in our daily papers that might actually influence the way people vote.

Advertisements
This entry was posted in Election 2011 and tagged , , , , , . Bookmark the permalink.

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s