Chance News 112

From ChanceWiki
Jump to: navigation, search

August 21, 2017 to December 31, 2017

Quotations

"To gain control of the State Assembly, the authors estimate, Wisconsin Democrats would have to beat Republicans by 8 to 10 points, a margin rarely achieved in statewide elections by either party in this evenly split state. As a mathematician, I’m impressed. As a Wisconsin voter, I feel a little ill."

-- Jordan Ellenberg, in: How computers turned gerrymandering into a science
Sunday Review, New York Times, 6 October 2017

"There is a widespread — and fundamentally incorrect — belief that a probability of 60 percent or 75 percent or 90 percent means it’s gonna happen. It does not, any more than a die is broken if it rolls a one. Sometimes, a candidate with a 62 percent chance of winning well into election night (as Jones had for a time last night) will lose. Otherwise, the percentage wouldn’t be merely 62."

--David Leonhardt, in: Roy Moore Is staying home, New York Times, 13 December 2017

Sugggested by Mike Olinick

Forsooth

"Of the 36 applicants that were interviewed, 20 were ultimately promoted... . Among the promoted individuals, 62 percent were female and 38 percent were male."

in: “Woefully thin statistics” doom adverse impact claim, JDSUPRA.com, 24 August 2017

The graphic below appears in: New study reveals bartenders, casino workers most likely to get divorced, Inverse Culture, 5 September 2017

Divorce scatter.png


Note: The scatterplot was originally created by FlowingData, where the relationship is correctly described as "downward slopey." The caption above is from the Inverse Culture article (emphasis added).


Facebook advertises that it "can reach 41 million 18 to 24-year-olds in the United States and 60 million 25- to 34-year-olds." But according to the U.S. Census, there are only 31 million and 45 million total people in those two demographic groups. Details are in this New York Times article.

Suggested by Steve Simon

Lecture on football probability

Margaret Cibes sent a link to the following YouTube video:

John Urschel-NFL Math Whiz: Real Sports Full Segment (HBO)

It features John Urschel, an offensive for the NFL's Baltimore Ravens, who is also studying applied mathematics at MIT. The video begins with John at a chalkboard using a decision tree to analyze a one-point vs. two-point conversion late in a football game.

John is already a published mathematician, as described in this 2016 article from the Notices of the AMS.

Redefining statistical significance

Scholars take aim at false positives in research
by Thomas Gaulkin, UChicagoNews, 1 September 2017

University Chicago economist John List is one of 72 collaborators whose commentary, Redfine statistical significance, was just published in Nature Human Behavior. The subtitle reads, "We propose to change the default P-value threshold for statistical significance from 0.05 to 0.005 for claims of new discoveries."

See also:

How bad/good were the predictions about Hurricane Irma

Irma Shifting Forecast: It's All a Matter of Probability.
by John Schwartz, The New York Times, September 10, 2017.

How surprising is it that Irma is heading up the "wrong" coast of Florida? Well, it changed the plans of one expert.

Brian McNoldy, a researcher at the University of Miama and respected blogger on tropical storms and hurricanes, decided on Thursday to evacuate from South Florida with friends and his two dogs and drive to the Tampa area.

Dr. McNoldy ended up travelling back to Miami once the forecast changed, showing Irma smashing into the west coast of Florida rather than hitting Miami dead on.

Was this a failure of the statistical model? Florida is such a skinny state that Dr. McNoldy admits that predicting where any hurricane will hit is problematic.

"A hundred miles is the difference between the east coast and the west coast-but a hundred miles in a three-day forecast is really good."

More accurate forecasts are unlikely to come anytime soon. The problem is that people don't understand the depiction of uncertainty in the graphic models. The focus is on the line that runs down the middle and they ignore the variation about that line, the cone of probability.

J. Marchall Shepherd, an atmospheric scientist at the University of Georgia exlained the fallacy in a Facebook post. "Anywhere in that cone is a possibility," Dr. Shepherd wrote, "and it has always been a challenge communicating what the cone 'means' versus what people 'think" it means."

Questions

1. There are different maps of the predicted paths of Irma here. One shows a cone of probability and another shows 52 separate predictions of the hurricane's path. Which one better depicts the uncertainty of the prediction?

2. Dr. Shepherd mentions in a Facebook post referenced by the New York Times article that people often confuse the concept of "percent probability of rain." What are some of the potential misinterpretations of this phrase?

3. Hurricane Harvey reintroduced us to the term "500 year flood." What are some of the potential misinterpretations of this phrase.

Submitted by Steve Simon

Hurricane Harvey and the 500-year flood

The “500-year” flood, explained: why Houston was so underprepared for Hurricane Harvey
by Dara Lind, Vox, 28 August 2017

The subtitle notes that Hurricane Harvey has produced Houston's third “500-year” flood in the past three years.

Harvey’s test: Businesses struggle with flawed insurance as floods multiply
by Ruth Simon and Cameron McWhirter, Wall Street Journal, 29 August 2017

"Twenty storms causing a billion dollars or more in damage have taken place since 2010, not including Hurricane Harvey, compared with nine billion-dollar floods in the full decade of the 1980s, according to inflation-adjusted estimates from the National Oceanic and Atmospheric Administration."

Brigitte Baldi sent the following article from FiveThirtyEight to the StatEd e-mail list:

It’s time To ditch the concept of ‘100-year floods’

What's going on in this graph?

Announcing a new monthly feature: What’s going on in this graph?
by Michael Gonchar and Katherine Schulten, New York Times, 6 September 2017

This article announces a new feature on data visualization with a view towards enriching classroom discussions of statistics. The project is a partnership with the American Statistical Association. The article says:

Teachers tell us these data visualizations are rich texts for classrooms across the curriculum, not just in the math or statistics class. Whether in a literature class analyzing Jane Austen’s language, a science class considering climate data, or a civics class studying gerrymandering, teaching students how to read, interpret and question graphs, maps and charts is a key 21st-century skill.

The inaugural post appeared on September 19 and features a rainfall map that was featured in coverage of Hurrican Harvey. Students are asked to consider the following questions

  • What do you notice?
  • What do you wonder?
  • What’s going on in this graph?

They are also invited to participate in an online discussion facilitated by the ASA.

(On a related note, the US Geological Survey's animated data visualization entitled Hurricane Harvey's Water Footprint.)

Monty Hall has died

Mike Olinick sent a link to the following:

Monty Hall, co-creator and host of ‘Let’s Make a Deal,’ dies at 96
by Dennis Hevesi, New York Times, 30 September 2017

Statisticians know Monty's name from a notorious conditional probability problem. As described in the article:

“Let’s Make a Deal” became such a pop-culture phenomenon that it gave birth to a well-known brain-twister in probability, called “the Monty Hall Problem". This thought experiment involves three doors, two goats and a coveted prize and leads to a counterintuitive solution.

The problem is related to Bertrand's box paradox, which dates from the 19th century, but it exploded into public awareness as the Game Show Problem in a series of columns from the 'Ask Marilyn' feature in Parade Magazine in 1990-1991.

A New Yorker cartoon inspired by the show was described by Laurie Snell in the inaugural installment of the Chance Newsletter in September 1992!

Gerrymandering

The new front in the gerrymandering wars: Democracy vs. math
by Emily Bazelton, New York Times, 29 August 2017

How computers turned gerrymandering into a science
by Jordan Ellenberg, Sunday Review, New York Times, 6 October 2017

These two articles describes how analytical models have been used with great effect to maximize the partisan advantages of redistricting. Ellenberg cites the following research article

Evaluating Partisan Gerrymandering in Wisconsin
by Gregory Herschlaga, Robert Raviera, and Jonathan C. Mattingly

Those authors analyzed a collection of potential Wisconsin electoral maps to demonstrate how extreme actual redistricting plan was in favor of Republicans.

Interpreting online product reviews

Don’t base all your Amazon purchases on the number of reviews a product has
by Corinne Purtill, Quartz, 23 August 2017

See also: The Love of Large Numbers: A Popularity Bias in Consumer Choice

Campus speech survey

Here are two views of a controversial survey on student view about free speech.

Dubious graphic on guns

Spurious chart, data on N.R.A. spending mislead in gun debate
by Linda Qiu, New York Times, 3 October 2017

In the wake of the mass shooting in Las Vegas, there were naturally calls for more stringent gun controls. Pro-gun arguments revived a graphic from the American Enterprise Institute, purporting to show that the increased number of guns has led to less gun violence over the last two decades.

International comparisons

What explains U.S. mass shootings? International comparisons suggest an answer
by Max Fisher and Josh Keller , New York Times, 7 November 2017

In a section entitled "Factors that don't correlate" we read

If mental health made the difference, then data would show that Americans have more mental health problems than do people in other countries with fewer mass shootings. But the mental health care spending rate in the United States, the number of mental health professionals per capita and the rate of severe mental disorders are all in line with those of other wealthy countries.

Statistic of the year

Statistic of the Year 2017: Winners announced
by "StatsLife" web news editor, Royal Statistical Society, 13 December 2017

The post announces that "We are delighted to announce our first ever UK Statistic of the Year and International Statistic of the Year, a new initiative that celebrates how statistics can help us better understand the world around us."

  • The UK statistic is 0.1%, representing the percentage of land in the UK that is densely built upon.
  • The International Statistic is 69, the average number of Americans killed annually by lawnmowers---compared to the 2 deaths on average attributed by Islamic jihadists.

Nominations are solicited for the 2018 contest.