What I’m Reading: July 2019

As always, Our World in Data provides some interesting numbers to think about, this time with food supply and caloric intake by country.

This article on chronic lyme disease and the whole “medical issue as personal identity” phenomena was REALLY good and very thought provoking.

Ever want to know where the Vibranium from Black Panther would land on the periodic table of elements? Well, now there’s a paper out to help guide your thinking. More than just a fun paper to write up, the professors involved here actually asked their students this on an exam to see how they would reason through it. I’m in favor of questions like this (provided kids know to have watched the movie) as I think it can engage some different types of critical thinking in a way that can be more fun than traditional testing.

I mentioned to someone recently that I have a white noise app on my phone, but after testing it out I found that brown noise tends to be more effective in helping me sleep than white noise. They asked what the difference was, and I found this article that explains different color noises. In my experience the noises that tend to be loudest and most likely to interfere with sleep tend to hang out at the low end of the spectrum, YMMV.

A new study “Surrogate endpoints in randomised controlled trials: a reality check” gives an interesting word of warning to the cancer world. It’s common in clinical trials to use surrogate endpoints like “progression free survival” or “response rate”to figure out if drugs are working. This is done because overall survival can take a long time to get and researchers/patients/drug companies want results faster and it seems like if the surrogate markers are good the drugs can’t possibly hurt.

Unfortunately, it appears this isn’t the case. A new drug venetoclex was studied and patients on it were eventually found to have better progression free survival, but eventually twice as many deaths as those treated with regular treatment. Ugh. The lead author has a great Twitter thread on his paper here, where he suggests this means that either the drug is a “double edge sword” with both better efficacy and higher toxicity than alternatives, or that it’s a “wolf in sheep’s clothing” that makes things look good for a while but causes changes that means relapse is swift and deadly. Lots to think about here.

Finally, SSC has a good post up on bias arguments here. I especially like his points about when they are relevant.

The Evangelical Voter Turnout that (Maybe) Wasn’t

There was an interesting graph in a recent New York Times article  that got Twitter all abuzz:

Visually, this graph is pretty fascinating, showing an increasingly motivated white Evangelical group, whose voter participation rates must put every other group to shame. I was so taken aback by this I actually did share it with a few people as part of a point about voter turnout.

After sharing though, I started to wonder how this turnout rate compared to other religious groups, so I went looking for the source data. A quick Google took me to this Pew Research page, which contained this table:

Two things surprised me about this:

  1. Given the way the data is presented, it appears the Evangelical question was asked by itself as a binary yes/no, as opposed to being part of a list of other options.
  2. The question was not simply “are you Evangelical” but “are you Evangelical/born again”.

Now from researching all sorts of various things for this blog, I happen to know that one of the most common ways of calculating how many white Evangelicals there are in the population is to ask people their denominational affiliation from a menu of choices, then classify those denominations in to Evangelical/Catholic/etc. That’s what PPRI (the group that got the 15% number) does.

For the voting block question however, they were only asked if they were a “White born-again or evangelical Christian?

Now to get too far in to the theological nuances, but there are plenty of folks I know who would claim the “born again” label who don’t go to traditionally “Evangelical” churches.  In fact, according to Mark Silk over at Religion News (who noted this discrepancy at the time), he’s been involved with research that “found that 38.6 percent of mainline Protestants and 18.4 percent of Catholics identified as “born again or evangelical.” So yes, the numbers may be skewed. It’s also worth noting that Pew Research puts the number of Evangelical Protestants at 25%, in a grouping that categorizes historically black groups separately (and thus is presumably mostly white).

So is the Evangelical turnout better than other groups? Well, it might still be. However, it’s good to know that this graph isn’t strictly comparing apples to apples, but rather slightly different questions given to different groups for different purposes. As we know slight changes in questions can yield very different results, so it’s worth noting. Caveat emptor, caveats galore on this one.

Fentanyl Poisoning and Passive Exposure

The AVI sent along this article this week, which highlights the rising concern about passive fentanyl exposure among law enforcement.  They have a quote from a rehab counselor who claims that just getting fentanyl on your skin would be enough to addict you, and that merely entering a room where it’s in the air could cause instant addiction. Given that it’s Reason Magazine, they then promptly dispute the idea that this is actually happening.

I was interested in this article in part because my brother’s book contained the widely reported anecdote about the police officer who overdosed just by brushing fentanyl off of a fellow police officer. This anecdote has been seriously questioned since. Tim expressed concerns afterwards that had he realized this he would have left it out. I’ll admit that since my focus was mostly on his referenced scientific studies, I didn’t end up looking up various anecdotes he included.

This whole story indicates an interesting problem in health reporting. STAT news has more here, but there’s a couple things I noted. First, the viral anecdote really was widely reported, so I’m not surprised my brother heard about it. It has never technically been disproven….outside experts have said “it almost certainly couldn’t have happened this way” but neither the police officer nor the department have commented further. This makes it hard for the “probably not” articles to gain much traction.

Second, the “instant addiction” part was being pushed by a rehab counselor, not toxicologists who actually study how drugs interact with our body. Those experts point out that it took years to create a fentanyl patch that would get the drug to be absorbed through the skin, so the idea that skin contact is as effective as ingesting or breathing it in seems suspect.

Third, looking at the anecdotes, we realize these stories are NOT being reported by the highest risk groups. Pharmacists would be far more likely to accidentally brush away fentanyl than police officers, yet we do not hear these stories arising in hospital pharmacies. Plenty of patients have been legally prescribed fentanyl and do not suffer instant addiction. The fact that the passive exposure risk seems to only be arising in those who are around fentanyl in high stress circumstances suggests other things may be complicating this picture.

While this issue itself may be small in the grand scheme of things, it’s a good anecdote to support the theory that health fake news may actually cause the most damage. While political fake news seems to have most of our attention, fake or overblown stories about health issues can actually influence public policy or behavior. As the Reason article points out, if first responders delay care to someone who has overdosed because they are taking precautions against a risk that turns out to be overblown, the concern will hurt more people than it helps. Sometimes an abundance of precaution really can have negative outcomes.

Asylum Claims and Other Numbers at the Border

There’s a lot in the news right now about border crossing, immigration and asylum claims, and I’m seeing all sorts of numbers being thrown around on Twitter. I wanted to do a quick round up of some numbers/sources to help people wade through it.

First up, every month US Customs and Border Patrol publishes the number of apprehensions they have at the Southern US Border and how that compares to the last 5 years. They do this relatively close to real time, we have the numbers for May, but not yet for June. If you want to know why you’re seeing so much in the news, take a look at this graph:

So with 4 months left to go in the fiscal year, we’re already 100,000 over the highest year on that chart. To give some context to that though, apprehension numbers have actually been relatively low for the last few years. They peaked at 1.6 million in the late 90s/early 2000s. However, there have been some changes to the makeup of that group….family crossings. Vox published this chart based on the CBP data that shows how this has changed:

I couldn’t find what those numbers were during the last spike, but it seems to be a record high.

So now what about asylum claims? Recently acting DHS Secretary Kevin McAleenan said the 90% of asylum seekers were skipping their hearings, but others were claiming that actually 89% show up. That’s a MASSIVE discrepancy, so I wanted to see what was going on.

First up, the 89% rate. The DOJ publishes all sorts of statistics about asylum hearings, and in this massive report they showed the “in absentia” rates for asylum decisions (page 33):

So for FY17, asylum claimants were at their decision hearing 89% of the time.

So where did the “90% don’t show up” claim come from? Reading the full context of McAleenan’s quote, it appears that he was specifically referencing a new pilot program for families claiming asylum. From what I can tell the pilot program is not published anywhere, so it’s not possible to check the numbers.

So is it plausible it jumped from 11% to 90%? I tend to doubt it, but it’s important to note the lag time here. The last published DOJ numbers are from FY2017, but those are for hearings that took place in FY2017. The average wait time for hearings in these cases for these cases is enormous….727 days so far in 2019.  These wait times are climbing, but if you toggle the graph around, we can see that the wait time back in 2015 was nearly two years:

So essentially those with decisions in FY2017 probably filed in FY2015. And a lot has happened to the stats since then. First, here are the number of applications over the last few years:

So compared to 2015, the number of applications have tripled but the number of approvals has barely budged. We don’t yet know what that will do to the percentage of people who show up, but it seems very plausible that it could increase the absentee rate. Additionally, because family migration is increasing so rapidly, it’s not clear what that will do to the numbers. Regardless, McAleelan’s reference was specifically to that group, so it was only a subset of the numbers that were previously reported. Still, 90% seems awfully high.

Complicating things further of course is the fact that this was a “pilot program”. That means it could have selected just one country or point of entry. One of the more interesting fact sheets from the DOJ site was the rate of asylum approvals by country. In the past few years, here are the top countries (page 29 of this report):

The rates of granting asylum from each of these countries were wildly different in 2018 though. Chinese asylum seekers were 53% granted, El Salvador was 15%, Honduras was 14%, Guatemala was 11%, Mexico was 6%. It seems plausible that a pilot program might have just been addressing those that arrive at the southern border, so it’s possible that individual countries have different profiles.

Overall, it’s clear that the data on this topic is worth watching.

Twitter and Sundry

I finally got a vacation this week, which means I spent not a lot of time near a computer but had a decent amount of time to scroll through Twitter. A few good accounts I found:

Any other good ones I missed?

Speaking of good accounts to follow, apparently one of the first papers focused on a prominent (and anonymous) Twitter account was published, with this one on the role of Neuroskeptic in calling out scientific ethical breeches. Only paper I’ve ever seen where the conflict of interest statement included “…was banned from commenting by Leonid Schneider at his blog “forbetterscience”.

Finally, Neuroskeptic also follows up on an analysis done by Susan Fiske of who research methods blogs go after and how often.  A few years ago Fiske had caused a kerfluffle when she referred to many science/research methods bloggers as “methodological terrorists” and defended the replication crisis. Her new paper looks at 41 bloggers and categorized which other researchers they talked about most often. Interesting findings: 3 out of the top 4 talked about researchers resigned (Diederik StapelBrian Wansink and Jens Förster), , and the #1 guy is pretty notorious. Other interesting note: there was no clear gender bias in who got talked about. Men were more likely to critique other men, and men who got named were posted about more often. However, it’s not particularly clear what this means about bias, as a few large cases of misconduct got a LOT of coverage, and Bem seems to have become the go to for bad methods/good researcher examples.

Still, super interesting work and good for Fiske for doing it. Gotta respect someone who gets mad, then gets data. I look forward to this being expanded on in the future.

Vaccination Rates by County, State and Around the World

There’s a lot in the news recently about vaccination rates, both in the US and in other countries. After seeing a few comments on Twitter this week about the topic, I got curious about the exact numbers.

First off, it’s important to note that for all the press coverage of the measles outbreak, vaccine exemptions are not evenly distributed across the country. For example, here’s just my state broken down by county. The map on the left is those with exemptions, the one on the right is those without exemptions not meeting requirements. Functionally, the vaccination rate is likely 100%-(the number on the left+the number on the right).


Now not everyone on the right may be missing their vaccinations, for some it may be paperwork. The Boston suburbs probably do well because most working parents have been providing vaccination documentation to day cares for years prior to this.

So how does Massachusetts compare to the country? Well, the CDC puts together a rather awesome interactive map for Kindergarten vaccination rates here. This map shows something interesting. Per the NYTs graphic of the measles outbreak, Washington, New York and Michigan have seen the largest measles outbreaks in the country. However, New York actually has one of the highest vaccination rates in the country at 97.5%. This suggests that it’s not just having a high vaccination rate, but where the unvaccinated are located that causes outbreaks. For example, in Massachusetts the exemption rates are highest on Martha’s Vineyard, a wealthy spread out community with not many year round residents. The other counties with higher rates of exemptions are not close geographically. In New York City though, it’s groups living in close proximity with low rates that are getting hit by the outbreak. This leads to high overall vaccination rates by state, yet still record setting outbreaks in targeted communities.

Now how about worldwide? Well, after working on their childhood vaccination program for years, Mexico actually now has a higher rate of children vaccinated against measles than the US (96% vs 92%). Interestingly, a similar number of people doubt vaccination safety in both countries. According to Our World in Data, France and Russia have the most vaccine skeptics:

Using Our World in Data again though, apparently vaccine skepticism doesn’t always correlate with not getting vaccinated. Both France and Russian have >95% measles vaccination rates.

I thought all this was interesting because overall, many people believe that fewer children are vaccinated than actually are. For example, Americans estimated that only 35% of the worlds children were vaccinated against measles, when in reality 85% are. While knowing that rates are high can sometimes convince people that it doesn’t matter as much if they get their kids vaccinated, thinking many people aren’t vaccinated can sometimes convince people that it’s not important to do so. “Everybody’s doing it” is an effective slogan for a reason.


Quote for Election Season

Life is a bit chaotic this weekend, but as we head towards what looks to be a rather drawn out election season, here’s a quote to keep in mind:

“The public’s ‘opinion’ on almost any issue will be a function of the question asked.”
-Neil Postman, Technopoly (as found on Twitter)

The exact wording here is important…”a function of” does not mean the question fully makes the opinion, but rather that it transforms it in ways that can often be predicted. A good thing to remember when you see headlines saying “the public supports ______” or “low public support for _______”.