Sunday, May 31, 2015

Take the A Train

I'm sure the terrible Amtrak derailment near Philadelphia that killed 8 people and injured 200 others captured your attention. Despite House Speaker John Boehner calling the idea “stupid,” there is reason to believe that Congressional budget cuts played a role in the accident by delaying the installation of positive train control, a signal system that would have automatically slowed trains traveling above the speed limit. In Mr. Boehner's simplistic world view, each event has one and only one cause. Therefore, if the accident can be attributed to human error, legislators are absolved of responsibility for cutting spending that would have protected passengers from human error. In sharp contrast to the usual political response to such tragedies, on the day after the accident, the House Appropriations Committee voted to reduce Amtrak's budget by another 15% next year. Highways and airports each receive about 45 times the subsidies that Amtrak does.

Amtrak's problems and our own will be compounded if the American people conclude that passenger trains are unsafe, especially if they decide to travel by car instead. Here are the data on annual number of deaths per mile travelled, supplied by Christopher Ingraham of the Washington Post.


Cass Sunstein, Administrator of the Office of Information and Regulatory Affairs during President Obama's first term, is a lawyer by profession, but his books and articles draw heavily on research in social psychology. In a recent op-ed, he mentions two reasons people might overreact to news of the Amtrak tragedy.
  1. Probability neglect. When deciding on a course of action, a rational decision maker should weight the utility of each alternative by its probability. Sunstein coined the term probability neglect to refer to people's tendency to completely ignore probabilities when making their decisions. As a result, our actions are influenced far too much by low probability risks, such as the chance of a plane crash, and low probabilty benefits, such as winning the Powerball lottery.
Sunstein overstates his case when he claims that probabilities are completely neglected, but there is evidence from psychologist Paul Slovic and others that average citizens do a poor job of estimating the risks posed by various activities and technologies, when the judgments of experts are used as the criterion of accuracy. An example is nuclear power, which is judged to be much more dangerous by non-experts than experts. When experts make risk assessments, their estimates are largely determined by annual numbers of fatalities, while lay people's judgments are influenced by other characteristics such as catastrophic potential.
  1. Availability heuristic. Sunstein's second reason that people might overreact to the Amtrak accident helps to explain some of our poor risk judgments. A judgmental heuristic is a simple rule that people use to make decisions quickly and without much thought. The availability heuristic, first identified by Amos Tversky and Daniel Kahneman, states that the frequency of some object or event is judged on the basis of how easily it can be brought to mind. For example, availability is influenced by proximity in space and time; that is, we are more likely to recall events that occurred nearby or recently. Please check out this brief explanation of the concept.

As noted in the video, one factor determining availability is the salience or emotional impact of an event. When asked to estimate the frequency of various causes of death, we overestimate those that are more dramatic, such as homicides and accidents, while underestimating more mundane causes, such as heart disease or diabetes. A followup study showed parallel differences in the amounts of attention given to these causes of death by the news media. Since the media usually devote more attention to infrequent events, such as plane and train wrecks, than ordinary, everday occurrences—the “man bites dog” phenomenon—they contribute to our tendency to overestimate the probability of rare events and underestimate more common ones.

My intent is not to criticize media coverage of events like the Amtrak accident. In fact, my first impulse was to include a picture of the accident, thereby making myself part of the problem. Our task, as critical readers, is to remind ourselves to place media publicity in its larger statistical context, and to start being less afraid of Ebola, for example, and more afraid of obesity.

You may also be interested in reading:

Death By Anecdote, Part 1

Good Studies Go to the Back of the Bus

Monday, May 25, 2015

Unsafe Exposure

A new study has found unsafe levels of polycyclic aromatic hydrocarbons (PAHs), linked to cancer and respiratory illness, near fracking sites in Carroll County, Ohio. The study, published in Environmental Science and Technology, was conducted by a research team headed by Drs. Kim Anderson of Oregon State University and Erin Haynes of the University of Cincinnati. The first author was Blair Paulik, an OSU graduate student. The article is gated, but secondary sources, including an OSU press release, give what appears to be a complete, though non-technical, description of the study.

Carroll County, 70 miles southeast of Cleveland, has 30,000 residents, lies on the Utica Shale formation, and had 421 natural gas wells at the time of the study, February 2014. (It now has 480.) The study was initiated by concerned citizens who contacted Dr. Haynes.  

Air samplers were placed on the properties of 23 volunteers who lived within 3 miles of gas wells. The monitors contained treated materials that absorbed contaminants. They were sealed in airtight bags and shipped to Dr. Anderson's lab for analysis. Volunteers were trained in proper collection and handling of the data.

PAHs are hydrocarbons (compounds of carbon and hydrogen), seven of which, i.e., benzopyrene, are classified by the Environmental Protection Agency as carcinogenic. Prenatal exposure is also associated with lower IQ and childhood asthma. Based on their data, the authors calculated that the lifetime cancer risk for those living closest (.1 miles) to the wells is 2.9 in 10,000. This is almost three times the EPA's acceptable level of 1 in 10,000. Levels of PAHs declined 30% at a distance of three miles—still well above the acceptable limits. Levels were about 10 times higher than a rural Michigan county containing no gas wells.

Researcher Blair Paulik in front of an air sampler
(© Oregon State University)
In googling the study, I found an article in Marcellus Drilling News, a pro-fracking website, calling this a “sham study.” Some of their objections were silly, i.e., that the OSU authors were “a long way from home,” and that the article “sneaks in the c-word.”  (Estimating cancer risk was the point of the study.) I found three possibly substantive criticisms: that there were too few data collection sites, that they were not randomly distributed throughout the county, and that the data were collected by “untrained volunteers.” The third objection is false, but if the samples were carelessly handled, it's likely they would have captured fewer, not more, contaminants. Since neither the gas wells nor the human population are randomly distributed within the county, I doubt the value of including data from locations that are far from either gas wells, people or both (although it certainly would have diluted the findings).

Future studies should randomly select fracking sites and collect more samples at varying distances from the wells, including some more than 3 miles away, to establish the physical boundaries of the contamination.

You might also be interested in reading:



Thursday, May 21, 2015

White People Don't Riot: A Manual of Style for Ambitious Young Journalists

Last week, two motorcycle gangs clashed at a restaurant in Waco, TX. They fought with fists, chains, knives and guns. Nine of them were killed, 18 were injured and 170 were arrested.

What is the proper terminology to use when referring to this incident in the corporate media? The New York Times referred to it variously as a “confrontation,” “fight,” “shootout,” “chaos,” and “problems.” To this, CNN added “melee,” “ruckus,” “fracas,” “brawl,” and “brouhaha.” Under no circumstances, however, is it to be referred to as a “riot,” since this term is reserved for the actions of people of color.

How are the participants in this incident to be called? As Charles Blow pointed out, the Times referred to them as “biker clubs” and “outlaw motorcycle gangs,” terms that have positive, even romantic, connotations to some Americans. However, they are not to be referred to as “thugs,” since this term refers to young black men demonstrating against police brutality. Furthermore, it would be ridiculous to assume that biker gangs are representative of white culture generally, or to speculate that their behavior was influenced by heavy metal music. It is equally inappropriate to question whether their fathers were present during their upbringing.



A similar principle applies to the use of the word “terrorist.” Terrorists are Muslims who plot or engage in violent acts against non-combatants. It follows then that white people engaged in similar actions are, by definition, not terrorists. In fact, when white people do these things, one may question whether they are newsworthy at all. Here is a useful case in point.

© Doggart for Congress
Robert Doggart, an ordained Christian minister and former Congressional candidate from Signal Mountain, TN was recently convicted of plotting—on tape and on the internet—to attack the residents of Islamberg, a small, rural Muslim community near Hancock, NY. To implement his plan, Doggart recruited followers and “battle tested” his M4 rifle. His statements, recorded by an informant, included the following:

Those guys [have] to be killed. Their buildings need to be burnt down. If we can get in there and do that not losing a man, even the better.

Yet Doggart was allowed to plead guilty only to interstate communication of threats, is out on bail, and faces a maximum penalty of five years in jail.

You've probably never heard of this case, and rightly so. Attempting to report such a crime in the corporate media would likely end the career of an aspiring young journalist. Fairness and Accuracy in Reporting did a search of the Nexis data base looking for media coverage of the Doggart case. It was reported in local Tennessee news media, and by a handful of mostly small newspapers in the US, UK and Pakistan. It was not covered by the New York Times, Washington Post, Los Angeles Times, USA Today, NPR or any of the national television networks.

Imagine the saturation coverage the corporate media would have given this incident had it been a Muslim group plotting a similar attack on a small Christian community.


Where did Doggart get the idea to attack Hancock, NY?

As the Daily Beast pointed out, one reason for the lack of coverage is that the FBI chose not to put out a press release about Doggart's arrest. This stands in sharp contrast to the extensive publicity given to cases of entrapment in which an undercover agent is able to persuade some hapless Muslim to agree to participate in an FBI-planned terrorist plot. When deciding what is important, journalists are well-advised to follow the lead of the FBI.

Aspiring reporters may question whether there are any exceptions to the “white people don't riot” rule. There are. The term “riot” can sometimes refer to the behavior of groups of predominantly white people, provided they are demonstrating in favor of a liberal cause. Occupy Wall Street is a recent example. A helpful cue for predicting a "riot" is when the police show up in "riot gear."

You may also be interested in reading:



Tuesday, May 19, 2015

And Then There Were Nones

Last year, I posted the results of the General Social Survey showing that, in 2012, 20% of American adults reported themselves as having “no religion,” and that these folks—referred to by demographers as the “nones”—are increasing by about .6% per year. This month, the Pew Research Group released the results of their 2014 Religious Landscape Study. This survey, undertaken every seven years, is based on the results of a quota sample of over 35,000 adults and has a margin of error of plus or minus .6%. Although Pew refers to their nones as “unaffiliated,” the results are strikingly consistent.

Pew reports that 22.8% of the population were nones in 2014, up 6.7% from 16.1% in 2007. Meanwhile, those calling themselves Christians declined from 78.4% to 70.6%. This amounts to an increase of 19 million nones, for a new total of 56 million. Among all religious groups, the nones are second in size only to evangelical Protestants. A followup question showed that, of this 22.8%, 3.1% said they were atheists, 4% claimed to be agnostic, and 15.8% believed “nothing in particular.” This leaves open the possibility that some of the nones engage in spiritual practices, pray, or believe in one or more gods. There was also a 6% increase in those saying that religion is "not too" or "not at all important" in their lives.


  1. Generational replacement. The most dramatic increases in the percentage of nones occurred among younger adults. However, they increased across all generations, even rising by 2% among people over 70.
  2. Switching religions. Among those who report switching their religious affiliation, the nones showed the largest gains of any group. Eighteen percent of Americans who were raised in a religious faith now claim to be unaffiliated.
Another perspective on this trend comes from the 2015 State of Atheism in America, a study by the Barna Group. It combines the results of four surveys of 23,000 adults conducted in 2011 through 2014. The Barna Group calls their nones the “unchurched,” meaning that they say they haven't attended church in the last six months. This results in a larger group, about 36% of the total. They find that 25% of the unchurched are either atheists or agnostics, a group they call “skeptics.” I find this a more interesting group than folks who believe “nothing in particular,” since they would seem to have given some thought to religion.

The Barna Group identifies five demographic shifts among their skeptics when they are compared to the same group in a 1993 survey they conducted.
  1. They are younger. In fact, 34% of them are between 18 and 30.
  2. They are more educated. Half of them are college graduates. Only 32% of Americans over 25 have graduated from college.
  3. A higher percentage are women. Male skepticism increased too, but not as much as among women.
  4. They are more racially diverse. This is due largely to an increase in skepticism among Hispanics and especially Asians.
  5. They are more regionally dispersed. The Northeast and West continue to have higher percentages of skeptics, but the differences are not as great as in the past.
The last three trends suggest a mainstreaming of skepticism. Rather than being a distinctive subgroup, atheists and agnostics are becoming more broadly representative of the population. One major exception, however, is the continuing underrepresentation of African-Americans.

The Barna Group's stated goal is to reconnect with skeptics and bring them back under the influence of superstition, so they asked skeptics what they thought of Christian churches. They note four common responses, although they don't report the percentages who gave each one.
  1. The church is group of people who share a physical space but are not otherwise connected to one another in meaningful ways.
  2. The church adds little of value to their community.
  3. The church stands for the wrong political policies: war, sexual and physical violence, prevention of gay marriage and reproductive freedom, etc.
  4. Church leaders are not trustworthy.
The third response supports the backlash hypothesis—that young people are leaving the Christian religion because they disagree with its conservative politics.

What can we conclude from all of this? Due to the positive association of religiosity with age, the percentage of nones is likely to continue to increase. This is good news for liberals. In an earlier post, I presented evidence suggesting that the most important causes of religiosity are poverty and lack of education. Barring an increase in the financial desperation of the middle class or a weakening of our system of public education—both very real possibilities—the future of organized religion does not seem very bright.

You may also be interested in reading:

The Revolt of the Nones, Part 1

The Revolt of the Nones, Part 2

The God Squad, Part 1: Religion as Selfish Individualism