Faith in numbers

Ever since I skipped most of my summer semester statistics class in college, I've been convinced that Mark Twain understated his classification of truth-tellers ("There are liers, damn liers and then there are statisticians."). For many months I've read with great interest the various election polls published by news networks and independent polling organizations, e.g., Rasmussen.

For the life of me I can't understand how these organizations are capable of publishing such wildly different results. The latest USAToday poll has Bush up by 8%! Rasmussen has the race at a literal dead heat with each candidate receiving a nod from 47% of those polled. I heard one pundit say that polling is what we do instead of thinking -- enough said.

I decided months ago that I would monitor Rasmussen in an effort to spot trends. I think this decision has served me well. Kerry trended upward after the Democratic convention and the first debate with Bush doing the same after his party's convention. Rasmussen provides additional insight by way of state-by-state polls that are updated regularly. Based on what I've chosen as my relative statistical standard, I call the race a dead heat that will only get "deader" as we get closer to the election.

In the final analysis, I think this statistical confusion favors Kerry (see previous post for details). We'll know if I'm right in 15 days.