Imagine if the NBA advertised tonight’s matchup between the Cleveland Cavaliers and the Charlotte Hornets this way: “Cleveland to beat Charlotte by four points tonight”.
How many people do you think would show up at Cleveland’s Fast Cash Payday Loans Arena to see Lebron and company beat the Bobkitties, featuring the miraculous stache of Gonzaga? Probably not a lot.
But in another contest, our US presidential election, the media feels quite comfortable telling us “this thing is over, Barack Obama has already won”.
Millions of Americans, who look at Yahoo’s home page every day, are presented with this graphic:
A casual observer (also known as a “voter”) looks at this graphic and says “I guess this thing is over. I’d be wasting my time to go and vote for McCain, plus, it would make me a LOSER. I guess I’ll stay home and watch According to Jim. ” (Note: This will also make you a loser.)
You may remember, back in 2000, George W. Bush’s disputed win in Florida, was originally called by TV networks as a Gore win, which presumably tipped the presidential election to the tree-hugging Tennessee robot. The problem was, voting booths in parts of Florida HADN’T EVEN CLOSED YET, meaning the networks were not just wrong (literally), they were wrong (morally). Again, the message to Republicans was “you might as well stay home, your guy lost.”
Mark Twain was fond of Disraeli’s quote “‘there are three kinds of lies: lies, damned lies, and statistics.'” If we’ve learned anything in presidential politics, it’s that for all the effort put into polling, it’s a snapshot, and nothing more.
Back in my college days, I thought I’d coast past the math requirement, by taking Statistics. How hard could it be? Recently, I came across the following statistics story problem:
“You take a simple random sample of 1000 balls from an urn containing 120,000,000 red and blue balls, and your sample shows 450 red balls and 550 blue balls. Construct a 95% confidence interval for the true proportion of blue balls in the urn.”
“the correct formula is: 95% confidence interval for P = p +/- 1.96 * sqrt( p*(1-p) / n) * FPC”
Seeing this problem brought back the SHEER TERROR I felt in that class…in other tough courses like Philosophy or Biology, at least I felt like I had a chance. But Statistics 125? My main reaction was “what in the hell are they talking about?” I’d flip the pages in the book over and over again, for hours, and get absolutely nothing out of it. I was helpless.
So I respect anyone who understands this stuff. But the numbers don’t mean a thing, if the underlying data is unreliable.