Now and in the future, we will have unprecedented access to voluminous amounts of data.  The analysis of this data and the conclusions drawn from it will have a major impact on public policy, business, and personal decisions.  The net effect of this could go either way–it can usher in a period of unprecedented efficiency, novelty, and positive decision making or it can precipitate deleterious actions.  Data does not speak for itself.  How we analyze and interpret that data matters a great deal, which puts a premium on statistical literacy for everyone–not just PhDs and policy wonks.

Thompson notes a number of statistical fallacies that many, including members of the media, fall prey to.  Using a single event to prove or disprove a general property is one spectacular one that we see all the time, particularly with large, macro-level events.  Regardless of what side of the climate change debate you are on a single snow storm or record-breaking heat wave does not rise to the level of hypothesis-nullifying or -verifying evidence.

There are oodles of other examples of how our inability to grasp statistics–and the mother of it all, probability–makes us believe stupid things.  Gamblers think their number is more likely to come up this time because it didn’t come up last time.  Political polls are touted by the media even when their samples are laughably skewed.

Take correlation and causation.  The cartoon below nicely illustrates the common fallacy that the correlation of two events is enough to prove that one causes the other:

Correlation and Causation

Bottom line: the importance of statistical literacy will only increase. Statistics will come to permeate our lives, more so than ever before. And if we want to truly take advantage of this we had better learn to speak the language.

Some folks are leveraging newly available data in urban areas as analytical shortcuts:

Ted Egan, chief economist in the San Francisco Controller’s Office, said he could wait six months for California to release the detailed sales-tax data he needs for city revenue projections. But it’s quicker to look at passenger tallies from the station closest to the Union Square shopping district, which generates roughly 10% of the city’s sales-tax revenue. The Bay Area Rapid Transit District releases the data within three days, he said: “Why should I have to wait?”

Mr. Egan is among a growing number of economists and urban planners who scour for economic clues in unconventional urban data—oddball measures of how people are moving, spending and working.

Egan has essentially found an analytical shortcut which allows him to see the world as it is more quickly.  Rather than wait for the actual tax receipts, Egan can look at a related measure and extrapolate from there about tax revenues.  Notice, Egan isn’t interested in why tax revenues and sales activity is low or high, he’s just trying to quickly get a handle on what the amount is.

This is illustrates the utility of correlational analysis and the promise of Big Data and data mining.  The danger, of course, is that many of these new indicators may be untested and more volatile.  On the other hand, new data may actually do a better job of capturing people’s patterns and motivations under certain circumstances and such a preview allows for business and policy makers to adjust more quickly to a rapidly changing environment.  For example:

Mr. Leamer discovered that truckers’ diesel purchases on Interstate Highway 5 from California to Oregon, a major timber-trucking route, are a leading indicator of construction employment in California. Diesel sales on Interstate Highway 80 from Sacramento to Salt Lake City, a trucking route for the San Francisco Bay area’s manufactured goods, can help predict California’s manufacturing employment, he said.

If only he had the diesel-fuel data in the first half of 2008, when major government-issued indicators failed to hint at the U.S. economy’s impending downward spiral. At the time, Mr. Leamer said, UCLA forecasters chose not to announce a recession because GDP was still growing and the Bureau of Labor Statistics was reporting relatively mild job losses.

Bad call. The government later revised the GDP and jobs data downward, and the National Bureau of Economic Research concluded that the recession started in December 2007. The jobs data are unreliable because they are based on sample surveys and don’t adequately capture company openings and closings, Mr. Leamer said in hindsight.

When the UCLA economists reviewed the fuel-purchases data late last year, they saw diesel buying had peaked in mid-2007, indicating that fewer goods were being made and moved across the country in the months after. “Had we been aware of that data in 2008,” Mr. Leamer said, “we would have made a different call.”

In business and government, sometimes quickly knowing “what” is happening is more valuable than waiting to know precisely “why” it is happening.

[Anderegg] says that merely mentioning terms like nerd or geek serves to perpetuate the stereotype. The words are damaging, much like racial epithets, he says, and should be avoided.

Why?

[Because] math, science and computer science, Dr. Anderegg said, are courses that young people too often associate with nerds and geeks. As a result, he added, “they sabotage themselves in these fields, and the nation’s work force is suffering.”

“The best way to combat this,” he said, “is put it to bed,” banishing “nerd” and “geek” to the linguistic dustbin.

I am not sure how much I agree with Anderegg.  On the one hand, I think he has a point that historically certain sets of skills did not confer young students with a great deal of popularity.  In many cases, this pressure to conform and thrive provided students with a disincentive to develop their skills in math and the sciences.  However, I think culturally we’ve seen a shift, especially in the last 20 years.  Being able to create cool applications, websites, etc, is a source of popularity these days (what some have termed the phenomenon digital natives).  Most students today revel in having some degree of basic computer science/coding literacy–those that have exceptional skills reap an even greater return socially.  Furthermore, many people now embrace the term ‘geek’ or ‘nerd’ (I often times use it to describe myself–e.g. ‘data geek’ or ‘analytics nerd’).

I tend to agree with Noah Brier that the bigger issue may be with how we teach and introduce children to computer science and, in particular, coding:

As I’ve mentioned in the past, this is something I believe in wholeheartedly. I think you can start with really little kids and make it fun (as the article points out, it’s important to give them the feeling of “magic” you get when you make a computer do what you told it). What’s more, I think it’s really important that the teaching of code be approached from the creative side: Introductory courses should be part of the art department, not science. This is about letting kids build what they dream of and code should be treated no differently than paint and crayons (of course it takes a bit more time to get the hang of than those other things).

Bottom line is stop focusing instruction on the code and start focusing on the output of that code. Let kids make stuff and they’ll be hooked.

He speaks from experience.

It was not surprising that Greinke won, since his earned run average, 2.16, was the lowest in the American League since 2000. But his decisive margin of victory over Seattle’s Felix Hernandez was a sign that voters overlooked his deficiency in another bedrock statistic: wins.

Greinke tied for seventh in the league in victories, with 16. The lowest previous total for an A.L. starter in a nonstrike season was 18.

To what extent voters are finally starting to incorporate the perspective of sabermetric analysts (e.g. many traditional statistics, like wins, distort the quality of individual performances) is still unclear.  However the vote for Greinke is pretty good evidence that progress is being made.

What’s more exciting for me is that not only was a pitcher voted the award who was dominant in newer, alternative statistical categories, but that he himself incorporates these statistics into his game plan and pitching strategy:

Bannister [Greinke's sabermetric-friendly teammate] said Greinke has learned to adjust his pitching based on the advanced defensive statistics. Because of the size of the outfield at Kauffman Stadium and the strength of the Royals’ outfielders, relative to their infielders, it sometimes made more sense to induce fly balls.

“David DeJesus had our best zone rating,” Bannister said, referring to the Royals’ left fielder. “So a lot of times, Zack would pitch for a fly ball at our park instead of a ground ball, just because the zone rating was better in our outfield and it was a big park.”

To that end, Bannister introduced Greinke to FIP, or Fielding Independent Pitching, the statistic Greinke named Tuesday as his favorite. It is a formula that measures how well a pitcher performed, regardless of his fielders.

Music to my ears.

How can we then decide that a campaign was “better” than another one? We rarely look at a campaign data – partly because the actual metrics data is proprietary and not available to anyone beyond walls of an agency and of their clients…

General, and generally available, feedback mechanisms and benchmarks for success don’t really exist. While it may not have been possible before to know exactly if a TV/print/outdoors/radio campaign influenced particular brand affinity and purchase decisions, digital lets us do things differently.

This means that we don’t have to judge works of others purely on elusive criteria of “creativity”, but on actual data on how this creativity fared with people (what did they do? and what did they do next?).

Read the whole thing.