In the American presidential election of 1936, a well known magazine, The Literary Digest, published a “scientific poll” showing that Alf Landon would defeat Franklin Roosevelt. As election night proved dramatically otherwise, an embarrassed pollster had to admit that a survey that was based on subscriber lists and telephone directories of people who had a private phone line might indeed produce a skewed result.

A young Canadian graduate student at the London School of Economics – who happened to be my father – did his PhD on the subject “Public Opinion and its Measurement.” His conclusion was that any reliable survey had to be marked by two things, a statistically reliable response rate and a public that was well informed, responsive and ready to decide.

Saul Rae teamed up with George Gallup to write The Pulse of Democracy and soon after joined the Department of External Affairs. One of his first assignments was to work on early government polling on public attitudes to the war.

In its early days, the Gallup organization did door to door surveys, and in time, as landline phones became universal, moved to phone surveys that were more or less reliable.

In the early sixties, the NDP pioneered the door-to-door canvass, returning three or four times, in order to identify a reliable pool of voters to be pulled on election day. Other parties with a large volunteer base borrowed the same technique.

As in so many parts of life, we now live in a different world. Canadian pollsters are reluctant to admit that “response rates” on the phone are way down, that the landline is no longer universal, that people aren’t at home, and that finding a reliable statistical base is in fact notoriously difficult and hard to measure.

Experienced political canvassers in all parties will also tell you that volunteers are not as numerous, that people are either not at home or don’t answer the door, and that it is more difficult to get the “I.D.’s” that are the key to a successful election day organization. The Harper Conservatives’ “machine” relies heavily on robo-calling and building a base of reliably identifiable supporters, a system that has now been imitated by the other parties. Barack Obama’s success was based on an unprecedented recruitment of volunteers in an effort to overcome voter apathy.

The polls in the Ontario election are bouncing around like crazy, and while they all claim great scientific accuracy, within a few points, nineteen times out of twenty, the reality is there is more guesswork and questionable assumptions than anyone is prepared to admit.

We cling to the “certainty” of polling, but in reality, nothing is certain. A growing group of citizens don’t vote, almost as a matter of principle. The number of “don’t know, don’t care” is on the rise, and the length of time that many are taking to come to judgment is growing as well. And political parties themselves are trying to figure out how to mobilize à la Obama – and realizing how tough a challenge it really is.

Chatting with my Dad about why he left the world of polling as a young man of 25, he said simply, “it was the war, there were more important things to worry about. Anyway, turning heads is more important than counting heads.”

Wise words. And always a challenge.

**A special to The Globe and Mail.