apathy

> The real predictor of success, almost no matter how you define success, is about the effort you put in.

That, and how well you choose your parents. Unless you honestly believe black people just didn't try hard enough to be president for the first 230 or so years of the USA's existence.

Christ these Horatio Alger "just so" stories are ridiculous. And if we repeal GINA, it'll make it that much harder for people to escape their supposed "destiny". Genetics interacting with environment makes us who we are. I'm of the opinion that the latter is more important, and I hold a doctorate in statistical genetics so this is not purely an off-the-cuff opinion. It may or may not be right in a specific case (evidence >> eminence, credentials are at best a proxy for expertise) but it happens to agree with yours, if you think about it.

The amount of grit you develop is going to be a function in large part of your childhood (or other major environmental influences; in the field, "environment" is shorthand for "everything that isn't genetics"). So locking in an ignorance of its importance is just astoundingly stupid. If we want a sclerotic European aristocracy or a caste system in America, this is the way to get it.

And of course the rich and powerful will do anything to perpetuate their position. On average, however, the American dynasty half-life is 3-4 generations. Compare that to Europe and I think we know which better favors social mobility.

This is severely fucked. Wellness programs do little or no good but do allow employers to offer incentives for participation. This on the other hand is GATTACA level fuckery. GINA took many years to be put in place; now the proposal put forward will potentially cause people to lose health coverage or their job based on things they cannot control. Potentially their kids, too, since these idiotic "wellness" programs are administered by poorly regulated third parties.

This is very, very bad stuff.

This is interesting -- suppose that the choke point of negotiation was handled by the government on behalf of citizens, but the administration of programs was privatized? That could be interesting.

Lord knows I've had about enough red tape for several lifetimes from NIH, NSF, and similar organs; Medicare as it now stands somehow manages to result in both medicine-at-a-loss and also fraud on a spectacular scale. I'm not a big fan of government but between consolidation and fragmentation, I don't think the current medical care solution is working, nor is it sustainable.

yeah the doc-in-a-box practice is rapidly becoming a thing of the past. However the corporatization of medicine is not doing anyone any favors save for the administrators, and even then only the executives make out like bandits.

Contracts and practice plans are considerably simplified ;-)

assuming you're asking in good faith, the answer is pretty simple. A larger population has more leverage, and if all of the citizens bargain as one bloc (i.e., if the government does so on their behalf), it drives the market. Note that a tremendous chunk of the market needs to be involved for this to have maximum effect -- if a physician can "opt out" of Medicare/MediCal and make more money, they will, by and large. (There's a reason that being on call instead of waiting for referrals is called "service")

Most physicians in (e.g.) the UK do participate in the single-payer market (there are a small number who make a living offering pay-as-you-go services, but they are the vanishing minority), since it dominates demand. In the US there are a great many physicians who simply won't accept Medicare rates (they're viewed as too low by most) and since there are alternative sources of patients, that's who they treat (typically privately insured). This leads to the cases that show up at County or the ER being a hell of a lot more expensive than necessary as they tend not to be survivable for long. (A running joke at most county hospitals is that conditions believed "incompatible with life" routinely walk or roll into the ER and clinics.)

If you have 1-3 insurance companies and MediCal/Medicaid and Medicare then you have different rates for different groups, almost all of it horrendously opaque, and the 3rd party insurers are not incentivized to pay for anything.

As far as private vs. public, the issue here is the same as for schools, a private insurer or school can choose not to insure or educate a "customer", the government by law cannot. In the handful of cross-over studies of charter schools or vouchers, after controlling for subject-specific effects, the children who switched from public to charter or private tended to do slightly worse than expected based on their test scores from public schools. (It is a difficult experiment to run for numerous reasons.)

Medical care, unlike most goods and services, is stunningly inelastic in demand -- you either need it and will do whatever is required to get it, or you don't and won't, by and large. (Elective surgeries for cosmetic purposes are a separate matter; nobody goes in for a stent "just because" or visits the trauma unit just to poke their head in) Furthermore, a substantial amount of the cost is centered on the first and last few years of a person's life. Unless you would like the "market driven solution" of even higher infant mortality and elderly culling to proceed, 3rd party insurers don't have the incentives to make it go.

wow, that sure looks like it could be automated, starting with the indexing passes (script them in Python maybe?) and slowly factoring out bits of it into sklearn modules or whatever. Alternatively, as a batch job spawned from within R, but that would get annoying over time. Still, the idea is great and it sounds like you guys end up with time to go kayaking etc.

I'd think that Microsoft would be bending over backwards to get you to run the R stuff on Azure, since they bought the Revolution operation to show off such things. In the long term I imagine that you'll move most of this to Python.

You might consider poaching some spatial statistics people from Uber. Like, say, some of the women. At least if the market is big and robust enough to bring in some heavy hitters. It seems like a lot of the steps might be automated with convolutions and validated against your expert pool, but I can't say for sure.

Neat projects. Best of luck.

It's mentioned at the end of Part III, in chapter 17, on survival models.

Answered my own question -- Part III of this book is what I am waiting for. I was going to send the authors a note, which seems silly, but my name is in an awful lot of standard textbooks just because I sent in corrections or notes, and for some reason I find that satisfying. Like I made some sort of a difference to some students somewhere.

Part II describes what you are asking for.

I wish abbreviations like inverse probability weighting and marginal structural models were expanded in the second book. It's annoying to have to look up "IP weighting" only to discover "oh, it's IPW, god damn you [authors] to hell".

MSMs are interesting. Now I remember why Robins' name stuck in my head. The book shows all the math explicitly, which is nice, and it delves into causal inference for time-to-event data, which is also nice.

Now I'm curious whether they look at piecewise constant survival models for time-varying coefficients. It's mentioned, but I didn't read enough to see if it's treated in detail. If it is, everyone who does A/B testing should read the book, because this is one of those "little details" from biostatistics that becomes super important at big retailers (like, say, Amazon, where the principal economist at the time pointed it out to me).

they seem to be fixated on how shiny and new L1 penalties are, in 2014. Greg Ridgway started using gradient boosting machines (GBM) for propensity scoring in the early 2000s, and I didn't see them cite him, so I kind of hate them already. On the other hand, at least GBM works well.

I'm no economist, though. Perhaps this is novel at NBER. It's just odd to see someone acting like using an ensemble to enable data-driven model selection is something new.

nb. I didn't read the entire 76-page paper (partly because it's obscenely verbose). A quick skim and here are my from-the-hip remarks. If they suck, I'll refund every cent you paid me ;-)

Quite possible -- Robins was pointed out to me as one of the pioneers by Sander Greenland, around 2008 or so? I'm more of an experimental/clinical statistician at this point. The work I did with propensity scoring and doubly robust estimation convinced me that shrinkage was a less-bad approach for the things I needed to do. YMMV.

The soul of marketing is getting people to overlook stuff like the laws of physics, the arrow of time, and compound interest. Any time you find yourself resorting to the word "should," it's a good idea to ask:

1) why would the universe naturally favor order? 2) why would the order it favors please my sensibilities?

Questioning the above premises will often yield opportunities that a rigid attention to preconceived notions would not.

Look up confounding by indication and some of the power/sensitivity studies for so-called doubly robust estimators. Counterfactuals are reasonable. A lot of "let's turn this observational study into a designed experiment with math" approaches turn out not to be. That's the gist of it; if you want rigor, you should read the papers. I don't have a reference to hand at the moment (on phone) but it shouldn't take more than a few minutes of searching google scholar to hit the appropriate vein. The bottom line is simply TANSTAAFL.

Also apparently the plague of undiagnosed Lyme Brain Cloud has spread to Antarctica.

Where will it end? Somebody quick dispatch an MA/RP to Bellingshausen Station!

Nah, Robins started (asterisk) all this causal/propensity-scored/pretend-experimental mania.

It's what drove me back to hard interventional experiments and eventually to adaptive clinical trial design.

(Asterisk): ok more like Pearl started it and Robins made it more practical with doubly robust designs etc. I still find most of the mechanics rather shady.

Prefer Mosaic. You'll see.

This. It would need to affect F2. It won't.

If they have an identifiable deficiency, sure. Sickle cell is a stem cell (erythroblast more precisely) disorder, so fixing the stem cells fixes the disease. If you already have the tissues you want to change, it'll be harder.

Yes, it's old by academic standards. And do read the comments. Remember that getting something through peer review is not unlike a siege, but once within the fortress walls, a mighty army guards glamorous findings against high-profile refutation.

But in the interest of "balance" here it is.

Oh -- you will :-)

With dplyr & purrr, if you can't find a DSL in R that works for you, the odds are you'll need lisp or Haskell to satisfy you.

Sacha Epskamp's tutorial has a nice section on this:

http://sachaepskamp.com/files/Cookbook.html

Also the section on penalized Ising models is great, I haven't seen anything else like it.

That's fantastic!

Nice job Thomas! I've gotten a hell of a lot of mileage out of ggraph already, your work has been tremendous. Eventually it will be nice to dump plots to JavaScript for poking purposes, but for publication there is nothing that comes close to ggraph. And I've tried everything.

Cre-Lox systems predate CRISPR/Cas9 or CRISPR/Cfp1 approaches by quite some time. If one exists for the system you want to study, they also tend to work better. (floxing the original mouse is, however, substantially harder)

I don't think it's entirely accurate to say that conditional editing of DNA is a new thing. The ready accessibility and combinatorial possibilities, yes, but for targeted conditional knockouts, floxing mice has been a thing for about 20 years now.

You don't need a ridiculously expensive rig to find that in physics, either: https://en.wikipedia.org/wiki/Double-slit_experiment

laser pointer plus a sharp enough carbide wheel

(trained as a chemist, grad school in biostats & genetics, now mostly design experiments & clinical trials... I have physics envy, except I don't envy their funding models!)

1) Tanya is an oncologist 2) Cachexia != fasting. Not much you can do about cachexia, it is an end stage phenomenon where the tumor is hijacking essentially all incoming nutrition. Feed the patient, feed the goober. 3) reversing myeloid bias (I'm dubious about this, even though Gregor was not the kind of fast and loose player that would fudge counts) would, if true, enhance innate and adaptive responses (subject to a million caveats because Immunology Is Really Fucking Hard)

If patients consent to trials and enough evidence accumulates it will become the standard of care. Of course it's going to be tough to patent not eating so don't hold your breath.

And refuse to serially update it ;-)

Which is why no working statistician is really 100% Bayesian (intractable in many cases) or 100% frequentist (obviously wrong in some cases). We all use Bayes' Rule (don't need to be "a Bayesian" to do that) and we all are forced to do Newman-Pearson-style power calculations now and then (holds nose). Even the latter have their uses, in preventing the worst of the worst abuses of frequentist techniques (it's not frequentism that is inherently bad per se, it's the profound abuses that turn it into a magnet for bad science; sometimes a Bayesian formulation of a problem is simply intractable).

This right here is why I read HN. Thanks!