As if we need more evidence.

The statistics community loves to think of our subject as highly practical and relevant to the general population. And this is true.

The average person has a poor grasp of basic statistical thinking, even if he or she has taken one or more statistics courses. This is true, yet many in our community are in denial.

Chapter 1 of **Numbers Rule Your World** deals with the most basic of basics... the use of statistical averages and the variance around the average. If you're designing a first course in statistics, or a course in statistical literacy for non-majors, you'd hope that students would at least pick up this one concept.

***

Apparently, neither the reporter nor the editors at LA Times has taken a basic statistics course, or more likely, has not learned about statistical averages and probability distributions in such a course well enough to apply to real-world data.

The depressing article discusses a tabulation of the "richest" cities in the world, using the metric number of millionaires per resident. New York City is the "richest" in the U.S. boasting 4.63% (389K) millionaires.

Here is how the reporter "enlivens" the boring statistic:

Walk down the street in New York and you're virtually guaranteed to see several millionaires. That's because more than 1 in every 25 New Yorkers is a millionaire, according to a study released Tuesday.

For this to be true, we have to assume that the population of New York is evenly distributed geographically. Further, we have to assume that millionaires are also evenly distributed. Both these assumptions are laughably bad. The statistical average is useless here when you start talking about "walking down every street".

Besides, what's the chance that a millionaire is walking down the street in Manhattan, let alone in Staten Island? Are they more likely found in a cab or limo or private car or helicopter?

The reporter should also have done homework on how the researcher classifies the residence of millionaires. Many of them have multiple homes, most likely in global locations. They may have an address in Manhattan, for example, but what if they spend most of their time in the Hamptons or Bermuda? You can count them as New York residents but surely you won't see them on the streets of New York!

***

When only a minority of students exhibit little to no ability to apply statistical thinking, one could blame the students; but when a majority of those who have taken statistics courses commit the most fundamental errors, one must blame the teaching.

If I live in a town with a guy who has $1B along with 1000 people who have nothing, then we each have $1M. Right?

If I live next door to a guy who makes $100M and I make nothing, then I'm rich: I make $50M a year. Right?

I have to say that particular paragraph in the NYT piece is just writing fluff, not intended to be actual serious thought.

Posted by: jonathan | 07/23/2014 at 09:46 AM

To judge "general statistics an utter failure", we'll have to

eliminate selection biasin the sample of journalists we're looking at.Following blogs like yours and Andrew Gelman's that focus on examples of statistics done wrong (AG: Am I too negative?), it's tempting to conclude that, yes, the general public is bad with statistical thinking.

But then there are also so many people doing stats well. Look at the people Alberto Cairo presents in "the functional art"...

"As if we need more evidence"?

Maybe. But in the form of n=1 examples?

They are valuable and important (and LA-Times should post an excuse and explanation, since they reach their readers while we likely don't), but let's not become too negative ;-)

Regards,

Berry Boessenkool

Posted by: Rclickhandbuch.wordpress.com | 07/23/2014 at 01:26 PM

Berry: I deliberately made this post "negative" because I don't believe the teaching profession has taken up this challenge. What are the most common comments we hear from students who have taken Stats 101? "Boring", "hard", "didn't understand it", "didn't like it", "hated it", "glad it's over", etc. If we document every instance of poor statistical thinking-not to pick on newspapers but also on blogs, on talk shows, in business meetings- the list will be long.

Posted by: junkcharts | 07/23/2014 at 03:25 PM