Marshall Brain's Blog

---

Polls

On Friday, January 16, CNN published its polling results for the Iowa caucuses in this article. Here's what the polls indicated:

• Kerry 21.6 percent
• Dean 20.9 percent
• Gephardt 20.9 percent
• Edwards 17.1 percent

When the results were tallied on Monday, January 19, they were:

• Kerry 38 percent
• Edwards 32 percent
• Dean 18 percent
• Gephardt 11 percent

The polls were not even close to predicting reality.

It is interesting to speculate on why the polls could be that far off just one weekend before the event. Some possibilities:

• Even with all the modern statistical science backing them up, polls really are that inaccurate.
• People intentionally lie to pollsters.
• Some new peice of information about one or more of the candidates came up over the weekend and drastically altered the landscape (seems unlikely since it was never reported)
• People wait until the last minute to make up their minds, and/or a whole lot of people changed their minds over the weekend.
• The caucus process allows people to be persuaded into changing their votes once they arrive at the caucus.

Since polls can be so inaccurate, it makes you wonder about all the news stories that use polls to bolster an argument, and all the politicians who listen to those polls. For example, a news story might say, "Polls indicate that 50% of Americans strongly support legalizing assault rifles (or whatever)." Really? How accurate is that number unless you really put it to a vote?