Tuesday, October 21, 2014

Safety First

OK, folks, raise your hands. Who's in favor of teenage pregnancy? How about teenage abortion? Childbirth?

Although the rate is declining, each year 600,000 American teenagers become pregnant; 30% will become pregnant at least once before the age of 20. According to the National Campaign to Prevent Teen and Unplanned Pregnancy, in 2010, births by teenage mothers cost the country $9.4 billion in health care, public assistance and lost income due to reduced educational attainment, to say nothing of the long-term public health and social costs to teenage mothers and their children.

Dr. Gina Secura and her colleagues at the Contraceptive CHOICE Project in St. Louis have published a case study of the effects of offering free long-term contraceptives to teenage girls. The participants were 1404 adolescents between 14 and 19 who were enrolled in the program between 2007 and 2011. They were recruited via referral from clinics, community flyers and word of mouth. To participate in the study, they had to be sexually active or say they were planning to be, and not want to become pregnant.

All participants received a standard counseling session in which the risks and benefits of alternative contraceptive methods were reviewed and questions answered. After making a choice, participants were provided with their chosen contraceptive option free of charge. Followup interviews were conducted by telephone at three months, six months and every six months thereafter for the next two or three years (depending on when they enrolled). They received a $10 gift card for each completed survey. The one, two and three year followup rates were 92%, 82% and 75%. Rates of pregnancy, birth and abortion were compared both to their age cohort in the general population and their sexually active age cohort (since enrollment was limited to those who were or planned to be sexually active). Here are the data, reported as the number of pregnancies, births and abortions per 1000 in the population.

CHOICE Project
U. S. Population
Sexually Active Teens
Pregnancy
34.0
57.4
158.5
Birth
19.4
34.4
94.0
Abortion
9.7
14.7
41.5

Comparing the CHOICE participants to sexually active teens in the general population—the more appropriate comparison group—shows that they became pregnant and gave birth at 21% the national average, and had abortions at 23% the national rate. The CHOICE Project reduced unwanted pregnancy more for older (18-19) than younger (14-17) girls, and more for African-American than White participants.

The project was effective primarily because the majority of participants—68.4% of the older and 77.5% of the younger girls—chose long-acting reversible contraceptive (LARC) methods, either the hormonal or non-hormonal IUD or the etonogestral implant, rather than birth control pills, rings, patches or injections. These methods don't require participants to remember to reuse them and have a close to zero failure rate. But fewer than 5% of teenagers in the general population use LARCs. The researchers report that most of their participants had never even heard of them.

The study doesn't separate the effects of counseling, the contraceptive method, or the fact that the contraceptives were free, and it's likely that all these factors played a role. Other limitations of the study include its less than 100% response rate, its reliance on self-report data about pregnancies, the possibility that the followups reminded some participants to use their contraceptives, and the requirement of parental consent for the 14 to 17-year-olds, who may have been at lower risk of pregnancy than those whose parents did not consent. These drawbacks may be balanced by the high percentage of African-American (60%) and poor teenagers in the CHOICE Project, who are at greater risk of unintended pregnancy than the general population.

Rarely do we have a chance to report a study that is so effective in solving an important social problem, and that provides such a vivid reminder of the superiority of prevention over remediation. The cost of these birth control methods varied, with the LARCs being more expensive—$400 to $1000 depending on the method and the individual. To this must be added the costs of recruitment, counseling and followup. All of this is a pittance compared to the medical and social costs of unwanted pregnancy. We have the technology; what we lack is the political will.

In theory, the Affordable Care Act requires insurance companies to cover all forms of contraception with no co-pay, but there is no reason for insurance companies to publicize this option.  And there are other problems:
  • Doctors probably don't follow as good a script as the CHOICE Project counselors, and are not as aggressive in recruitment and followup. In general, doctors get paid less for time spent talking to patients than other activities.
  • Teenagers, who are usually covered by their parents' insurance, may not seek contraceptives because the visit will show up on their parents' statements.
  • According to the National Health Interview Survey, 18.4% of Americans aged 18-64 were uninsured during the first three months of 2014. Girls from uninsured families are probably at greater risk of unwanted pregnancy than those whose families are insured.
The Secura study argues strongly for a single-payer health care system, not only to guarantee that everyone is covered, but also because government can prescribe a uniform set of procedures to be followed by doctors when discussing birth control with teenagers.

Newspaper articles about the Secura study mention anticipated resistance from conservatives who believe that providing teenagers with contraceptives encourages sexual activity. But these articles fail to mention that research doesn't support the claim that education about contraceptives encourages teenagers to have sex. (That would be rude.) In fact, the reverse is true. Comprehensive sex education is associated with delayed or reduced sexual activity.

You may also be interested in reading:



Saturday, October 18, 2014

"Fetch!"

The Willie Horton ad used by George H. W. Bush in his 1988 presidential campaign is considered to be one of the most effective political ads in U.S. history. It's also the prototype for the use of symbolic racism in political campaigns. Brad Ashford is a Democrat running for Congress in Nebraska's 2nd District. The National Republican Campaign Committee is spending $170,000 to run the following TV spot in support of his opponent, incumbent Lee Terry.


The ad is technically correct. Nikko Jenkins was released from prison under a Nebraska law that allows early release for good behavior, and went on to murder four people. Ashford supported this law while a member of the state legislature--as, presumably, a majority of Nebraska politicians did.

To call this an example of "dog whistle politics" may be to credit it with more subtlety than it actually shows.

By the way, the U.S. held 2.23 million people in prisons and jails in 2012, seven times the number in 1972, and by far the highest rate of imprisonment in the world. It will be difficult to reduce these numbers if politicians who vote for prison term reductions are held responsible for any recidivism that results.

You may also be interested in reading:

Dog Whistle Politics

Another Dog Whistle

Voter ID and Race, Part 2

In Part 1, I presented evidence that racially prejudiced individuals are more likely to support voter ID laws. Please read it before continuing.

Of course, public opinion does not automatically translate in social policy. In 2012, Bentele and O'Brien published a study in which they used multiple regression to examine what state-level variables were most strongly associated with both the proposal and passage of voter ID laws. The following variables predicted proposal and passage of restrictive legislation:
  • Republicans controlled both houses of the legislature and the governorship. 83% of voting restrictions were passed by Republican-controlled legislatures.
  • The state had become increasingly competitive in the last presidential election. Note that passage of voter ID laws required both increased competitiveness and Republican control. Those states that had become more competitive but had Democrats in control were less likely to pass voter ID laws.
  • The state had a higher proprtion of African-American and Latino residents, and minority and low income turnout had increased in the last presidential election.
  • There were more allegations of voter fraud. This variable had less impact than the other three. Note also that very few allegations of voter fraud have been substantiated.
Their study is entitled “Jim Crow 2.0.”

What about the individual legislators who vote for voter ID laws? Earlier this year, Mendez and Grose reported an experiment in which 1871 state legislators from 14 states with relatively high Latino populations received an e-mail from an apparent constituent asking whether a driver's license was required in order to vote. The legislators were randomly assigned to one of four treatment groups. The e-mail either came from an Anglo (Jacob Smith) or a Latino (Santiago Rodriguez) man, and was written either in English or Spanish. The dependent measure was whether or not the legislator replied. Here are the results:

Supports Voter ID
Does Not Support Voter ID
Anglo Name/English
45.0%
50.3%
Latino Name/English
27.5%
43.4%
Anglo Name/Spanish
10.1%
12.5%
Latino Name/Spanish
1.1%
11.7%

The results are clearest with the English-language e-mail. Those who supported voter ID were 17.5% more likely to respond to the Anglo constituent, while those who did not support voter ID were 6.9% more likely to respond to the Anglo constituent. Response rates were depressed considerably when the e-mail was in Spanish, but the same pattern was obtained. Supporters of voter ID were more likely to discriminate against Latino constituents.

Did the legislators themselves decide whether or not to answer the e-mail, or was it a staff member? It probably doesn't matter, since legislators hire like-minded staff members.

The authors claim that the experiment shows discriminatory intent on the part of legislators who supported voter ID. It could be argued that Republican legislators cast their vote for partisan reasons, and the fact that they also happened to be prejudiced was coincidental. However, not all Republicans supported voter ID, and further analysis showed that those Republicans who voted in favor of voter ID were more likely to discriminate against the Latino constituent than those Republicans who voted against it. (No Democrats in the entire sample supported voter ID laws.)

I'm not suggesting that prejudice and political partisanship are separate alternative explanations for support for voter suppression, either among legislators or the general public. It is likely that Republicans support these laws both because they harbor racial resentment and because the laws are advantageous to their party. In fact, in the last six years, attitudes toward a variety of political issues, such as health care, have become more strongly correlated with racial attitudes. The Republican Party has gradually evolved into an unabashedly racist party, much like the neo-Nazi parties of Europe.

I woke up yesterday morning to the following headline: “GOP calls for travel ban for West Africa over Ebola.” Will Republican Congresspeople be willing to support adequate financial aid for Ebola-stricken countries, or will they use fiscal austerity as an excuse to throw West African Black people under the bus?

You may also be interested in reading:

Voter ID and Race, Part 1

With the election coming up in two weeks, voter ID laws are back in the news. The Supreme Court has accepted or rejected state voter suppression laws in a seemingly random pattern, without explaining their rationale. In news stories about voter ID, the corporate media typically demonstrate false balancing. They fail to examine Republican claims that voter impersonation is a problem. (It is not.) They report but don't evaluate the Democrats' argument that voter ID laws suppress turnout among key Democratic constituencies, such as the poor, minorities, and college students. (There is evidence to support this claim.) While the media sometimes imply that these laws are politically motivated, new evidence suggests that they may also be motivated by racial prejudice. These studies confirm every bad thing you suspected about voter ID laws.

In a new study, David Wilson and his colleagues at the University of Delaware's Center for Political Communication (CPC) used cognitive priming to test the prejudice hypothesis. Their experiment was embedded in the 2012 Cooperative Congressional Election Survey, managed by computer by YouGov/Polimetrix. They had 1436 US adult respondents, 1100 of whom were White. The White participants were randomly assigned to one of three versions of a question asking them whether they favor or oppose voter ID laws. For one-third of the respondents, the question was accompanied by a photo of a White voter and poll worker. For another third, the voter and poll worker were African-American. The remainder of the participants were not shown any image. The photos and the wording of the question are shown in the results table below.


Support for voter ID laws was quite high. Most importantly, when given the Black prime, a reminder of the fact that African-Americans vote, the White respondents were more in favor of voter ID laws than when given the White prime or no prime at all. The fact that the race of the prime was manipulated in a true experiment suggests that race plays a causal role in attitudes toward voter ID.

These results are consistent with an earlier correlational study by Wilson and Brewer. This was a 2012 telephone survey of 906 adult Americans conducted by the CPC. Particpants were asked their party affiliation (Democrat, Republican or Independent) and their political ideology (liberal, moderate or conservative). White respondents were also given a three item measure of racial resentment similar to the Symbolic Racism Scale:
  • I resent any special considerations that African Americans receive because it's unfair to other Americans.
  • Special considerations for African Americans place me at an unfair disadvantage because I have done nothing to harm them.
  • African Americans bring up race only when they need to make an excuse for their failure.

As indicated in the chart, Republicans and conservatives supported voter ID laws more than Democrats and liberals. There was also a significant correlation between racial resentment and support for voter ID. Racial resentment was higher among Republicans and conservatives than Democrats and liberals, with Independents and moderates in the middle. However, racial resentment still had a significant effect on attitudes toward voter ID laws even when controlling for political partisanship, ideology and several other demographic variables. Interestingly, racial resentment had a greater effect on the attitudes of Democrats and liberals, since Republicans and conservatives overwhelmingly support voter ID laws regardless of how much racial resentment they express.

Of course, public opinion does not automatically translate in social policy. Please see Part 2 of this post.

You may also be interested in reading:



Thursday, September 18, 2014

Is Fracking Safe?

It May Depend on What You Mean By "Fracking"

A new study by Thomas Darrah of Duke University and four colleagues, published in the Proceedings of the National Academcy of Sciences, suggests that there has been significant contamination of the ground water from natural gas drilling in Pennsylvania and Texas, but this contamination is not due to the process of hydraulic fracturing itself. Rather it is caused by leakage of methane, the primary ingredient in natural gas, and other chemicals near the surface due to faulty construction of the well.

The researchers took 133 water samples from drinking water wells—113 of them from along the Marcellus Shale in Pennsylvania and 20 from the Barnett Shale in Texas. Through chemical analysis, they were able to tell whether each sample was contaminated, and if so, from where the contimation had come. I'm not knowledgeable enough in chemistry to understand the testing that was done, so I rely on the fact that this was a peer-reviewed journal.

They found eight clusters of contaminated wells—seven in Pennsylvania and one in Texas. The cluster in Texas consisted of five of the twenty samples they took. They don't say how many samples were included in the seven Pennsylvania clusters. From eyeballing their charts, it looks like about 20-25% of their samples were contaminated. However, the wells were not randomly selected. The researchers had focused on areas where contamination was suspected.

The chemical analysis indicated that four of the eight clusters were due to leakage around the cement used to seal the outside of the well, three were due corroded or poorly joined steel tubing used to drill into the ground, and the last was due to “underground mechanical well failure.”

Diagram by Howarth and Engelder (2011)
In hydraulic fracturing, or fracking, fracking fluid is forced into the shale at high pressure in order to release the natural gas contained in it. The concern is that fracking might create breaks in the undreground rock formations, through which either natural gas or fracking fluid might leak upward and contaminate drinking water. None of the contamination came from far enough below the surface to indicate that this type of contamination had occurred.

Unlike last week's study of the correlation between fracking sites and health problems, this study received generous coverage by the corporate media. The general spin seemed to be that “fracking is safe.” The Pittsburgh Post-Gazette ran a McClatchy article about the study on the front page. Two days later, they reprinted a Bloomsberg op-ed entitled “Fracking is not the threat.” The threat, it seems, is drilling company carelessness, which can be avoided through government regulations, regular inspections, and fines for poor performance.

The coverage is superficial in at least two respects. First, while there is no evidence that fracking in the narrow sense--that is, hydraulic fracturing, per se--has caused contamination, fracking in a broader sense--the fracking boom--is certainly responsible for ground water contamination. Regulation and oversight of the drilling industry will not be accomplished easily, especially in a state like Pennsylvania, where the state government is a wholly-owned subsidiary of the natural gas industry.

Secondly, the absence of evidence of leakage due to fracking, per se, does not necessarily mean that there is none and the process is safe. Methane and fracking fluid from deep in the shale may simply take a longer time to reach the surface than chemicals that leak from at or near the surface. Maybe decades. The Darrah study may simply have been done too soon.

Meanwhile, there is accumulating evidence of another problem caused by natural gas drilling. A new report published in the Bulletin of the Siesmological Society of America shows a significant increase in earthquakes in the Raton Basin of Colorado and New Mexico since 2001. Fracking began there in 1999. As in previous studies, most of the earthquakes are located within 5 kilometers of waste water injection wells, rather than near sites where hydraulic fracturing has taken place. I guess that means that fracking is “safe,” but it doesn't solve the problem of what to do with the waste water.

You may also be interested in reading:




Sunday, September 14, 2014

Fracking and Public Health, Part 2

The Coverage

Please read part 1 about the Rabinowitz survey of fracking and health.

With the exception of USA Today, the response of the corporate media to the Rabinowitz study has been a collective yawn.

Since Washington County is part of the area served by the Pittsburgh Post-Gazette, one might expect the survey to be front-page news. On the contrary, the study was buried in a short article on page 4 of the local section, with what seems like a deliberately uninformative headline, “Study looks at gas wells, health.” Three of its nine paragraphs were devoted to rebuttal from Travis Windle, a spokesman for the Marcellus Shale Coalition, an industry group.

© All rights reserved by jeffs4653
In an unusually harsh attack on the integrity of the researchers, Mr. Windle stated, "This study, done in partnership with a local activist group, was designed to put selective and unproven data behind a predetermined and biased narrative." The activist group is presumably the Southwest Pennsylvania Environmental Health Project, which is opposed to fracking. They are credited with assistance with the survey; specifically: "The survey was pre-tested with focus groups in the study area in collaboration with a community-based group and revised to ensure comprehensibility of questions." If this is all they did, they played only a limited role in the study.

Pittsburgh's avowedly right wing newspaper, the Tribune-Review, offered an unusual take on the Rabinowitz study. First of all, they claim that this study is balanced by a “confllicting” 2014 study by Penn State geologist Terry Engelder and colleagues. This study found no evidence that fracking fluids, injected deep underground into the Marcellus shale, were migrating toward the surface and contaminating the ground water. Rather than conflicting, the two studies are actually irrelevant. Rabinowitz attributed the symptoms he discovered to air pollution, pollution of ground water from fluids leaking from around the drilling site, and stress. Migration of deep underground fluids, if it occurs, is presumably a problem for future generations to deal with, long after the drilling companies have taken their money and run.

The Tribune-Review also suggested that both studies are biased due to their source of funding. The Engelder study was supported by the Marcellus Shale Coalition and included an employee of Shell International as one of its co-authors. The primary supporter of the Rabinowitz study was the non-profit Heinz Foundation. The Heinz Foundation underwent a turnover of its executive staff beginning in August 2013, which is widely interpreted as a shift toward opposition to fracking. Previously, Heinz had been a member of the Center for Sustainable Shale Development, a coalition of environmental and industry groups that was supposed to establish voluntary safety rules for drilling. However, the Rabinowitz study was completed in 2012, while Heinz was still friendly with the drilling industry.

It's clear that we need an improved vocabulary for talking about real or imagined conflicts of interest in research, one which, at a minimum, distinguishes between financial and ideological conflicts of interest. If your research is supported by a for-profit corporation and its results are contrary to corporate interests, two consequences are likely. First, the research results will never be made public, and second, you will never again receive financial support from that corporation. Those outcomes are unlikely when your research is supported by a non-profit foundation.

The Tribune-Review seeks to expand the category of conflicts of interest to include those cases when the researchers or their sponsors have an opinion about the research topic. But this is unrealistic, since investigators almost always have opinions about their research, opinions which often explain why they were interested in the topic in the first place. Even when the research topic is completely non-political, scientists are often professionally committed to their research hypotheses. Good research design ensures that the researchers' expectations, and those of their financial supporters, do not influence the results.

I can't comment on the design of the Engelder study because I lack expertise in geoscience. I described the Rabinowitz study in some detail because it seems to me that the authors have done a good job, under the circumstances, of minimizing researcher bias and evaluating alternative explanations. Right now, the best advice to people on both sides of the issue is replicate, replicate, replicate. For example, studies of fracking and health are in progress at the Geisinger Health Center in central Pennsylvania. Ultimately, whether researchers have successfully eliminated bias is an empirical question. If studies are replicated often enough, it should be possible to determine through systematic reviews whether researchers with different sources of funding have obtained inconsistent results.

You may also be interested in reading:

Saturday, September 13, 2014

Fracking and Public Health, Part 1

An Important First Study is Published

Corporations that engage in mining and manufacturing are extremely fortunate that Americans cannot be randomly assigned to places of residence. This prevents researchers from doing true experiments--that is, randomly assigning people to experimental and control groups to study the effects of air, water and soil pollution on public health. As a result, when epidemiologists, who study the spatial distribution of health and illness, find evidence of illness concentrated in the vicinity of sources of industrial pollution, it its always possible for corporations to claim that something else—known or unknown—is responsible. Correlation does not mean causation.

On Wednesday, what is to my knowledge the largest scientific study relating proximity to natural gas wells to human health, conducted by a team led by Peter Rabinowitz of the Yale School of Medicine, was released online by the peer-reviewed journal Environmental Health Perspectives. Previous evidence of an association between natural gas drilling and illness had come either from case studies or convenience—that is, nonrandom—samples.

Photo by Ruhrfisch
The study was done in Washington County, south of Pittsburgh, which had 624 active natural gas wells, 95% of which utilized hydraulic fracturing. A set of 38 contiguous rural communities were selected for study, to avoid urban areas unlikely to have ground-fed wells. Within these communities, 760 geographic points were randomly chosen and the nearest residence was selected for study. Some sites were excluded due to duplication, because there were no homes nearby, or because the house did not have a ground-fed well. This left 255 eligible households, 180 (71%) of which agreed to participate. Forty-seven households refused to participate, and researchers were unable to contact residents of the remaining homes.

Participants were interviewed in their homes in the summer of 2012 by trained survey researchers who followed a prepared script. The survey made no reference to natural gas drilling. Researchers were unaware of the home's proximity to the nearest gas well. (Of course, this precaution failed if a drilling rig was visible from the front porch.) An adult member of the household was asked questions assessing the health status and symptoms of each member of the household—a total of 492 people. Participants were paid $25. Participating households were divided into those within 1 kilometer (.6 miles) of a natural gas drilling site, those between 1 and 2 km, and those more than 2 km from the nearest site.

Respondents were asked about six types of health problems: dermal, upper respiratory, lower respiratory, gastrointestinal, neurological and cardiovascular. The following variables were statistically controlled: age, gender, education, occupation, the presence of smokers or pets in the home, and awareness of nearby environmental hazards. Statistically significant results were obtained in two of the six health areas, dermal and upper respiratory.
  • Upper respiratory symptoms, such as coughing, itchy eyes and nosebleeds, were reported by 39% of those living within a km of a drilling site but only 18% of those who lived more than 2 km away. Both the near and intermediate groups reported significantly more upper respiratory symptoms than the far group.
  • The number of people suffering from dermal problems was smaller, but the association was stronger. Thirteen percent of those living nearest to a gas well suffered from skin problems, such as rashes, itching and burning, but only 3% of those living the farthest away. The near group reported significantly more dermal symptoms than the far group.

The authors do not claim that their data show a causal relationship between drilling and health problems. However, they speculate that, if natural gas drilling is responsible for these skin and respiratory issues, it could have happened through three pathways: (1) pollution of ground water, including the water in their wells; (2) air pollution from the drilling site; and (3) stress due proximity to a drilling site. Noise is a particularly potent stressor. Supporting the stress hypothesis is the fact that those who were aware of nearby environmental hazards reported more skin and respiratory symptoms, although the relationship between proximity and symptoms was still significant when the effect of awareness was statistically removed.

Whenever a relationship between industrial pollution and illness is found, the most likely alternative explanation is the indirect involvement of social class. Locations that are close to sources of pollution are usually undesirable places to live. Property values are lower and people living nearby have lower incomes. We know that, for a variety of reasons, lower income Americans are in poorer health. However, this explanation is weakened somewhat by the fact that almost all the natural gas wells in Washington County were drilled within the last 5-6 years. This allowed less time for migration to occur and for pockets of poverty to form around the drilling sites.

Because the researchers relied on self-report data, it's possible that people living close to drilling sites exaggerated their symptoms—although they might also have had reasons to minimize them. The fact that the results were still strong when awareness of environmental hazards was controlled argues against this interpretation. Future researchers should consider conducting actual medical examinations of the participants. This would of course be much more expensive and logistically challenging.

The 71% completion rate is well above average for surveys. However, it might be argued that people experiencing illnesses that they attributed to local drilling were more anxious to tell their story, and that this contributed to an overestimation of the percentage with health problems. However, the refusal rate did not vary significantly with distance from the nearest drilling site.

The health problems detected in the study were among the least serious included in the survey. Since most of those surveyed had only been living near a gas well for a short time, the authors say they were not surprised that diseases with longer latencies, such as heart conditions or cancers, were not yet associated with proximity to a drilling site.

The research was supported by a grant from Pittsburgh's Heinz Foundation, along with four other non-profit sources. The authors thanked the Southwest Pennsylvania Environmental Health Project for “assistance with the community survey.” They conclude by calling for further research, including “longitudinal assessment of the health of individuals living in proximity to natural gas drilling activities, medical confirmation of health conditions, and more precise assessment of contiminant exposures.”

To be continued in Part 2.

You may also be interested in reading:



Thursday, September 4, 2014

Reefer Madness Revisited

Chronic pain is a major problem in this country. To combat it, doctors are prescribing—some would say overprescribing—opioids such as Vicodin and OxyContin. Americans consume about 90% of the world's supply of these dangerous drugs. According to the Centers for Disease Control, opioid overdose deaths have increased from 4,030 in 1999 to 16, 651 in 2010. That's an average of 46 deaths per day, more than heroin and cocaine combined. Sixty percent of these folks have a legitimate doctor's prescription. If only someone could come up with a safer way of treating pain. . . . .

Marijuana has been shown to be effective in treating chronic pain. A survey of medical marijuana users in Canada found that 68% of them used it as a substitute for prescription drugs. The three main reasons given were fewer withdrawal symptoms, fewer side effects, and better symptom management. Marijuana can also be effectively combined with opioids to allow patients to use lower doses of opioids. If you can locate someone who has died from a marijuana overdose, you have a scoop. In July, New York became the twenty-third state to legalize marijuana for medical use.

Marcus Bachhuber and his colleagues published a new study of the effects of marijuana legalization on opiate deaths in the Journal of the American Medical Association this week. It's not a controlled experiment; it's a before-after comparison group design. While there is some ambiguity about how to interpret it, the study has a number of important strengths.

The authors counted opioid deaths on death certificates from all 50 states between 1999 and 2010. By 1999, three states (California, Oregon and Washington) had already legalized medical marijuana. Another ten states legalized it while the study was in progress. (The remaining ten states changed their laws after 2010.) The death rates from opioids were compared between those states that had legalized medical marijuana (the marijuana states) and those that had not (the comparison states). They also compared death rates within the marijuana states from before to after legalization. The design statistically controlled two variables that are known to affect the opioid death rate, the unemployment rate and state policies regulating prescription drugs.
  • Those states in which medical marijuana was legal at the time had a 24.8% lower mean annual opioid mortality rate than the comparison states. The marijuana states had 1729 fewer death than would otherwise be expected in 2010 alone. The results were unchanged when suspected suicides were eliminated.
  • In the ten states that legalized marijuana between 1999 and 2010, the drop in the death rate from opioids coincided with the change in the law. It took a couple of years for the full effect of the legal change to be realized. On average, the death rate dropped 20% in the first year, 25% in the second, and peaked at 33% in the fifth and sixth years.

In searching for possible hidden confounds, our attention is naturally draw toward possible differences between the marijuana and comparison states that might have affected overdose deaths. There is a possible selection bias. The marijuana states are more progressive, and it's possible, but not obvious, that political liberalism might reduce painkiller deaths in other ways. However, when the death rates are compared from before to after the legalization, the marijuana states become their own controls.

It is important to note that in order to explain the results, a potential confound would not only have to be more prevalent in the marijuana states, it would also have to have occurred at about the time legalization took effect. One of the strengths of this study is that the drop in deaths coincides with the onset of legalization even though the law changed at different times in different states. This makes it less plausible that historical events affecting several states simultaneously could account for the results.

Almost all news reports about this study quote drug experts who urge caution when drawing conclusions from the study, although they usually don't specify what they think is wrong with it. One expert, Dr. Andrew Kolodny, is quoted in Newsweek as suggesting that states that legalized medical marijuana might also impose more restrictions on the prescribing of painkillers, apparently not realizing that the authors had anticipated this possibility and found that prescription monitoring laws were not associated with lower overdose death rates. (The reporter did not correct his error, allowing readers to assume that he had made a valid point.) Of course, most articles also contain the obligatory vague words of warning about the alleged dangers of the devil's weed.

The study makes a fairly strong case for the immediate legalization of medical marijuana at the federal level and in the remaining states. Why is medical marijuana still illegal? As the Bachhuber study suggests, medical marijuana is a serious threat to the profits of pharmaceutical companies. Given the opportunity, chronic pain sufferers are likely to substitute marijuana for more expensive and dangerous prescription drugs. Journalist Lee Fang points out that the groups leading the fight against relaxing marijuana laws, such as the Partnership for Drug-Free Kids, receive a large portion of their funding from the pharmaceutical companies that market analgesic opioides. Several leading anti-marijuana academic experts also serve as paid consultants to big pharma. Here is Fang being interviewed by Chris Hayes.


As usual, we can say (this time, with enthusiasm) that more research is needed. The good news is that the Drug Enforcement Administration has increased its annual allotment of cannabis for clinical research from 21 kilograms to 650 kilograms—still too little—in response to demand from investigators.

You may also be interested in reading:

Legalized Bribery

Evergreening

Taming the Gift Culture

Thursday, August 28, 2014

A Jazz Metaphor

In an interview conducted by Thomas Frank (“Cornel West: 'He posed as a progressive and turned out to be counterfeit'”), Dr. Cornel West of Union Theological Seminary expresses his disappointment with Barack Obama's presidency.

It's a sad thing. It's like you're looking for John Coltrane and you get Kenny G in brown skin.

You may also be interested in reading:

If you're interested in jazz and blues, you might want to check out my other blog, The Blues and the Abstract Truth.

Wednesday, August 27, 2014

Eyewitness Blues, Part 2

The Controversy

Before reading this post, please read Part 1, which presents the evidence that sequential lineups are more accurate than simultaneous lineups.

Why do the police prefer the simultaneous method? First, it produces more identifications, which is their immediate goal. Since they are likely to believe that whomever they have arrested is the guilty party, they argue that the greater number of misses by the sequential method results in guilty suspects being allowed to go free, roam the streets, commit further crimes, etc. They are not as sensitive to the costs of false alarms since they underestimate how often they occur.

In his statement refusing to adopt sequential lineups, Mr. Bucar refers to a meeting of the National Academy of Sciences (NAS) in which there was “dissent among scientists who claim that the analysis methods used in the research between 2009 and 2011 was faulty and when corrected will actually show that . . . the simultaneous method is superior.” NAS's Committee on Science, Technology and Law held three meetings on eyewitness identification in 2013 and 2014. The program of these meetings, along with slide shows submitted by the participants, is available online. However, I couldn't determine the exact nature of the controversy to which Mr. Bucar refers. I am puzzled by his reference to analysis methods used between 2009 and 2011, since I don't detect any change in methodology. NAS is preparing a report on eyewitness identification to be released in the near future. I've signed up to receive a copy and I'll let you know their recommendations.

Meanwhile, here's my best guess as to what Mr. Bucar is concerned about. One of the great unknowns in the real world use of eyewitness identification is the base rate of culprit-present lineups. What percentage of lineups actually contain the prepetrator? Is it 50%? 75%? 90%? Neither the identification rate nor the conviction rate really answer this question.

In the studies referred to earlier, the base rate is 50%, since the researchers run an equal number of participants with culprit-present and culprit-absent lineups. In the real world, the lower the percentage of culprit-present lineups, the greater the danger of false alarms. For example, suppose the police conduct a near-random sweep of the neighborhood and show eyewitnesses many people who are not identified as suspects by any other type of evidence. Under these circumstances, it becomes more important to use the sequential method.

The police probably believe that close to 100% of their lineups are culprit-present. If they are right, the number of true identifications—or more likely, lucky guesses—lost in these culprit-present lineups using the sequential method might exceed the number of false identifications avoided in the (presumably) smaller number of culprit-absent lineups. Using Bayes' theorem, Steven Clark has identified the crossover point at which the misses exceed the false alarms using the sequential method. If the true base rate is greater than 85%, the simultaneous method will identify more guilty suspects.

Of course, no one will ever know what the true base rate of culprit-present lineups is. Your guess is likely to be influenced by whether you trust the fairness and efficiency of police investigative procedures or are more skeptical. One of the more important contributions of DNA testing is that it has greatly increased our estimate of the number of innocent people who are convicted of crimes in this country.

However, simply comparing the rate of false alarms in the simultaneous condition to the rate of misses in the sequential condition doesn't do justice to the true superiority of the sequential method because a false alarm is a more costly error for society than a miss. In both cases, the real culprit escapes detection, but when a false alarm occurs an innocent suspect is likely to be arrested and charged, with costs to that individual that range from considerable inconvenience to the complete ruin of his or her life. This is presumably the basis of Lord Blackstone's famous quote: “It is better that ten guilty men escape than that one innocent suffer.”

In a classic case of bad timing, Allegheny County police recently released a man arrested based on a mistaken identification from a culprit-absent lineup using the sequential procedure. Another man confessed to the crime. Of course, no one claims the sequential procedure is foolproof. Our best guess is that, when the culprit is absent, it will produce false alarms almost one-third (32%) of the time, which is disturbing, but not as bad as the 54% false alarm rate using the simultaneous procedure.

Monday, August 25, 2014

Eyewitness Blues, Part 1

The Issue

On July 31, Pittsburgh public safety director Stephen Bucar had an unpromising first day on the job. A week before, Allegheny County District Attorney Stephen Zappala sent him a letter requesting changes in the way the Pittsburgh city police conduct eyewitness identifications. Mr. Zappala said his office would no longer prosecute cases unless (1) the identification was made using a sequential rather than a simultaneous lineup and (2) the lineup was conducted by an officer not involved in the investigation. His demand may have been motivated by an abstract desire for justice, but it's more likely the result of two recent lawsuits filed by men who were mistakenly arrested based on faulty identifications using the simultaneous method.

But the Pittsburgh police, notorious for their resistance to change, said something like, “That's not how we do things around here.” Unfortunately, Mr. Bucar decided to back them up, citing what he believes to be conflict in the scientific literature over whether sequential or simultaneous lineups are more likely to lead to error. This is disappointing news for those of us who had hoped new Mayor Bill Peduto would bring more rationality to Pittsburgh government.

Stipulating that the identification be conducted by an officer not involved in the case assumes that this officer won't know which person in the lineup is the suspect. If so, this double blind procedure prevents bias in which the administrator unintentionally communicates the identity of the suspect to the eyewitness. Since Mr. Bucar did not mention blind testing in his statement, I presume he has no objection to it.

When an eyewitness views a lineup, there are four possibilities. The culprit is either present in or absent from the lineup, and the eyewitness either does or does not make an identification. If the culprit is present and the eyewitness correctly identifies the culprit, this is called a hit. When the eyewitness either identifies someone else or makes no identification, this is a miss. When the culprit is absent and the eyewitness erroneously makes an identification, this is a false alarm. Failure to identify anyone is a correct rejection. Obviously, the goal is to maximize hits and correct rejections, thereby minimizing misses and false alarms.
Culprit Present
Culprit Absent
Identification
Hit
False Alarm
No Identification
Miss
Correct Rejection
In the traditional simultaneous procedure, the eyewitness is shown an array of (usually) six photographs and asked which one, if any, is the perpetrator. In the sequential procedure, the eyewitness is shown the photographs one at a time, and is asked to state whether each one is the perpetrator. In this procedure, eyewitnesses usually don't know how many photos they will be shown.

In studies that compare these methods, participants unexpectedly witness a simulated crime, either live or on video, and are later asked to make an identification. In those studies with complete designs, the perpetrator is present in half the lineups, while in the other half, he or she is absent. Half of each of these lineups are conducted using the simultaneous and the sequential procedure. Since the researchers know whether the culprit is present in each lineup, they can determine whether a correct identification was made.

Several decades of research show the sequential procedure to result in greater accuracy. In 2011, Nancy Steblay and her colleagues published a meta-analysis of all 72 known studies (from 23 labs involving over 13,000 participants) comparing simultaneous and sequential lineups. Simultaneous lineups produce higher choosing rates (76%) than sequential lineups (61%), suggesting than the simultaneous method encourages guessing. More importantly, identifications from a sequential lineup contain fewer errors.

To illustrate this, here are the data from the 23 studies with complete designs (as described above). Since all four conditions are present in these studies, it is possible to compare their average results.

Hits from culprit-present lineups
          Simultaneous                     52%
          Sequential                         44%
          Difference                           8% fewer hits in sequential

False alarms from culprit-absent lineups
          Simultaneous                     54%
          Sequential                          32%
          Difference                          22% fewer false alarms in sequential

There is a tradeoff. If all lineups were culprit-present, the simultaneous method might appear to be superior, since it produces 8% more identifications. However, since the simultaneous method encourages guessing, some researchers argue that this 8% difference should be treated as lucky guesses rather than true identifications. When the culprit is absent, the simultaneous method leads to 22% more false alarms. Therefore, the sequential method is more accurate.

Why is the simultaneous method more error prone? When an eyewitness is asked to view a lineup, he or she probably infers that the police think they have arrested the perpetrator. In a simultaneous array, eyewitnesses make a relative judgment. They compare the photos to one another and choose the one which most closely resembles their recollection of the perpetrator. In the sequential method, however, because they don't know how many suspects they will be shown, they must make an absolute judgment, comparing each suspect to their independent memory of the perpetrator.

Here is a 61-min lecture by researcher Gary Wells that I posted once before. Between minutes 14 and 30, he talks about the relative judgment process and the need for sequential lineups.



Wednesday, August 20, 2014

Modeling Road Rage

One of my guilty pleasures is watching auto racing on television, an interest that traces back to my childhood, when my father took us to midget, sprint car and stock car races throughout the Northeast. Auto racing suffered a tragedy this month when Tony Stewart, three-time NASCAR Sprint Cup champion, driving a sprint car on a dirt track in Canandaigua, NY, ran over and killed fellow-driver Kevin Ward, Jr., who was standing on the track. The incident was unusual only by virtue of its sad ending.

In racing, it's common for contact to occur when one car passes another. It doesn't always cause a accident, but when it does, it is most often the slower car that is left spinning to the side of the track. The passing driver typically denies that the accident was intentional; his victim usually has a different interpretation. It is not unusual for the victim to exhibit signs of road rage, for example, standing on the track and yelling or shaking his fist at the other driver as he passes the scene of the accident.

On the lap prior to the fatal accident, Stewart (#14) spun out Ward's (#13) car, bringing out the caution flag. The video shows what happened next.


You can hear Stewart hit the throttle just before he drives past Ward. (The video doesn't permit us to locate the source of the sound, but eyewitnesses claim it was Stewart.) Stewart may have responded to Ward's anger with some road rage of his own. On a dirt track, drivers will sometimes deliberately spray dirt onto someone standing on or near the track. However, accelerating suddenly on dirt can cause the car to fishtail, sending its rear wheels toward the outside of the track. Stewart's right rear tire struck Ward, pulled him under the vehicle, and killed him.

Stewart's actions were reckless but almost certainly not intentional. It's doubtful that he will be prosecuted, since Ward assumed a great deal of risk by walking on the track. He may, however, suffer professional consequences, especially given his long history of on-track aggression. (For example, when Stewart was taken out in an accident he attributed to fellow NASCAR driver Matt Kenseth, he threw his helmet at Kenseth's front windshield. Later he threatened Kenseth with retaliation in a television interview.) Stewart's greatest risk will be loss of corporate sponsorship and income from television commercials.

AAA reports that aggressive driving, broadly defined, is a factor in 56% of fatal auto accidents. A 2013 article by Christine Wickens and her colleagues summarizes what is known about the causes of road rage, which they define more narrowly as aggressive actions on roadways resulting from hostility toward another driver. Since little is known, the article functions as a road map for future research. They divide the causes into personal, i.e., men exhibit more road rage than women, and situational, i.e., it occurs when aggressive intent is attributed to another driver. Not mentioned among potential causes is observational learning and imitation of the behavior of role models.

NASCAR and other racing sanctioning bodies have been quite tolerant of aggressive displays on the track, probably because they think they increase the sport's popularity. But drivers like Tony Stewart are liked and respected by racing fans, and their actions might serve to legitimize similar aggressive actions on the roadways. Are changes in the popularity of auto racing on television correlated with changes in the frequency of aggression on our highways? (NASCAR's ratings have tailed off in recent years. Has road rage also declined?)

There has been relatively little research on the effects of specific events shown or reported in the mass media. An exception is research on suicide contagion, a problem which has become timely following the death of Robin Williams. Suicides increase in the country in the week following the suicides of well-known people. The effect is greater when person who commits suicide is real rather than fictional and there is evidence suggesting that contagion is more likely to occur when there is a demographic match between the model and his or her imitator. These studies could point the way for future research on the connection between televised incidents and road rage among racing fans.

The Stewart incident has brought about at least a temporary change. NASCAR has announced that drivers will not be permitted to leave their cars following an accident until they are escorted off the track by the safety crew. (Of course, there are exceptions, such as when the vehicle is on fire.) NASCAR claims that this has always been the rule, but if so, it hasn't been enforced. If this change is permanent, it could impact not only driver safety, but audience safety as well.

Now, NASCAR, what about those threats of violence in post-race interviews?

You may also be interested in reading:

Situation Alarming--But Not Serious

Friday, August 8, 2014

Americans For Racial Injustice

In spite of declines in violent crime during the past two decades, the United States imprisons more people than any other country in the world, and at any time in its history. Our punitive criminal justice policies have had a grossly disproportionate impact on racial minorities. While African-Americans make up 12% of the population, they are 40% of our criminal inmates. These disparities exist even where evidence suggests there are no racial differences in commission of the offense.

Both President Barack Obama and Attorney General Eric Holder have argued for reducing the length of prison terms for nonviolent offenses, giving racial disparities in incarceration as one of their reasons. The President commented on this issue when the administration announced new guidelines for public school discipline intended to reduce racial disparities in school suspensions and dropout rates.


I remember thinking at the time that this argument could backfire. In a racist society, stating racial discrimination as a reason to reduce mass incarceration carries the risk of increasing support for punitive crime policies. Ian Haney-Lopez has noted that get-tough crime policies are a "dog whistle" used by politicians to appeal to white prejudice without referring directly to race. These coded racial appeals are used because they are effective. A new article by two Stanford social psychologists, Rebecca Hetey and Jennifer Eberhardt, supports these speculations.

Hetey and Eberhardt note that racial disparities in imprisonment are usually treated as an unintended outcome of punitive policies. They propose instead that knowledge of racial discrimination in the criminal justice system is one of the causes of widespread public support for these policies. They conducted two field experiments.

In the first, white registered voters were approached by a white female experimenter at a train station and asked to sign a petition for a ballot initiative to reduce the punitiveness of California's "three-strikes-and-you're-out" law. Before being presented with the petition, participants were shown a 40-sec video about the prison system. The video did not refer to race, but in one version, 25% of the inmates shown were black, while in the other version, 45% were black. As predicted, fewer people (27%) were willing to sign the petition in the More-Black condition than in the Less-Black condition (52%).


The second study was an online survey of white New York City residents conducted by Survey Sampling International. Near the beginning of the survey, participants were given demographic information stating that the prison population was either 40.3% or 60.3% African-American. (Both statements were true. The first referred to the U. S. prison population, while the second referred to New York City.) Later in the survey, participants were shown a petition calling for an end to the New York police's "stop-and-frisk" policy, and asked whether, if they had been approached, they would have signed it. The results confirm those of the first experiment.


A measure of fear of crime was included in the survey. The relationship between racial composition of the prison population and support for less punitive policies was mediated by fear of crime.  That is, the More-Black condition produced greater fear of crime, which led to less willingness to sign the petition.  However, this was not the whole story.  The effect of racial composition on policy acceptance was statistically significant even when fear of crime was statistically eliminated.

The authors admit that their results pose "quite a challenge for those striving to create a more just and equal society."  It appears that presenting people with evidence that a social policy leads to racial injustice has the effect of increasing, rather than reducing, support for the policy.  Maybe the President should emphasize the fact that reducing the prison population would save money.

You may also be interested in reading:

Dog Whistle Politics

Another Dog Whistle

Old-Fashioned Racism