Happily, my paper “Does ‘right to work’ imperil the right to health? The effect of labour unions on workplace fatalities” is attracting a lot of attention, which makes me feel good as an academic who cares at least a little about producing things that others find valuable (we will see how the sheriff stuff is received, once it’s finally done, though it seems like there’s something both quaint and gripping about them!). Although I wrote that short article in August 2017 — and had really no sense of the Janus case at that time — it’s really seemed to have caught the moment. It helps that there’s yet a non-trivial, vested constituency in organized labor that finds something useful, or reassuring, in these results. Anyway, in response to a journalist query about, essentially, what “14.1%” (the coefficient of the reduced form regression of the RTW variable) means exactly,  I wrote something like this:


There were 138,736 total deaths on the job in the 50 states over 1992-2016 recorded in the Census of Fatal Occupational Injuries. Using a negative binomial regression (with the same predictors as the “reduced form” regression in the paper, but with the number of workers as an offset variable and the state/annual count of workplace deaths as the dependent variable) to model the incidence of occupational deaths,  the expected count of occupational deaths in a state during this time-period is about 118.7 in a right to work state and about 104.9 in an otherwise similar state without RTW (figure above; Stata margins command with coefplot for figure). Reassuringly, this 13.1% increase in RTW states in the negative binomial model is essentially the same result as in the “reduced form” model in the paper (about a 14% increase in RTW states relative to others). The Incident Rate Ratio on the right-to-work coefficient is also  1.1314 — which makes sense, as that is the ratio of the predicted values!

Over this same 1992-2016 period, 529 state-year dyads had right to work legislation (721 did not). Therefore, the model-implied count of occupational fatalities attributable to RTW  (assuming no confounders) is (118.7-104.9)*529 or about 7,300.That is, if no states had RTW over these years, the model implied counterfactual is that about 131,436 occupational deaths would have occurred instead of the observed 138,736 occupational deaths.

Semi-related: Using the original (non-logged) rate of fatalities, here’s a predicted effects plot of occupational fatality rate at different levels of unionization; I shared this on Twitter previously. This was used with an OLS model containing all the parameters of model 3 in the paper (or replication code), but with the non-logged dv (in the replication code labeled “robustness check using original unscaled variable”). This shows how the actual, observed decreases in unionization rates we’ve seen in states like Wisconsin in recent years may translate into meaningful increases in occupational mortality.


While a student, I wrote several articles for Vanderbilt’s campus newspaper, but all of these were lost from the Vanderbilt Hustler website (along with all historic website content) sometime in the summer of 2016. I’m going to try to upload these articles here while I can still find scanned PDFs of the print issues that existed at the time. Unfortunately, the articles’ original links to sources are lost. 

This article did end up prompting a conversation about the tuition insurance policy. Vanderbilt Student Government passed a resolution encouraging a policy change. The Bursar (who was very genial when I met with him about this with the Student Body President) negotiated a new tuition insurance policy that reimburses mental and physical illness equitably, which took effect the following year (2015-2016).

Equality in name only? How Vanderbilt promotes a discriminatory insurance policy

(Published August 27, 2014)

Like many universities, Vanderbilt aspires to promote diversity and inclusion. The university requires its first-year students to discuss racism and the importance of pluralism, and its official policies include a promise to not discriminate in university programs. There is even an Equal Opportunity, Affirmative Action, and Disability Services Department whose “core values include equity, diversity, inclusiveness, accessibility and accommodation.” In light of the university’s seeming commitment to equality, I found it particularly vexing to receive a letter from Vanderbilt, accompanied by a signed note from the bursar, promoting a policy that is patently discriminatory.

Vanderbilt offers a “tuition-insurance” plan issued by A.W.G. Dewar, Inc., which provides reimbursement of education fees should a student withdraw from the university for medical reasons. The plan explicitly specifies that withdrawals caused by “mental health conditions” are given a 60 percent refund while a 100-percent refund is given for “injury and sickness.” The plan also indicates that withdrawals precipitated by addiction or substance abuse receive zero compensation.

On the face of it, the plan suggests either that mental illness is equivalent to three-fifths of a physical illness or that 40 percent of doctor-certified mental illnesses are feigned. Neither of these claims is borne out by any evidence. Rather than aiding a vulnerable student, the plan adds another layer of financial stress to a person likely facing a cascade of problems, possibly impeding his or her return to Vanderbilt.

This practice makes for an incoherent insurance policy, since it provides significantly less coverage for the illnesses most likely to warrant withdrawal. A 2009 study at 10 universities found that having clinical depression was the strongest predictor of withdrawing from college. So giving less coverage for mental illness is rather like having car insurance that provides 40 percent less coverage when your accident involves another car.

Moreover, distinguishing between mental and physical illness to provide unequal service perpetuates stigma against psychiatry and psychiatric illness — the idea that mental illness is “less real” than physical illness, or worse still, that someone is more blameworthy for having a mental illness than a physical one.

The practical implications of this policy are jarring to the point of ineffability. Why should a student suffering from depression because a person raped her receive a 40-percent-reduced reimbursement compared to a per-son struck by the “kissing disease”? Is post-traumatic stress caused by being robbed at gunpoint a less valid reason for withdrawing from college than a football injury?

The policy also violates the spirit, if not the letter, of federal law. In 2008, Congress passed the Mental Health Parity and Addiction Equity Act, which requires equal coverage for mental illness, including substance abuse, in health insurance plans. Hence, as The New York Times writes: “This would probably be illegal if tuition refund policies were deemed health insurance, instead of insurance that just happens to be based solely on your health.”

Since 2010, Vermont has required that colleges in the state provide equal coverage for withdrawal due to mental illness. Prompting the change was a complaint by a University of Vermont student named Sherry Williamson, a registered nurse who suffered from depression. “I couldn’t believe that UVM, which tries to promote diversity and be all-encompassing, would take on a policy that was clearly discriminatory,” Williamson said in an interview.

By choosing to contract with A.W.G. Dewar, Inc. under these terms, Vanderbilt University is complicit in unambiguous discrimination against people with mental illness. This is unacceptable for an institution that claims to be forward-looking and antithetical to the welfare of its students. Renegotiating a nondiscriminatory plan is essential if the university’s commitment to equality is more than skin-deep.

While a student, I wrote several articles for Vanderbilt’s campus newspaper, but all of these were lost from the Vanderbilt Hustler website (along with all historic website content) sometime in the summer of 2016. I’m going to try to upload these articles here while I can still find scanned PDFs of the print issues that existed at the time. Unfortunately, the articles’ original links to sources are lost. 

Suicide is not inevitable
A look at how small changes can make all the difference 

(published October 23, 2013)

The decision to commit suicide may be the most important choice a person could ever make. Suicide is unrivaled in its finality, permanence and, of course, tragedy. Death is forever, unknown and unknowable, surely a state a person would want to think about carefully before embracing. And to kill oneself may be the only decision one cannot take back. As a result, it is commonly understood that suicide is a deliberate act, the capstone of years of unrelenting suffering.

But this characterization of suicide is almost always wrong. In the spring 2013 edition of
its seasonal magazine, the Harvard School of Public Health called the notion that suicides
are “long-planned deeds” the “biggest fallacy” about the act. The magazine reports: While “people who attempt suicide often face a cascade of problems,” empirical studies suggest that “they act in a moment of brief but heightened vulnerability.” In 2001, epidemiologists at the Centers for Disease Control and Prevention interviewed 151 people who nearly died while attempting suicide. Shockingly, a quarter of interviewees reported that less than five minutes elapsed between the decision to commit suicide and the attempt itself. Half said less than 20 minutes had passed. Seventy percent said less than one hour, and 86 percent said less than eight hours. Suicide thus tends to be an impulsive reaction to a crisis. Many people kill themselves, for example, on the days they get fired, get divorced or fail a test, and the availability of suicidal means is often the trigger for suicide.

The suicides on the Golden Gate Bridge demonstrate this fact. As described in a recent Slate article, the bridge’s “mythic beauty, easy access, and promise of near-certain death” beckons many to suicide. To jump off the bridge, one need only surmount a 4-foot barrier; suicide notes left on the bridge often ask, “Why did you make it so easy?” More than 2,000 people have leapt off the bridge to their deaths since it opened in 1937. Nevertheless, efforts to erect a barrier on the bridge have not gained traction, largely because of the belief that it wouldn’t prevent suicide. But there is good reason to believe that stopping a person from attempting suicide very often saves his or her life. A famous 1978 study at the University of California Berkeley tracked down 500 people who were restrained just before they leapt off the Golden Gate Bridge between 1937 and 1971. At the time of the study, 94 percent were alive or had died of natural causes.

Furthermore, there is ample evidence that even simple measures can stop suicides. A meta-analysis of the effect of bridge barriers at suicide hotspots around the world found
that, on average, they reduce suicides on that bridge by 85 percent; though in some cases
suicides shifted to nearby bridges, barriers still decreased the number of local suicides by 30 percent. Gun-owning households in the U.S. are at least three times more likely to be the site of a suicide, but simply using gunlocks or securing ammunition can reduce the risk of suicide by two-thirds. As the Harvard School of Public Health reports, whether those who attempt suicide survive “depends in large part on the ready availability of highly lethal means, especially firearms.” (More than half of the 40,000 Americans who kill themselves each year do so with guns.) In the U.K., a 1998 law restricting the pack size of acetaminophen (the active ingredient in Tylenol) reduced the number of suicides by 200 in the first three years alone.

Hence, restricting access to suicidal means is an effective way to prevent suicide. Heeding this fact, as Cornell University did, can save lives. Between 1990 and 2010, 27 people committed suicide by jumping off one of the five bridges on Cornell University’s campus. In 2010, responding to three student suicides in a month, Cornell installed safety nets on campus bridges.

Moreover, the Golden Gate Bridge example shows that suicidal thoughts, even when acted on, are often fleeting. Only 30 people have survived a jump off the Golden Gate Bridge; of those, 27 lived out the rest of their lives without killing themselves, according to the Bridge Rail Foundation. In interviews with The New Yorker, many survivors of attempted suicide reported that they felt regret the moment they jumped. Ken Baldwin, a survivor who later found his calling as a high school teacher, recalled: “I instantly realized that everything in my life that I’d thought was unfixable was totally fixable —
except for having just jumped.” Perhaps many of the thousands who did not survive felt the same way.

These findings echo nationwide studies, which suggest that 90 percent of people who attempt suicide and survive ultimately do not kill themselves. Nevertheless, many people cling to the erroneous idea that suicide is inevitable, an opinion commonly held even by psychiatrists, the people trusted with caring for the mentally ill. How many people must die on the altar of fatalism before that belief fades?

Using the same municipal employee data from data.nashville.gov as the previous post, here I am looking at whether there is evidence of gender discrimination in the salaries of Metro Nashville government employees. Obviously this is a pretty complicated issue, and I am only really going to scratch the surface of it. Mostly this is just a fun exercise for me.

Okay, so where to begin: First, I think, it makes sense to begin by just comparing mean salaries by gender. If this doesn’t show evidence of discrimination, then there isn’t much merit it going further.

In fact, as the next two figures show, there’s a pretty big difference in mean salaries between men and women.

salary v gender

t test salaries

The mean salary of men among municipal employees is $7,487 higher than for women. Not surprisingly, this is statistically significant, so the observed gap is more than what we would expect due to chance if mean salaries were equal. (It’s not shown but an f-test showed unequal variances between the two groups of salaries, which is why I assumed unequal variances for the t-test).

But wait – maybe men tend to be over-represented in full-time work while women tend to be overrepresented in part-time work. That could explain the observed difference, not discrimination (although I think that could still show evidence of systematic disadvantages [or at least disincentives] for women in employment).

And, indeed, it is the case that there is this kind of difference employment status between these two groups.

women in part time

chisq emply status gender

However, this doesn’t fully explain the difference in salaries. Even within these categories, men are paid more than women.more emply gender

A difference remains. Mean salary is $46,000 for full-time women vs $52,000 for full-time men – a difference of $6,000 or about 80% of the difference we observed without any adjustment. Tangent: why part time wages for women are slightly higher is a puzzle – when I looked at median instead of mean, this went away, and so the median part-time man earns more. This probably means there are outliers distorting the picture for part-time (some part-time women is getting paid much higher than average, perhaps??).

Another important check is difference in job. Using EEOC reported job categories, I can look at whether women and men tend to be working different sorts of jobs, thus explaining the different salaries.eeoc

And indeed, there’s a visually apparent (and statistically significant; Pearson’s X^2 < 0.0001) difference in the frequencies of jobs by gender. A much larger percentage of women are doing “administrative support” ($35,000 annually) and “para-professional” ($26,000) work. A larger percentage of men are doing “Professional work” ($53,000) or “technicians” ($58,000).

So this is another thing to account for, since it could be driving differences in salaries. I think the differential prevalence of women in the professionals category could signal some important inequities but is not per se discrimination in the sense of unequal pay for the same work. To try to give a more comprehensive account of what’s driving salary, I can do a multivariate regression.

I will do this tomorrow! Oh, the suspense.

I found that the government of Nashville has some open data at https://data.nashville.gov. I messed around with a couple of employment datasheets and made some graphs.

Mean Salary vs Ethnicity

Mean annual salary by ethnicity

Mean Annual Salary by Job Category

mean annual salary by job category

Interestingly, elected officials are not the highest paid category.

Metro Nashville Salaries

metro nashville salaries

In total, annual salaries for Metro Nashville government employees is more than $400 million.

[Featured Image Credit: http://images.fineartamerica.com/images-medium-large-5/nashville-skyline-dan-holland.jpg]

Today one of my coworkers and I got in an interesting discussion (or maybe an argument) about why relatively few medical students become psychiatrists. She just finished her first year at a pretty prestigious medical school.

I claimed that it boiled down to stigma against mental illness, while she argued that the difference is explained by other fields of medicine being “cooler,” “more interesting at least to some people,” and “more hands-on” (which I pointed out does not at all contradict the stigma explanation).

 I (like most people I imagine) am not very good at faithfully and fairly recounting how arguments played out, especially when they get a bit heated. So I’ll from here on try to make this objective (by which I mean independent of my conversation with my co-worker) and just flesh out a few interesting points that came up.

Psychiatrists are paid less and research is funded less than for other specialties

As New York Times writes, “Psychiatrists rarely earn enough to compensate for their additional training. Most would have been better off financially choosing other medical specialties.” Psychiatry is among the least paid medical specialties, though it looks like the pay is increasing faster than in most other areas. Interestingly, HIV/Aids specialists receive the lowest salary of all, and they also treat a stigmatized population.

Psychiatrists are least likely to accept insurance plans, possibly because they are systematically under-reimbursed, even for the same procedures  (that article also says that half of counties lack a regular psychiatrist!). Research on mental illness is significantly underfunded relative to the disease burden it creates.

Are other specialties more interesting, hands on?

Surely for some people and not others – like everything else. Maybe some people really like feet or skin or hearts or reading x-rays.  But for this to make sense in explaining the shortage of psychiatrists, it has to take the stronger form that psychiatry is systematically less interesting than other forms of medicine. And that sounds a little difficult to believe to me. (I am also inclined to think that if the average psychiatrist made $340,000 a year like the average orthopedic surgeon, interest wouldn’t matter much).

 The brain is an incredibly complex organ. Mental pathologies are extremely powerful and disruptive forces – they ruin lives and cause all kinds of bizarre behaviors. Additionally, considering that the lifetime prevalence of mental illness is one in two, and depression and substance abuse are higher among medical students than the general population (discussed later), it is fairly likely that a medical student has personally experienced a mental illness or observed its effects. It seems to me that, all else equal, you would be more likely to go into a field of medicine which has affected you or someone you know than not.

Anyhow, psychiatry sounds a lot more interesting to me than Orthopedics, for example (the highest paid specialty by far) or Plastic Surgery (another high payer). Surgery is a lot sexier than psychiatry (in addition to being much better paid) and I think this is a product of the stigma.

Are future doctors too sophisticated for stigma?

The hypothesis for this line of thinking is pretty intuitive: stigma comes from misunderstanding and misperception of mental illness. Medical students and doctors, who know a lot about mental illness, don’t make these mistakes, so they are less likely to stigmatize mental illness.

Unfortunately, it’s not true. Not at all.

According to an Journal of the American Medical Association study (specifically of med students at UMichigan), mental illness stigma may actually be higher among medical students, who, in addition to being more likely to be depressed than average, are more likely to attribute depression to “weak coping skills” and more ashamed about revealing their negative emotions. This sort of makes sense to me, as medical students strike me as the kind of people who endorse a very strong “protestant work ethic”, “internal locus of control” type of outlook. One that says if you want something, work hard and you get it; if you have a problem, work hard and you’ll solve it.

The implication is, unfortunately, that if you can’t solve a problem, you aren’t working hard enough. It’s your fault, something to be ashamed of.  It’s not hard to see how this attitude, coupled with a high-stress, hypercompetitive environment, would be ripe for high levels of depression and self-stigmatization. After graduating, doctors tend to avoid seeking mental health care.

And in the population at large, while more people attribute mental illness to neurobiological factors than in the past, stigma is actually higher, according to an American Journal of Psychiatry report. Being able to attribute illness to neurobiology doesn’t make people more accepting of illness; in fact, it may make it seem intractable, a rigid characteristic about a person that will never go away.

Mental illness stigma deters people from Psychiatry

Furthermore, there are plenty of published anecdotes (eg here, here, here) of psychiatrists describing the stigma they received for choosing their profession. A Columbia Professor of Psychiatry writing for Scientific American describes this comment made by another faculty member: “Tell all students who get low scores on their board exams not to worry, they just need to change their career plans and go into psychiatry.”

Psychiatry is also, as far as I know, the only branch of medicine with a dedicated movement opposed to its existence.  Ever heard of anti-cardiology?