Declining to “Respect Your Opinion”

When is it reasonable to decline to “respect” somebody else’s opinion? I’ve done this before. I have two specific examples in mind, and in both cases I could not bring myself to say or type the words, even though I felt like some sort of olive branch was being extended to me. This is true even though I respect most of the people I know as people. You can be a good person, a loving parent, a hard worker, and I deeply respect those qualities in my friends and acquaintances. I just might not particularly respect their political commentary on topics they might be ignorant or dogmatic about.

I don’t think I’m being completely unreasonable or prickly here. If someone has at least a plausible argument in their defense, I usually respect their opinion. But there is a bar to clear. If someone adopts a position that I find morally repugnant, I don’t respect their opinion (unless it is particularly well argued or somehow interesting, and even then it is the approach and not the conclusion that I respect). If someone utterly fails to bring any argument to bear on the discussion, I probably don’t respect their opinion. If someone just asserts “The earth is the closest planet to the sun” and has no response when presented with contrary information, I feel no obligation to respect that. Some political arguments are literally this terrible. Likewise, if the arguments presented are just bad, and the presenter of these arguments refuses to budge after their total dissection, I don’t feel obligated to respect their opinion. If any of this sounds really snobbish, try imagining yourself saying “I respect your opinion” to a neo Nazi, or an intransigent child making a factually incorrect claim, or a person adamantly claiming “2+2=5, because my hair is a bird.” Everyone has a bar to clear. In fact, depending on how repugnant or how poorly argued your interlocutor’s position is, you may well say, “I don’t respect your opinion, and I don’t particularly respect you either.” 

Learning the Front Handspring

I thought I’d do a post about a physical skill I’ve been learning: the front handspring. I always wanted to learn those cool gymnastic moves, and now at the age of 35 I’m finally doing it. Sort of. It’s been a long, exhausting, frustrating journey but I finally have a serviceable front handspring.

This won’t be a tutorial; there are plenty of good ones out there. Watch a few Youtube tutorials (here is my favorite, but I've watched many others) and read this Crossfit document and this series of three blog post by a gymnastics coach. I had to watch every tutorial I could find about fifty times and re-read the blog posts and Crossfit document about a dozen times to catch all the mistakes I was making. So have your library of tutorial videos and reading materials ready to go.

This post will be more of a motivational and trouble-shooting post, perhaps filling in some holes or emphasizing different problems than the tutorial videos.

First of all, get a yoga ball and maybe some gym mats. If you have a very soft carpeted floor (what I started on), that might work, but it’ll be murder on your knees, hips, and back to do handsprings repeatedly on a hard surface. You can also try it in the grass outside. Position your yoga ball somewhere on the floor, take a hurdle step toward the yoga ball, kick up into the tallest handstand you can manage with your hands placed just in front of the ball, and fall forward over the yoga ball. The ball will serve as your spotter and bounce you up onto your feet. Trust me, you won’t get bounced off in some odd direction. You’ll bounce up in pretty much the direction you were headed. Take a deep breath and try. Then try a few more. You’ll get used to going over yourself, which is an important confidence builder. You'll probably start off flopping onto the ball at the small of your back and bouncing off with most of your weight. This means your arms aren't straight enough. As you progress, your arms will straighten until you're barely grazing the ball on the way over, probably grazing with your shoulder instead of your hips. 

Now record yourself doing a few. Review the tape, then review the tutorials on Youtube. On an iPhone you can drag yourself through the video at any speed you like by sliding your finger at whatever speed and freeze on the frames you need to. Notice everything you’re doing wrong. You’re probably bending your arms too much when you’re upside down with your weight on your hands. This is wrong; your arms are supposed to be straight. You’re probably reaching down to the ground with your arms. This is wrong; your back leg is supposed to kick up and drive your hands to the ground. Your arms and kicking leg are supposed to be in a straight line, as if you were one of those drinking bird toys dipping down for a drink. Your forward leg is supposed to start bent, then spring to give you some lift just as your hands touch the floor. Think about this as your lunge leg is helping to spring you up into a really tall handstand, it’s just that you’re going to go over yourself rather than staying up. You’re probably tucking forward, trying to look for the ground in front of you. This is wrong. You’re supposed to curl your legs under you and arch your back powerfully, like doing a back-bridge in midair. Lie on the ground for a moment. Try tucking your chin forward and doing a back bridge, then try bridging with your head flexed back. It’s a lot harder to arch your back with your chin tucked forward, but in order to land properly you need this arch in your spine. When you go over, you should be looking backwards and down at your hands, not trying to look forward to spot your landing. Then, there’s the “block” with the hands, where you push yourself up off the floor. Make sure you’re doing this at the right time.

I went through this about a thousand times, looking at my terrible front handsprings in slow-motion on my iPhone video and trying to spot my errors. I noticed that I was bending my left arm while I was upside down in a handstand. To correct this error, I had to make sure I was jumping off my lunge leg, which gives you upward momentum. The force of my back-kicking leg was driving me down, so I was forced to bend my arm to absorb the impact. I was indeed tucking my chin on the way up, rather than looking back so that my back could properly arch. As a result, I was landing very hard on my heels with my knees bent and my weight far behind my feet, rather than landing more softly on the balls of my feet. Doing it the wrong way is murder on your joints, so if you have a problem here fix it fast. I repeatedly missed one or another of these steps and had to record myself to see what I was missing. I repeatedly drilled the one part of the movement that was missing, and then promptly forgot another piece. It’s frustrating as hell, because I could do them fine one day then forget how to do it the next. After several days of messing up badly, I went back to doing them over the yoga ball, which would briefly fix my problem before some other problem arose. But I think I’m finally at a point where I’m reasonably proficient in this move. It took about three months of practice, but I got there.

If possible, get a good coach and schedule a few practice sessions under their tutelage. They’ll be able to correct things that you don’t see yourself doing, suggest useful drills that address your deficiencies, and tell you if you’re doing the move safely or not. I really wish I had had a few sessions with a good gymnastics coach, but it was fun trying to learn it on my own. 

Private Drug Prohibition is Adequate. Government Enforced Prohibition Unnecessary.

 I’m not opposed to all forms of drug prohibition. There are some forms that I like and that I believe to be necessary. But in those situations where prohibition is called for, it will virtually always be enforced privately. I’m not talking about some shadowy private security force doing armed SWAT raids on civilian residences looking for drug stashes (think Black Water, the private military company, or Lone Star Security from the game Shadowrun). I’m talking about boring everyday life making drug use an impossibility for most normal people.

Most people spend around half of their day in a restrictive institution of some sort. Most adults go to work. Most children go to school. Even outside of these environments, you may go to a store to do some shopping or go to a restaurant to eat. Many of these environments make drug-induced intoxication unfeasible. You could get drunk and go to work, but you’d most likely be found out and severely punished or, if it’s a repeat offense, almost surely fired. Likewise, if you go to a grocery store rambling incoherently and bothering the other customers, you will most likely get ushered out. Work and school are going to be far less forgiving than the restaurant or the store, but virtually everywhere you go there will be something constraining you from full-on drug intoxication. Even if we assume that workplaces and schools aren’t very good at spotting and punishing drug use, there are intrinsic penalties for these behaviors. Your work will suffer until you eventually get fired or kicked out of school, or at the very least you will not prosper as much as you otherwise could. Even if your boss is oblivious to your high-functioning alcoholism or your methamphetamine habit, if your work suffers your boss will eventually notice that.

Supposing your drug use doesn’t affect your work, there’s no cause for concern anyway as your habit isn’t causing a problem. If someone is very privately using intoxicants in a way that doesn’t affect their work or school, it’s hardly any of their concern. In this sense it’s unreasonable to do drug testing; your boss or school can observe your work product directly and decide whether it’s adequate. With the caveat that some drug use might have no observable effect on day-to-day work output but slightly raise the risk of a catastrophe (think an airline pilot using cocaine), directly observing someone’s work is the sensible solution. Testing for something that isn’t causing an obvious problem is a waste of time.

I can feel your objection welling up. “But…people do drugs anyway and ruin their lives. Clearly private deterrence is inadequate, right?” And to that I say, You’re missing the point. Yes, some people feel the bad consequences of their destructive drug habits. That is how incentives work. The people who behave irresponsibly are hurt by their irresponsibility. Observe that most people don’t have self-destructive drug habits. They see ahead of time that there is an enforcement mechanism in place and decide not to, say, smoke weed before reporting to work, or take heroin before sitting though a class lecture. Perhaps the negative consequences are so obvious that the idea of starting a drug habit doesn’t even occur to most people. It’s not a conscious decision not to smoke a bowl in the morning; the need to get your kids to school and get yourself to work rules that completely out of the question. For most people. And the guy who does try it goes to work smelling of their obvious habit, unable to focus, and gets punished or fired.

Perhaps I haven’t answered the objection fully. The argument is that government enforced prohibition enhances the already-existing (if sometimes tacit) private enforcement of drug prohibition. Does it make sense to go after those remaining screw-ups for whom the implicit penalties for drug use are inadequate? Not really.  Not at all .  It simply does not make sense to try to penalize people out of harming themselves, particularly when the demand for the drug is inelastic (as it must be for any self-destructive drug habit). The intuition here is that I have to harm you more than the drug harms you to get you to stop. Even admitting that a few potential drug users are deterred, the harm to the remaining users is so exacerbated that it doesn’t justify the costs. This is an outcome of very straightforward economic reasoning, requiring no actual grounding in economics or assumption of weird econ theorems or anything. A bit of logic alone will get you to this answer. Unless you posit some very strange, stilted assumption about how drug users respond to legal penalties much more strongly than implicit penalties (of comparable magnitude), you basically have to accept this conclusion: drug prohibition does more harm than good. It’s completely unnecessary. But such deterrence as everyday life naturally provides keeps most of us clean most of the time.

If you’re going to bring up harm to third parties (intoxicated motorists, neglected children, alienated friends, etc.), I answer that point here. In short, the externalities are 1) grossly exaggerated and 2) they are already internalized to the extent that the drug-induced misbehaviors are already illegal. I wish the prohibitionists would take their own arguments a little more seriously. My impression, having thought this through, is that they don’t really have a leg to stand on but they don’t have the patience to think the arguments through to their conclusions. 

No Moral Trump Cards in Health Policy

Apparently some people think you can answer questions about health policy with their moral outrage alone. At least, that’s the impression I get from my Facebook feed. I see arguments every single day of the variety, “It’s just the right thing to do…” or “It’s just wrong to let someone go without healthcare” or (and this one is a paraphrase from memory, but it may be the exact language) “It’s called being a fucking decent human being!” These are bad arguments. They aren’t even arguments, really. These are simply assertions by the speaker (or writer or meme-sharer) that they hold the moral high ground and will make pronouncements thusly. (At the moment I’m talking about the potential repeal of the ACA, but I could be talking about any public flare-up over health policy.)

I’m sorry, but you simply cannot draw conclusions about health policy without being a little bit analytical. How much would it cost to save one quality-adjusted life-year (QALY, a standard unit of relative health)? Is this close to what most people objectively value a year of life? Is it an order of magnitude higher, thus rendering a strong verdict against the health policy in question? Is it an order of magnitude smaller, thus rendering a strong verdict in favor of said health policy? Based on our best evidence, might the health effects be *negative*, thus making the whole policy morally dubious? How are other values, like freedom of choice and property rights, to be weighted against a policy that forces you to purchase something? If the cost per life saved came out to something absurd, like a trillion dollars, would you still favor it (as a “moral trump card” argument would commit you to doing)? Is there massive uncertainty in even our best estimates of the benefits, thus making it impossible to justify the enormous expense of the law? To play a moral trump card is to dodge all these important questions.

People often use the following hypothetical: “If we can save the life of an uninsured man with no resources for a million dollars, aren’t we obligated as a society to do so?” This is the equivalent of making about 20 median households work an entire year for the sake of one person. (20 households x $50,000/year income per median household.) You can try to make this sound noble, but to my ear it sounds like we’re enslaving 20 households for a year for the benefit of a single person. I’m not trying to support any particular answer to the question posed by the hypothetical (mine is “No”). I’m simply pointing out that there is a trade-off of rights here, and it isn’t clear a priori whose rights should dominate. I think even the moral trumpists on my Facebook feed would balk at the idea of spending a billion dollars to save a life, or the idea of *literally* enslaving 20 people, conscripting them to work exclusively on the task of saving one man’s life. In fact I think they’d balk at the idea of enslaving even *one* person for significantly shorter than a year, even if they objectively valued the man’s life higher than the conscript’s freedom. If a slight re-framing of the hypothetical causes you to answer differently, then it’s no more than a superficial rhetorical trick.

You can continue to grandstand on some all-trumping moral principle, but if you're playing that game your opponents can simply grandstand on some other moral principle that you are neglecting (freedom of choice, property rights, etc.) and they would be on equal footing. There is no framework for discussing anything in this kind of exchange. However, if you engage a little bit with the concept of trade-offs, if you are a little bit empirical about your assessment of government policy, if you employ a little bit of numeracy and mathematical thinking to compare the relative values in conflict, you can have a reasonable conversation about the topic under contention. 

Here, Let Me Think That Through For You

You know how sometimes people ask a silly question that’s easily checkable with a Google search? Or make a factual claim that is easily correctable with such a search? “Here, let me Google that for you” is a wonderful snarky response to this behavior. Sometimes it’s a technologically incompetent person who really doesn’t know how to find the information they want, and sometimes it’s a technology native who just forgets that he has all the world’s knowledge at his fingertips. Either way, it’s fun to remind people that there’s an easy solution to their problem.

A different version of this problem is the following. Someone blurts out whatever arguments come to mind without really thinking about them, usually done in rapid-fire succession. It may be someone who has very poor impulse control and no filter. Or it might be someone who is losing an argument dead-to-rights and is flailing in panic. Whatever the reason, some people are prone to producing this kind of blather. These interactions can be interesting, as this sort of free-association might by luck hit upon a nugget of truth. And a thoughtful person can benefit from thinking through someone else’s bad ad hoc arguments and dissecting them. It’s like a game of “’Why?’ Boy”, in which a child keeps asking “Why?” to each response in the iteration and you have go deep down to the foundations of your knowledge. You learn something from this process, even if the ‘Why?’ boy doesn’t.

Mostly, though, this is a huge distracting waste of time. I wish that people would think through their arguments a little more clearly before sharing. Don’t tax the patience of your correspondent with half-baked ideas. I realize there is some ambiguity as to whether this is happening or not. (After all, isn’t there another person on the other side of the argument who thinks you’re doing the same thing?) But there are some tell-tale clues. If you present someone with novel, relevant information that they haven’t digested and they have an immediate response, they are blathering. If someone makes a claim of a statistical nature but can’t offer any numbers (or even a hint of numeracy), they are blathering. If someone hasn’t thought through your argument but blurts out the first objection that pops to mind, they are blathering. Beware of people who are too quick on the draw with the “post” button.

Examples help illustrate a point, so I’ll offer some recent ones. The purpose of this isn't to revisit the argument or embarrass anyone, just to focus on something specific. The other day I got several very absurd responses from someone who clearly wasn’t thinking. It was a discussion of health policy and how to evaluate whether it’s working or not. This person asserted several things that made no sense. His claim was that it would take a generation to evaluate the effectiveness of a health law. On the contrary, if you’re looking for the effects of a recent policy change, you look *closer* to the time of the policy change. A longer timeline allows more confounding factors to dominate the trend. Another bizarre claim was that it’s hard to evaluate policy that’s applied to millions of people. True enough, but the *size of the sample* is not the issue. A larger sample makes small effects *easier* to see, not harder. I said exactly this, and my interlocutor disagreed directly with this statement. No, it’s difficult to analyze policy because of *confounding factors*, i.e. all that other stuff that’s going on at the same time. A true ceteris paribus comparison isn’t possible. Perhaps this person actually had this in mind and did a bad job of explaining his argument, but then this would have been contrary to his claim that we’d have a better estimate of the benefits if we waited a generation to see the results (a generation being a long enough timeline for a lot of confounders to build up). This person seemed to have gotten some basic principles of statistics and social science backwards. (A larger sample size is better, and an effect that is *closer* in time to the policy change is easier to attribute to the policy.)

I presented some information about randomized controlled trials in which lots of people get free healthcare while a control group doesn’t, and the “free medicine” group doesn’t appear to get any healthier. (See the RAND healthcare experiment and the Oregon Medicaid experiment if you want to know more about this.) This evidence bears directly on the policy question. What’s the sense in giving away a lot of extra medicine if there isn’t an appreciable health benefit? I don’t know if he was familiar with this body of scholarship. He didn’t say. He simply asserted that the health effects would be real but too small to measure.  I cannot prove but strongly suspect that this was a case of someone rejecting out of hand *extremely* relevant information for the item under discussion.

Now, “The health effects are real but too small to measure” is a statistical claim. It’s not just something you can blurt out. The exact size of “Too small to measure” can be determined using statistics. (Actuaries like myself use something called "credibility theory" to determine how much trust to put into a dataset, depending on the sample size and strength of signal vs. noise.) It’s possible to calculate how small an effect would be “too small to measure” and perhaps show that it’s absurdly costly to save a single life. Would anyone want a health policy that cost, for example, a trillion dollars per life-year saved? A million dollars even? I wouldn’t, and most people don’t objectively value their lives that much, as determined by willingness to purchase safety features or take dangerous jobs (to list two ways that economist try to "value" a human life). Something that should have been an explicit calculation was instead just asserted. I could have just as easily asserted that the health law had *negative* health effects that were too small to measure, and we’d have been on equal footing. As it happens, I *did* open up an Excel workbook with some mortality-by-age data in it. Very crudely, I convinced myself that you’d need to see a few percentage points change in mortality for very young people, about a 1% change in mortality for 50-year-olds, and less than a 1% change in mortality for 60+ year-olds, to see a statistically significant mortality effect. That should be a starting point for anyone offering a “the benefits are too small to measure” kind of argument. My “analysis” was totally back-of-the-envelope and is probably wrong for various reasons; someone else might get another answer for what “too small to measure” means. My point is that you have to do some work, you have to use figures (real or at least made-up-but-plausible), and you have to use some math when you’re making a statistical argument. It’s inconsiderate to just assert something and make someone else do the legwork for you. “Here, let me think that through for you.”

If my interlocutor is reading this, my apologies for using you as a foil, but this isn’t about you. I’m not trying to be gossipy. The point I’m making has nothing to do with any one person or any particular argument. It just helps to have an example at hand, and this one serves nicely. I have probably wasted way too many hours of my life thinking through bad arguments left on my FB page. A healthier habit might be to simply ignore distracting blather and only communicate with people who state their arguments clearly. It bothers me to think that a bad argument can just be allowed to stand. Someone might see it and think it's superficially plausible. Unfortunately this puts me in a position of sometimes arguing with thoughtless people and reasoning through their arguments for them, and I always feel a little bit cheated when I get sucked into one of these discussions. As a clever friend of mine put it, the thoughtless commenter has a huge cost advantage over the thoughtful one. 

Healthcare Policy Changing. Falsifiable Predictions Wanted.

When the ACA first passed, I remember posing the following challenge: “Ok, ACA supporters, you got your law. Now what’s going to happen? How will the benefits manifest themselves? Make a prediction that we can check. In a few years you should recant if you’re way off. I may take a bet against your prediction, if it’s well-specified enough.”

It would have been interesting to hear people’s actual responses to my question, assuming anyone had thought about it. I was fishing for something like, “Mortality rates will fall for poor minorities by X deaths per 100k population,” or “Diabetes rates will decline by Y%, high blood pressure by Z%, among the currently uninsured population.” The point here is that a healthcare law should somehow affect health, and if it doesn’t noticeably do so it has failed as public policy. It’s fake medicine. The people who were singing the praises of Obamacare should have been willing to make some kind of falsifiable prediction the moment it passed. I took their unwillingness to do so to imply a lack of serious thinking. I never got a meaningful response, nor did I see any prominent pundits or bloggers independently come up with the same idea.

(By the way, it’s not enough for someone to point out that an improving trend has been at work since the law was passed. It could be a pre-existing trend that can’t possibly be attributed to the law. In fact, life expectancy has risen since 2000 so you have to be really careful about attributing any improvements in health to a bill passed in 2010. One has to show that any pre-existing trend has improved above and beyond the trend-line.)

One might have answered something like, “The world is really complicated, and a lot of things affect the overall health of any population. I’m unwilling to make such a prediction.” This would have been an interesting admission. If the major supporters of the ACA thought that the benefits are too small to measure, or so small they would be swamped by noise, they should have at least said so. Some of us think it's unwise to waste massive resources in pursuit of benefits that are speculative or invisible. 

One might have given another sort of answer, something like, “The law would work if implemented, but stupid Republican states will fail to implement it and stupid Republicans in a future Congress will hamstring the law.” I can imagine someone being tempted to issue this hedge, but again this is something that needs to be stated ahead of time. The person giving this kind of answer is predicting failure, and failure based on something endogenous to the system. If the failure of public policy is that predictable, we should oppose such policy. Your political initiative might have worked if it weren’t for that incorrigible opposition party, just as your lunar program might have worked were it not for stupid gravity!

Today it looks like some kind of repeal is imminent, and doomsayers are predicting some kind of blood-bath as people lose their insurance coverage. I seriously doubt it, but I am willing to hear from anyone capable of discussing this at room temperature. What *precisely* do you think will happen?

My own view is that health policy doesn’t matter all that much in terms of getting actual health outcomes, although bad policy can certainly be very expensive and saddle us with enormous burdens. The social science is pretty clear on this point. An individual’s “insured status” has little correlation with health after you’ve made the appropriate demographic adjustments; it simply isn’t true that insuring someone grants them “access to healthcare” in a way that makes them healthier. (Or more precisely, a population of such people won’t get healthier; any one such someone might get healthier or sicker. But it’s the population effect, not some individual’s health outcome, that tells you something about causation.) I have a long reading list for anyone who doubts this. Start with Cut Medicine In Half by Robin Hanson, and do go on to read the entire discussion. (Cato Unbound, where the essay was hosted, is a forum. In this particular one there are three other health policy experts who, while they never actually say Hanson is wrong, take issue with his claims.) Read In Excellent Health by Scott Atlas; here is a podcast of him discussing the book on Econtalk. Also check out Overtreated by Shannon Brownlee, Catastrophic Care by David Goldhill, Crisis of Abundance by Arnold Kling, Affordable Excellence: The Singapore Healthcare Story by William A Haseltine, Priceless by John Goodman, and I’m sure I’m forgetting a few others. Also listen to any episode of Russ Robert’s Econtalk that talks about healthcare, especially anything with Arnold Kling or Robin Hanson. Most people balk at the claim that people don’t get healthier when you give them a bunch of free medicine, but I’m on very solid ground here. Go review the literature a little if you're skeptical. I'll wait. Given all this, the obsessive fixation on getting healthcare to “the poor” is misguided. This is mostly not a fight about access to healthcare, but rather about who pays for what and how. 

My own prediction would be something like: Policy tweaks of the "pass/repeal the ACA" kind won't noticeably affect health outcomes (although much larger restructurings might). Policy tweaks that cause patients to face a larger proportion of the bill would result in cost savings with no measurable effect on health outcomes. I could try to be more specific if a specific proposal is on the table. But the people who are predicting disaster from the repeal of a seven-year-old healthcare bill look positively daffy. In terms of evidence-based policy, they don't have a leg to stand on. 

Pet Peeves of a Data Analyst

[Edit: I wrote this a few days ago then reread just before sharing. It comes off as more sneering and angry than I would like, though I was mostly in a good mood when I wrote it. I'm sharing anyway, recognizing that it could probably be written in a way that's more understanding to the peeve-inducers. I tried, when I could, to be constructive rather than abusive.]

I work with large datasets for a living, doing all manner of statistical analysis. My work ranges from simple summary statistics to model building and feature engineering on big datasets. I thought I’d list a few pet peeves of mine.

1)      Fake Precision on Noisy Data. Someone always pipes in with “Did you adjust/control for X? (Which happens to be a hobby horse of mine?)” Sometimes this is welcome, but sometimes the dataset is too small and therefore not credible enough for refined adjustments to matter. Sometimes the “analysis” is basically fitting a line through some random dots. If you had infinite data, the dots would *not* be random, but would be a noticeable pattern of some sort (linear or otherwise). But you have to work with the data you actually have. If your data are too sparse to see the real pattern, any “adjustment” you make, even if there’s some good theoretical reason for making it, isn’t going to matter. It just means you’re fitting a line though *this* set of random dots rather than *that* set. I’ve also seen maddening arguments about whether some set of points should be fitted with a line or a curve or some other sort of grouping, when in reality the data points are too noisy to make any such determination.

2)      Last Minute Demands on a Big Dataset. It’s often said that data modeling is 90% data gathering/cleaning and 10% model building. So it’s a huge headache when someone has a bright idea for a last minute insertion. Sometimes this is the fault of the modelers, but usually it’s wishy washy management deciding at the very last second that something they just thought of just now is very, very important. “Wait, will we be including social media history in our analysis of auto accident frequency? I didn’t see it in the list of variables. Let’s add it!” The data modeling people sigh at these kinds of requests, because it usually means a few days of additional data gathering and a delay in a (perhaps already determined) modeling schedule. Data modelers should keep lines of communication open and set some kind of “no further adjustments” date so that this doesn’t happen. But it probably will anyway.

3)      “Well, I did this at my former company…” Sometimes a person will move from a large company with millions of customers and a huge dataset to a small or medium-sized company. They often bring with them unrealistic expectations. With larger datasets it is easier to see very weak patterns that might be invisible on a smaller dataset. The pattern may be real, but you won’t see it in your data. At my company, you have to assemble several years' worth of customer data, in countrywide aggregates, to see real patterns. A much larger company might have 20% of the adult population as its customers, and can thus see real patterns in a single quarter’s worth of data in a single state or region. If a manager gets used to the latter, they will have unreasonable expectation if they moved to work at a smaller company. Sometimes this might be a matter not of data size but of expertise. If your previous company had three data science PhDs, a team of actuaries, and a bunch of SQL experts, but your new company has a bunch of Excel jockeys who barely know how to run an Excel regression, you won’t be able to implement all your awesome ideas at the new company. At the very least, it may take some time to develop the appropriate skill set.

4)      Ignoring the modeling results and expecting there to be no consequences. Data models sometimes give us surprising results. That’s why we do them. If we knew all the patterns ahead of time, there would be no need for the modeling. But I sometimes get requests such as: “I’m going to ignore a piece of your modeling results. Re-run your model so that I’m still right.” For example, a model might tell you that people above age 60 are at increased risk of an auto accident, but a manager objects that this is their target market. They may ask you to re-run the model without including age as a predictive variable, hoping that the new model will offset their (possibly unwise) business decision. (This is a completely made up example that is vaguely similar to something that might actually happen.) There may be some rare instances where this is appropriate, but managers need to understand when they are throwing away predictive power with their business decisions. If you throw a predictive variable out of your model, you are throwing away predictive power and you can’t get it back. Worse, the model will try to find the influence of that variable elsewhere, so you may end up with a model that is altogether weaker. To take an example, suppose I have a multivariable regression including driver age and other variables on auto accident frequency; if I throw driver age out of my model, the other variables that correlate with driver age will adjust to try to pick up the lost signal. It’s probably better to keep driver age in the model and consider it a “control variable” that won’t count against the driver (because you won't charge the customer based on their age or something). At any rate, it’s delusional to think that you can throw away predictive information and somehow totally offset this decision.

5)      Hobby horsing (again). As in, “Hey, I did X once and it was important and really made a big difference. Did you do X?” This overlaps with 3), obviously. Just because your brilliant insight saved the day once doesn’t mean it’s going to matter every time. It’s extremely annoying when people try to shoehorn their awesome idea into every single project.

6)      Finding a bullshit reason to discredit something. Often someone will object to an analysis the conclusion of which they dislike. A much more toxic dynamic is when someone dislikes a specific *person* and looks for stupid reasons to discredit their work. This is incredibly demoralizing to the data people and managers need to be very aware of when they are doing it. To guard against this, a data modeler should anticipate such objections and be ready to answer them, even going so far as to prepare for a specific person known to be an incredulous hard-ass.

7)      Someone gets mad at you for finding something inconvenient. I understand when someone responds incredulously to incredible results. Sometimes the data modeler really did goof. But sometimes the incredulity persists after all the objections are answered. “Yes, I adjusted for this. Yes, I controlled for that. Yes, I filtered for those.” After all this, perhaps you *still* conclude that your target market doesn’t deserve that huge discount you’ve been giving them, or your giant marketing initiative didn’t work. Accept the results and move on. I once had a boss who simply could not accept inconvenient results. He would come up with bullshit adjustments or filters or something hoping to get the results he wanted. It felt like I was being punished for bringing him bad news. Don’t be that guy.

8)      The regulatory state. I work in the insurance business as a research actuary. I am often in charge of crafting responses to regulators, sometimes filling out standard filing forms. Every single state (except Wyoming) requires that you file your rate plan with the state department of insurance (DOI), and every state DOI reserves the right to object to any filing. There is often a painful back-and-forth where regulators ask annoying questions and the insurance company tries to answer them. Sometimes the objections are based on the violation of a specific statute, and sometimes the reasons for objecting are far more capricious. They often have no statutory authority for their objections, or authority that comes from a lame catch-all (such as a vague law saying that rates must be “actuarially sound”, “not unfairly discriminatory”, etc.). This is a major pain for data modelers. I’m not making a libertarian stand here; if the state outlaws race-based discrimination and asks for reasonable proof that a model is not engaging in any such discrimination, I don’t object to that. My major gripe is that many of these departments are decades behind the latest modeling methods. Their standardized form questions often betray their ignorance.* A standard filing form from one state (see footnote below for additional detail) was littered with questions that looked like they were copied from a standard textbook on traditional linear models (very old school) but which are irrelevant to generalized linear models (glms, very commonly used in my industry). It’s like the actuaries at that department went to a predictive modeling seminar for a day or two and came back thinking they were experts on the topic. Then they copied some wording from the session handouts and turned it into an official state document. The non-standard questions we get from state DOIs are no better. Almost every filing is met with an “objection letter,” in which a DOI employee asks questions specific to the filing. One such question was (and I swear to you I am paraphrasing only slightly here): “What is a multivariate model?” Anyone remotely knowledgeable would have known that the term "multivariate" was a reference to glms, which almost every insurance company uses. These non-practitioners (I nearly said non-experts, but that would be a woeful understatement) have pathetically little knowledge of what they are actually regulating. I suspect this is the same in other industries. The latest, most cutting-edge methods must be justified, but must comply with decades-old language written for a different purpose. Such cutting edge tech must be explained to laymen who have the final decision. Once again I’m not here to critique government regulation in general. I just think it’s not too much to ask that government employees understand the thing they are regulating. And if a government agency cannot afford to keep such expertise on retainer, they need to relax or repeal those regulations.  This single factor is a huge barrier to innovation. If we can’t use a model unless it can be explained to an ignorant non-practitioner, that severely restricts what we can do. Another problem is that regulators sometimes demand a very specific statistical test when in practice something might be more of a judgment call. Data modeling is a process that requires a great deal of human judgment. What variables should I include? How should I group things? What kind of curve should I fit? Regulators often demand an unreasonable sort of rigor in the model-building process where all of this judgment is stripped away and replaced with an unalterable decision tree. Such a process will often lead to models with nonsensical results that any reasonable person can spot and fix, but the regulators interpret any insertion of human judgment as an attempt to be devious.

This list is by no means exhaustive. Obviously it could be longer, or shorter, but these were the peeves that occurred to me on one evening. 

* A questionnaire on generalized linear models (glms) in one state ask about tests for “homoscedasticity”, which means that the variance does not vary with the expected mean. (On a residuals plot, your residuals will spread out more on one side of the graph, rather than having a roughly constant standard deviation across the range.) This is a topic in traditional linear models, but the power of a glm is that you can relax the “constant variance” assumption. (You can make your error term gamma, poisson, tweedie, etc, rather than the traditional Gaussian that results in a constant expected variance.) Apologies if these technical details are confusing to the uninitiated, but it is really basic stuff as far as glms go. And every company is using these now.