Statistics Don't Lie

Statistics don't lie. They're cut and dried, and come from calculators or excel formulations. People, however, lie and make mistakes quite often. I have 2 examples of lying: market research on the Hispanic community and the Wal-Mart class action lawsuit.

One of my professors was talking about PRIZM research. This company conducts demographic and psychographic research for the United States and creates categories for different types of people. (Click here: http://www.claritas.com/MyBestSegments/Default.jsp?ID=20&SubID=&pageName=ZIP%2BCode%2BLook-up , put in your zip code and see where they place you, it's hilarious.) This company did some research on Hispanic Americans' consumption habits and found that they love BET, Jet Magazine, The New York Times, and canned chili, but don't like pork sausage. Interesting huh? Sounds like some companies need to adjust their marketing strategies.

Here's the problem.

Hispanics are most prominent in Los Angeles, Miami, Texas and New York. What Hispanic person in Los Angeles, Miami or Texas reads the New York Times?? What countries do Hispanic New Yorkers come from? What ethnicities, outside of Hispanic, do New York Hispanics belong to? The research shows a strong bias toward reporting done in New York, where people of all ethnicities read the NYT, and Hispanic people are more likely to have Dominican or Puerto Rican roots, along with African roots.

Then there's food. Do Mexicans really like chili? The reporting indicates that Hispanics were asked, "Do you eat chili?" to which they answered an emphatic, "Yes!" because they eat chiles in almost anything they make. And pork sausage? They eat it all the time and call it "chorizo."

Now to the Supreme Court's denial to hear the Wal Mart class action suit, in which 1.5 million women sued the behemoth for gender discrimination. Something like 70% of Wal-Mart's hourly workforce is women, but less than 1/3 of management jobs are held by women. That's statistics!

Almost.

First of all, this compares management jobs to hourly jobs, which are not the same thing.

Second, the women in the class action suit were not part of the same "class" of worker, as the Court pointed out. Some are looking for management jobs, while others are grandmas trying to make a couple extra bucks, and other are in between.

Third, similarly to the last point, women often have different orientations toward work than men. True, gender is a protected class as it should be, but many women choose to place family as a higher priority than work. The kind of women who work at Wal-Mart likely over-index when it comes to this. Women who want to have management jobs should be able to reach them if they have the skills and put in the work. (And I do believe that Wal-Mart discriminated.) However, to use statistics to say that men are overrepresented in management jobs to prove systematic discrimination is systematically flawed. What if these hourly workers simply weren't trying to get promotions? Or continually requested schedule changes in favor of family obligation? That's fine if they can keep their jobs and do this, but it doesn't make them promotion material. And it shouldn't make them part of a discriminated-against class worthy of a large judgment.

The point is that the specifying kind of women who work in hourly Wal-Mart jobs matters, just like the location of Hispanic market research matters, and specifying food names matters. Otherwise statistics can't uncover actual lies.