Why Bush will win easily - bias in polls

Most political opinion polls are conducted by telephone. There has been a growing backlash against telemarketers (do-not-call legislation, etc) and they are generally viewed as an annoyance. Most people, when they answer the phone and hear "I would like you to answer some questions about your political views", will immediately hang up. This in itself is not a problem because the polling agency can just call more and more homes until it receives the required number of responses.

However, a problem can arise due to a "self-selection bias". For example, if 90% of Bush voters hang up on the pollsters and 85% of Kerry voters hang up on them, the polls will show higher numbers for Kerry than the true opinions of the general population (and the results of the eventual election).

Now the question is, why would one group be more likely to talk to the pollsters than the other group? The answer lies in the strength of an individual's political opinion. Suppose each citizen were to rank his/her political opinion on a scale from 0 (indifferent) to 10 (extremely strong). The 0s will not even bother to vote, and the 10s will do everything they possibly can to get their voice heard. The threshold required to induce a person to vote will be much lower than the one required to talk to someone who interrupts your dinner with a phone call. In general (this doesn't have to be true for EVERYONE, just for the population in general) the 9s and 10s will be overrepresented in the polls, and the 1s and 2s will be underrepresented.

Now...who are the 9s and 10s, and who are the 1s and 2s? In general, the Democratic voters seem to have stronger opinions than the Republican voters, due to the anti-Bush sentiment that is prevalent among some groups (and aided by the Michael Moore types). Note again that I am speaking generally...of course some of the 10s will be Kerry voters and some of the 1s will be Bush voters, but the Kerry voters make up a higher proportion of the 9s and 10s than they do of the general population.

I think this phenomenon can be generalized to other elections...for example, in the 2004 Canadian elections, the polls showed the incumbent Liberals neck and neck against the Conservatives. However, the Liberals won the election easily. In general, the voters who want change will be angrier and thus voice their opinions more strongly than the voters who want to keep the status quo. Therefore, the polls should always show a bias AGAINST the party currently in power.

Agree or disagree?