When studying statistical workflow, or just doing applied statistics, we think a lot about propagation of uncertainty. Today’s post is about something different: it’s about propagation of responsibility in a decision setting with many participants. I’ll briefly return to workflow at the end of this post.
The topic of propagation of responsibility came up in our discussion the other day of fake drug studies. The background was a news article by Johanna Ryan, reporting on medical research fraud:
In some cases the subjects never had the disease being studied or took the new drug to treat it. In others, those subjects didn’t exist at all.
I [Ryan] found out about this crime wave, not from the daily news, but from the law firm of King & Spaulding – attorneys for GlaxoSmithKline (GSK) and other major drug companies. K&S focused not so much on stopping the crime wave, as on advising its clients how to “position their companies as favorably as possible to prevent enforcement actions if the government comes knocking.” In other words, to make sure someone else, not GSK, takes the blame. . . .
So how do multi-national companies like GSK find local clinics like Zain Medical Center or Healing Touch C&C to do clinical trials? Most don’t do so directly. Instead, they rely on Contract Research Organizations (CROs): large commercial brokers that recruit and manage the hundreds of local sites and doctors in a “gig economy” of medical research. . . .
The doctors are independent contractors in these arrangements, much like a driver who transports passengers for Uber one day and pizza for DoorDash the next. If the pizza arrives cold or the ride is downright dangerous, both Uber and the pizza parlor will tell you they’re not to blame. The driver doesn’t work for them!
Likewise, when Dr. Bencosme was arrested, the system allowed GSK to position themselves as victims not suspects. . . .
Commenter Jeremiah concurred:
I want folks to be careful in giving the Sponsor (such as GSK) any pass and putting the blame on the CRO. The Good Clinical Practices (GCP) here are pretty strong that the responsibility lies with the Sponsor to do due diligence and have appropriate processes in place (for example see ICH E6(r2) and ICH E8(r1)
The FDA has been flagging problems with 21CFR 312.50 for years. In fy2021 they identified multiple observations that boil down to a failure to select qualified investigators. The Sponsor owns that and we should never give a pass because of the nature of contract organizations in our industry.
Agreed. Without making any comment on this particular case, which I know nothing about, I agree with your general point about responsibility going up and down the chain. Without such propagation of responsibility, there are huge and at times overwhelming incentives to cheat.
It does seem that when institutions are being set up and maintained, that insufficient attention is paid to maintaining the smooth and reliable propagation of responsibility. Sometimes this is challenging—it’s not always os easy to internalize externalities (“tragedies of the comments”) through side payments—but in a highly regulated area such as medical research, it should be possible, no?
And this does seem to have some connections to ideas such as influence analysis and model validation that arise in statistical workflow. The trail of breadcrumbs.
> “tragedies of the comments”
Maybe I’m missing something but is that spelling intentional?
Or is “tragedy of the comments” a clever way of referring to poverty of the imagination?
> I agree with your general point about responsibility going up and down the chain…
I think one issue there is determining how far up the chain to go.
A some point, lawyers earning enormous incomes by helping pharma executives protect even enormously larger incomes by shirking accountability for criminal fraud (or even just bad science) doesn’t just fall down to individuals or even law firms or corporations or law enforcement entities or our political institutions.
Doesn’t accountability at some point have to rise up to implicate the neo-liberal superstructure where those at the very top don’t think there’s a price to pay for easy money and an explosion in asset values – under the umbrella philosophy that a rising tide lifts all boats and a “free market” and basically unfettered capitalism is better than anything else as proven by a global rise in standards of living.
It all works until it doesn’t work. Maybe at some point the chickens will come home to roost. Maybe de-coupling growth at the top end from a stagnant bottom end and a dropping middle has a limited lifespan, and maybe we’re beginning to the evidence of such now in the stock market and crypto-currency and inchoate public frustration that seems to focus indiscriminately in pretty much all directions including, not in significantly, each other?
Maybe focusing on crooks or perverse corporate (or academic) incentives isn’t enough and we have to look higher up the chain?
+1
-1
The people who commit crimes should be held responsible, regardless of where they are in the “superstructure”.
The assumption that employees at the bottom level are by definition innocent and taken-advantage-of, and that management is always responsible because they unduly encourage lower-level people to cheat (or whatever twisting of reality advocates try to apply), is just plain BS. I’ve seen first-hand that lower-level employees cheat frequently, often at the company’s expense, even though they may already be very well compensated. Sometimes they cheat for the simple stupid reason of just being lazy, sometimes to make more money, sometimes to screw other employees…the list of reasons is endless.
I’ve seen management cheat too, and they should be held accountable when they do. But a big reason the system of gigging and contracting has come about is because it’s almost impossible to fire problem people for cause.
BTW, if anyone cares, I’ve commented in the past as “anonymous”. But there are so many anonymi that I suppose it’s appropriate to choose a pseudonym, so people can direct their comments at the appropriate source.
chipmunk –
> The assumption that employees at the bottom level are by definition innocent and taken-advantage-of, and that management is always responsible because they unduly encourage lower-level people to cheat (or whatever twisting of reality advocates try to apply), is just plain BS.
I guess from that statement you think I’ve made such an assumption?
Do you mind pointing to what specifically it was that I said that led you to that conclusion?
It’s not the case that I hold any such belief.
To restate more concisely: I think that it’s not enough to merely assign blame at the individual or institutional levels. I think there’s a more systemic problem in play, where accountability is an undervalued commodity at the societal level.
I often find that in online comments, if someone has misinterpreted my views, when I offer clarification they don’t acknowledge that clarification.
But it’s particularly notable in a comment thread on the topic of responsibility.
Why do people comment if they aren’t interested in good faith discussion?
Joshua:
Speaking in general and without reference to this particular case: sometimes we get misinformed commenters that I or someone else respond to. Here I’m talking about a factual error or trolling or someone flat-out missing the point; I’m not talking about difference of opinion. Anyway, five things that can happen:
1. The commenter explains further, and it turns out that he or she had a good point after all.
2. The commenter apologizes for the error or for trolling or whatever.
3. There’s some intermediate outcome: the commenter got something wrong but still has a valuable point after clarifying the error.
4. The commenter keeps arguing and won’t let go.
5. The commenter doesn’t return.
Option 5 is perhaps the most common. Of course my favorites are options 1, 2, and 3, but option 5 is better than option 4, no? I think sometimes these commenters are expecting a twitter-like flamewar and are kind of disappointed that we respond calmly.
Andrew –
With respect to the overall quality of the blog, if nothing else, 4 is the worst.
But I just don’t get why people write a response and then don’t follow up if you clarify a misunderstanding. What is their objective?
“I think sometimes these commenters are expecting a twitter-like flamewar and are kind of disappointed that we respond calmly.”
Probably giving yourself a little too much credit – maybe you should mind your assumptions :)
Maybe the commenter who didn’t come back:
1) wants to make a point and doesn’t want to bother with the original commenter’s endless dodging and hiding and trying to rearrange the argument or drag it into the weeds on another topic – a famous path of many discussions on this blog.
2) has other matters to attend to
3) doesn’t want to explain the wheel – people here seem to have a broad knowledge of data sets and Bayesian modeling but a very limited knowledge of the actual world – some of which **gasp** might not be represented in their data set – which they’re surprisingly resistant to acquiring.
But as Phil likes to say: “It is difficult to get a man to understand something when his salary depends upon his not understanding it.” Data analysts will analyze data and offer their conclusions, sensible or otherwise, because they want to get paid.
4) doesn’t want to go over fundamentals like assumptions for the umpteenth time to the experts who are supposed to know them.
5) doesn’t care that a random commenter somewhere on some blog is upset that his comment may have been misinterpreted
chipmunk –
Thanks for the response.
Earlier, I asked you for a direct response. I’ll repeat:
I guess from that statement you think I’ve made [the assumption that “employees at the bottom level are by definition innocent and taken-advantage-of, and that management is always responsible because they unduly encourage lower-level people to cheat (or whatever twisting of reality advocates try to apply)?”]
Do you mind pointing to what specifically it was that I said that led you to that conclusion?
It’s not the case that I hold any such belief.
Now determining responsibility lies for your misinterpretation of what assumptions I made (if I’m right and your comment about “BS assumptions” was directed at me) is kind of hard and doesn’t ultimately matter anyway. But regardless, I think it’s a sign of good faith discussion if, when asked to respond directly to something, an interlocutor responds.
Please clarify if your statement about “assumptions” was directed at me and if so, acknowledge my clarification that I hold no such assumption. At that point a productive discussion could continue.
Chipmonk,
I think you are dodging the point. Clearly the doctors who do the fake trials should be held responsible for what they did, but the question at hand is, what about the people who set up the system that motivates the cheating? In this case, according to the article, the lawyers involved are more interested in protecting the companies funding the cheating than in stopping the cheating.
To make what will seem to you like an off-the-wall argument, I’ll claim that one way to deal with this sort of thing is to return the income tax structure to something like what it was when I was young, when the top marginal tax brackets were high enough that as some point, increasing your pre-tax income wasn’t worth the cost to your self-respect.
+++++1
This may be slightly off topic, but the propagation of responsibility fits this perfectly legal paradigm that is all too familiar. If you run into an issue streaming anything (this applies more generally to almost any computer issue, but streaming is perhaps the best example), try figuring out if it is your hardware (including computer, router, modem, video device), software, network provider, or streaming service. Try calling each so that they can tell you the problem lies somewhere else.
+1
Looks like maybe you tried watching something on Apple TV?
To boil it down even further, try calling when you have an issue, say, with your phone service and you get passed around endlessly in recursive loops between different departments WITHIN A SINGLE CORPORATE ENTITY.
And that’s assuming when you first call customer support, you can get out of a loop of prerecorded messages because you can’t phrase your problem in such a way that an AI-inspired voice recognition software will interpret it as intended.
Check out https://gethuman.com/
That might at least get you to a person rather than an AI to pass the buck.
Apple is a good example here because you’re running into the problem where Apple cannot own the whole stack. One of the reasons to buy into the Apple ecosystem of vertical integration, from the very transistors on the CPU to the final Apple TV Oscar-winning series you are watching, is so you have one entity to blame when something goes wrong: Apple. You neither know nor care where in the stack something goes wrong, and it is not mathematically possible for you to care less about which division blames which for the glitch in the TCP stack that broke your stream. In fact, you cannot know where to assign blame, and Apple itself might be unable to. They are supposed to get the job done, and you will not pay if they do not. (Conversely, them being able to get the job done where other things so often fail to is also why you pay them so much when they do.) So, they will figure out some way to make things work, which may not involve blame-assignment and buck-passing at all, but finding another blackbox they can buy from (which itself may be purchasing blackboxes).
Similarly, there is no reason the person ordering dinner should have to consider pizza delivery logistics. They should just get a refund from Uber. And then Uber will either figure out an answer, like a fix to its planning software or firing the delivery driver in question, or it won’t, and it’ll go out of business and be replaced by something that will do it right.
This kind of data fabrication is very hard to discover. These people don’t want to be caught. Unless they are lazy, you won’t catch them. I worked on a study where a data coordinator at a clinic sent the same EKG for every patient. It was only by luck that a medical resident discovered it. Sadly, this is a fact of life. Just like faked results of all kinds. We obviously don’t like it and it is very dangerous. But, blaming GSK for lawerying up seems silly. The FDA enforcement actions are legalistic. Of course, they need lawyers.
Here’s my favorite example of responsibility-passing: Tesla. It’s “self driving AI” requires the human driver to have their hands on the wheel at all times. When an accident happens, either the driver’s hands are on the wheel or not. In the first instance, “self driving” can’t be at fault since the human is in control; in the second instance, “self driving” can’t be at fault since the hands should have been on the wheel. Thus, “self driving” is never faulted. Did some law firm come up with this?
It gets even more suggestive
https://static.nhtsa.gov/odi/inv/2022/INOA-EA22002-3184.PDF
I’m curious if you’ve read Leonid Schneider’s post on Retraction Watch a few years ago entitled “What if universities had to agree to refund grants whenever there was a retraction?”
In it he argues that research institutes should bear some financial responsibility for the research carried out under their roof, and could be forced to return funds back to granting agencies in the event that resulting work is retracted.
Dace:
The trouble is that there are very few retractions so this wouldn’t do much.
I suspect there would be even fewer retractions under that proposal. If the financial burden was severe, there would be great pressure to not retract articles – more than the inertia that is already built into the system. I’m thinking that editors will then be wondering, “well it was a mistake, but was it really that serious?”