Chance News 115

September 1, 2018 to December 31, 2018

Quotations

"The Supreme Court does not compute. Or at least some of its members would rather not. The justices, the most powerful jurists in the land, seem to have a reluctance — even an allergy — to taking math and statistics seriously."

--Oliver Roeder, in The Supreme Court is allergic to math, FiveThirtyEight blog, 17 October 2017

Sent to the Isolated Statisticians list by Jason Wilson

Some basketball forsooths?

UNC’s generous 3-point defense is a cost of its championship style
by Quintin Schwab, Tar Heel Blog, 4 September 2018

Popular sports reporting often includes dubious statistical reasoning. The present story reports that the North Carolina Tar Heels have allowed opponents to shoot 357 of 940 (38.0% )on 3-point field goal attempts, "the worst figures of any ACC team in at least the last nine years (122 individual team seasons)."

Here are two statistical passages from the post:

• "The stat is understood in the analytics world as one of the more “fluky” sports data points — Ken Pomeroy famously declared that defending a three-point shot doesn’t matter so much as limiting attempts, and with a shot that goes in, on average, about 35% of the time, there’s a ton of variability just by the low hit percentage of the endeavor."

• Gives a list of ACC conference teams showing, for the last nine seasons, the average number of successful 3-point shots allowed/ average number of attempts (total number of games).

1. Duke: 5.1/16.3 (331 games);

2. Virginia: 5.8/18.0 (305 games);

3. Clemson: 6.0/18.0 (294 games);

4. Louisville: 6.1/19.2 (137 games);

5. North Carolina State: 6.2/18.7 (309 games);

6. Maryland: 6.3/19.2 (168 games);

7. Georgia Tech: 6.3/18.8 (298 games);

8. Miami: 6.4/19.1 (309 games);

9. Pittsburgh: 6.6/19.3 (168 games);

10. Boston College: 6.7/18.8 (292 games);

11. Florida State: 6.9/20.4 (308 games);

12. Wake Forest: 6.9/19.9 (285 games);

13. Virginia Tech: 7.0/21.2 (298 games);

14. Notre Dame: 7.1/20.1 (178 games);

15. Syracuse: 7.4/22.8 (173 games);

16. North Carolina: 7.5/22.0 (337 games).

The blog says, "When the analysis is on a per-game basis, it really highlights how close most of the teams are. In general, an ACC team gives up between six and seven 3-point makes out of about 19 tries per contest. The outliers, or at least the poles, to avoid claiming statistical significance, are, obviously, Duke and North Carolina."

Discussion
For each of these passages, what is the author trying to say? Does the statistical language support the point?

New data search tool

Eric Reyes sent this link to the Isolated Statisticians e-mail list:

Google unveils search engine for open data

by Davide Castelvecchi, Nature News, 5 September 2018

More on alcohol risks

Remember when a glass of wine a day was good for you? Here's why that changed.
Popular Science, 10 September 2018

In Alcohol risks from the last installment of Chance News 114, we looked at a recent study on alcohol risks which announced the surprising finding that no level of drinking was safe. This story describes how the thinking on moderate drinking has evolved.

1 in 20 deaths globally is a result of alcohol use
by Nina Avramova, CNN, 21 September 2018

This according to Global status report on alcohol and health 2018 from the World Health Organization (WHO). in the news release accompanying the report, we read

Of all deaths attributable to alcohol, 28% were due to injuries, such as those from traffic crashes, self-harm and interpersonal violence; 21% due to digestive disorders; 19% due to cardiovascular diseases, and the remainder due to infectious diseases, cancers, mental disorders and other health conditions.

What's going on in this graph

Here is the first installment for the fall.

Life expectancy by neighborhood

The Strawberry Capital of the World is the early death capital of the U.S.: lessons from a landmark dataset
by Andrew van Dam, Washington Post, 14 September 2018

DNA forensics

The dangers of DNA testing
New York Times, 21 September 2018

Subtitle: In a new study, 74 out of 108 crime laboratories implicated an innocent person in a hypothetical bank robbery.

How to persuade children to eat more vegetables and fruit

Association of Plate Design With Consumption of Fruits and Vegetables Among Preschool Children
JAMA Pediatrics, 6 August 2018

Abstract: Children in the United States do not consume enough fruits and vegetables. One potential approach to change health behaviors such as fruit and vegetable consumption is to change the decision environment to make certain behavior more likely without altering incentives. For example, placing pictures of fruits and vegetables on lunch trays is associated with an increase in fruit and vegetable consumption among school-aged children. This potentially powerful intervention, however, has not been explored among children in early childhood (3-8 years of age), and dietary behavior during this period can be associated with life-long dietary practices. This study addresses this gap and tests the association between use of plates with fruit and vegetable pictures and consumption of fruits and vegetables among preschool children between 3 and 5 years of age.

Here is a news report on the paper from a Utica NY television station.

Discussion
1) How would you design a study to test the hypothesis that placing pictures of fruits and vegetables on lunch trays is associated with an increase in fruit and vegetable consumption among school-aged children?
2) What is your opinion on the study design as presented in the JAMA paper
3) If you have access to the full paper, comment on the results obtained.

Murder rates and temperature

A rise in murder? Let’s talk about the weather.
by Jeff Asher, New York Times, TheUpshot blog, 21 September 2018

"The correlation between heat and crime suggests the need for more research on shootings in American cities."

Food research and p-hacking

Cornell Food Researcher's Downfall Raises Larger Questions For Science
by Brett Dahlberg, NPR, 26 September 2018