Thursday, January 29, 2015

Chemi & Giorgi's "The Pay-for-Performance Myth"

UPDATE: The link listed below is currently not working. I've talked to Ariana Giorgi about this, and she is working to get her graph up and running again via Bloomberg. She was kind enough to provide me with a provide me with alternate URLs to the interactive scatter plot as well as a link to the original text of the story. Ariana is doing a lot of interesting work with data visualizations, follow her on Twitter or hit up her website.


This scatter plot (and accompanying news story from Bloomberg News) demonstrates what a non-existent linear relationship looks like. The data plots CEO pay on the x-axis and stock market return for that CEO's organization on the y-axis. I could see where this graph would also be useful in an I/O course in discussions of (wildly unfair) compensation, organizational justice, etc.

If you go to the website, the scatter plot is interactive in that you can cursor over the individual data points in order to see the person who it represents.

Monday, January 26, 2015

Saturday Morning Breakfast Cereal and statistical thinking

Do you follow Saturday Morning Breakfast Cereal on Facebook or Twitter? Zach Weinersmith's hilarious web comic series frequently touches upon science, research methods, data collection, and statistics. Here are some such comics. Good for spiffing up a power point, spiffing up an office door (the first comic adorns mine) or (per this post) testing understanding of statistical concepts. a good example of the availability bias!

Monday, January 19, 2015

Pew Research's "Global views on morality"

Pew Research went around the globe and asked folks in 40 different countries if a variety of different behaviors qualified as "Unacceptable", "Acceptable", or "Not a moral issue". See below for a broad summary of the findings.

Summary of international morality data from Pew

The data on this website is highly can break down the data by a specific behavior, by country, and also look at different regions of the world. This data is a good demonstration of why graphs are useful and engaging when presenting data to an audience.

Here is a summary of the data from Pew. It nicely describes global trends (extramarital affairs are largely viewed as unacceptable, contraception is widely viewed as acceptable).

How you could use this in class.

1) Comparison of different countries and beliefs about what is right, what is wrong. Good for discussions about multiculturalism, social norms, normative behaviors, influence of religion on social norms, etc.
2) Comprehensive information on the survey methods used in the different countries (good for a research methods class and discussion of data collection according to the technology one has available).
3) Non-interactive PDF version of the could have your students input this data and make their own graphs. I did this in class last week. First, they played around with the interactive data, then they created frequency tables and bar graphs of just the "unacceptable" gambling data medians using SPSS.
4) In a psychometrics class, you could discuss the virtues (pun!) of using a 3-item response scale (like Pew did) or if it would have been more appropriate to use a Likert-type scale to understand attitudes. Also, from an interpretation view point, this data is per country...does this accurately represent the number of humans in the world that hold these views?

Monday, January 12, 2015

Emily Oster's "Don't take your vitamins"

My favorite data is data that is both counter intuitive and tests the efficacy of commonly held beliefs. Emily Oster's (writing for 538) presents such data in her investigation of vitamin efficacy. Short version of this article: Data that associates vitamins with health gains is based on crap observational research. More recent and better research throws lots of shade on vitamin usage.

Specific highlights that could make for good class discussion:

-This article explains the flaws in observational research as well as an example of how to do good observational research well (via The Physician's Health Study, with large samples of demographically similar individuals as described in the portion of the article featuring the Vitamin E study). This point provides an example of why controlled, double-blind lab research is the king of all the research.

-This is an accessible example as most of your students took their Flintstones.

-The article also demonstrates The Third Variable Problem. As illustrated in the graph above, taking vitamins also correlates with higher levels of education and lower levels of obesity. Hence, pure observational research that touts the benefits of vitamins fails to address the health benefits associated with education and good general health.

Monday, January 5, 2015

Improper data reporting leads to big EPA fines for Kia/Hyundai

On November 3, 2014, Hyundai and Kia were fined a record-setting $100 million dollar for violating the Clean Air Act. They were fined for cooking their data and misreporting their fuel economy, using the unethical (cherry-picking) techniques described below by representatives for the federal government:
""One was the use of, not the average data from the tests, but the best data. Two, was testing the cars at the temperature where their fuel economy is best. Three -- using the wrong tire sizes; and four, testing them with a tail wind but then not turning around in the other direction and testing them with a head wind. So I think that speaks to the kinds problems that we saw with Hyundai and Kia that resulted in the mismeasurement." Video and quote from Sam Hirsch, acting assistant attorney general. 

 Here is EPA's press release about the fine

How to use in class:
-Hyundai and Kia cherry-picked data, picking out the most flattering data but not the most representative data.
-Here, a company was felled by the simple, modest statistical mean.