Tuesday, April 28, 2009

87.3% of the time, you trust numbers


Why do people believe every number they see? Tonight social media guru David Armano rebroadcast a tweet claiming Razorfish generated $18.8 million in incremental revenue for H&R Block. It's possible; H&R Block takes in about $4 billion annually, so a 0.47% lift from online chatter could happen.

So we tweeted David back, asking how agencies such as Razorfish measure social media results, and he responded he didn't know. Meanwhile thousands of David's friends were forwarding the stat around to others online. This is not to poke at David, who is a truly brilliant and generous agency mind and has illustrated the nuances of social networks in ways that might make Edward Tufte blush. Rather, it points out the human fallacy of numbers.

We like numbers. 50% off. Buy 1, get 1 free. Perfect 800 on SAT. You're a 10. Numbers make us feel good, and if tipped just so -- say, $3.99 instead of $4.00 on a mocha latte -- we rush to respond. Numbers can even be fake and we take action. Look at the cover of Oprah or Men's Health or other quasi-entertainment magazines and you'll see lots of numbers -- 99 ways to lose weight, 7 steps to better sex, 3 new brownie recipes. Are 7 steps to sex too much or too little? Doesn't matter. We'll buy the magazine, read about it, and then lose weight while eating brownies.

Numbers lie. Yet numbers illuminate the truth about our inner needs. We yearn to predict the future exactly and if you tell us you can, we'll believe you -- at least 87.3% of the time.

Update: David Deal, VP of marketing at Razorfish, wrote us to note the $18.8 million figure was wrong -- an error by a panelist who pulled unrelated data and attributed it by mistake to H&R Block. Thanks, David, for the update.

6 comments:

F. Andy Seidl said...

Numbers lie because so many people let them. John Allen Paulos' "Innumeracy" is a fast, fun read that should be required reading to qualify for... well, adulthood. And, its already sold over 2.2 billion copies! -- more or (a lot) less

Maximo said...

Great post, Ben. Never trust numbers---the more detailed the number the more dangerous unless you're in a hospital.

Andy Jukes said...

I agree 110%!

Kelly said...

I make a living out of numbers (and also the qual stuff) and I know how dangerous numbers can be in the wrong hands, and also how people abuse them. I hate number abuse. I am going to print that on a t-shirt.

sk said...

I agree that it's easier to deceive people with a fake stat or two, but I'm not sure it's a big mystery why. As Ian Hacking, among others have shown, stats have long (150 years or so) been viewed as the most valid form of evidence in our culture, providing the decisive data point for the majority of business, policy and judicial decisions. And that's not necessarily a bad thing. (They aren't all fake) It's true that not all objects of inquiry can be turned into algorhythms. (art, eg) And then it's best to stick with a good description, or what we still, i think, call a story.

safemeds said...

I think that it is social media is huge. It is quite simple to realize that because in the increment of the companies working in this kind of media.