The brms package provides an interface to fit Bayesian generalized (non-)linear multivariate multilevel models using Stan, which is a C++ package for performing full Bayesian inference (see http://mc-stan.org/). The formula syntax is very similar to that of the package lme4 to provide a familiar and simple interface for performing regression analyses. A wide range of response distributions are supported, allowing users to fit—among others—linear, robust linear, count data, survival, response times, ordinal, zero-inflated, and even self-defined mixture models all in a multilevel context. Further modeling options include non-linear and smooth terms, auto-correlation structures, censored data, missing value imputation, and quite a few more. In addition, all parameters of the response distribution can be predicted in order to perform distributional regression. Multivariate models (ie., models with multiple response variables) can be fit, as well. Prior specifications are flexible and explicitly encourage users to apply distributions that actually reflect their beliefs. Model fit can easily be assessed and compared with posterior predictive checks, cross-validation, and Bayes factors.
2000-cook.pdf: “How Complex Systems Fail: Being a Short Treatise on the Nature of Failure; How Failure is Evaluated; How Failure is Attributed to Proximate Cause; and the Resulting New Understanding of Patient Safety”, (2000; ):
- Complex systems are intrinsically hazardous systems.
- Complex systems are heavily and successfully defended against failure.
- Catastrophe requires multiple failures—single point failures are not enough.
- Complex systems contain changing mixtures of failures latent within them
- Complex systems run in degraded mode.
- Catastrophe is always just around the corner.
- Post-accident attribution accident to a ‘root cause’ is fundamentally wrong.
- Hindsight biases post-accident assessments of human performance.
- Human operators have dual roles: as producers & as defenders against failure.
- All practitioner actions are gambles.
- Actions at the sharp end resolve all ambiguity.
- Human practitioners are the adaptable element of complex systems.
- Human expertise in complex systems is constantly changing.
- Change introduces new forms of failure.
- Views of ‘cause’ limit the effectiveness of defenses against future events.
- Safety is a characteristic of systems and not of their components.
- People continuously create safety.
- Failure free operations require experience with failure.
“You and Your Research”, (1986-03-07):
[Transcript of a talk by mathematician and Bell Labs manager Richard Hamming about what he had learned about computers and how to do effective research (republished in expanded form as Art of Doing Science and Engineering: Learning to Learn; 1995 video). It is one of the most famous and most-quoted such discussions ever.]
At a seminar in the Bell Communications Research Colloquia Series, Dr. Richard W. Hamming, a Professor at the Naval Postgraduate School in Monterey, California and a retired Bell Labs scientist, gave a very interesting and stimulating talk, ‘You and Your Research’ to an overflow audience of some 200 Bellcore staff members and visitors at the Morris Research and Engineering Center on March 7, 1986. This talk centered on Hamming’s observations and research on the question “Why do so few scientists make substantial contributions and so many are forgotten in the long run?” From his more than 40 years of experience, 30 of which were at Bell Laboratories, he has made a number of direct observations, asked very pointed questions of scientists about what, how, and why they did things, studied the lives of great scientists and great contributions, and has done introspection and studied theories of creativity. The talk is about what he has learned in terms of the properties of the individual scientists, their abilities, traits, working habits, attitudes, and philosophy.
“Chess Masters' Hypothesis Testing”, (2004):
Falsification may demarcate science from non-science as the rational way to test the truth of hypotheses. But experimental evidence from studies of reasoning shows that people often find falsification difficult. We suggest that domain expertise may facilitate falsification. We consider new experimental data about chess experts’ hypothesis testing. The results show that chess masters were readily able to falsify their plans. They generated move sequences that falsified their plans more readily than novice players, who tended to confirm their plans. The finding that experts in a domain are more likely to falsify their hypotheses has important implications for the debate about human rationality.
“Posterior Sampling for Large Scale Reinforcement Learning”, (2017-11-21):
We propose a practical non-episodic PSRL algorithm that unlike recent state-of-the-art PSRL algorithms uses a deterministic, model-independent episode switching schedule. Our algorithm termed deterministic schedule PSRL (DS-PSRL) is efficient in terms of time, sample, and space complexity. We prove a Bayesian regret bound under mild assumptions. Our result is more generally applicable to multiple parameters and continuous state action problems. We compare our algorithm with state-of-the-art PSRL algorithms on standard discrete and continuous problems from the literature. Finally, we show how the assumptions of our algorithm satisfy a sensible parametrization for a large class of problems in sequential recommendations.
Individuals influence each others’ decisions about cultural products such as songs, books, and movies; but to what extent can the perception of success become a “self-fulfilling prophecy”? We have explored this question experimentally by artificially inverting the true popularity of songs in an online “music market,” in which 12,207 participants listened to and downloaded songs by unknown bands. We found that most songs experienced self-fulfilling prophecies, in which perceived-but initially false-popularity became real over time. We also found, however, that the inversion was not self-fulfilling for the market as a whole, in part because the very best songs recovered their popularity in the long run. Moreover, the distortion of market information reduced the correlation between appeal and popularity, and led to fewer overall downloads. These results, although partial and speculative, suggest a new approach to the study of cultural markets, and indicate the potential of web-based experiments to explore the social psychological origin of other macro-sociological phenomena.
The tendency to make unhealthy choices is hypothesized to be related to an individual’s temporal discount rate, the theoretical rate at which they devalue delayed rewards. Furthermore, a particular form of temporal discounting, hyperbolic discounting, has been proposed to explain why unhealthy behavior can occur despite healthy intentions. We examine these two hypotheses in turn. We first systematically review studies which investigate whether discount rates can predict unhealthy behavior. These studies reveal that high discount rates for money (and in some instances food or drug rewards) are associated with several unhealthy behaviors and markers of health status, establishing discounting as a promising predictive measure. We secondly examine whether intention-incongruent unhealthy actions are consistent with . We conclude that intention-incongruent actions are often triggered by environmental cues or changes in motivational state, whose effects are not parameterized by . We propose a framework for understanding these state-based effects in terms of the interplay of two distinct reinforcement learning mechanisms: a “model-based” (or goal-directed) system and a “model-free” (or habitual) system. Under this framework, while discounting of delayed health may contribute to the initiation of unhealthy behavior, with repetition, many unhealthy behaviors become habitual; if health goals then change, habitual behavior can still arise in response to environmental cues. We propose that the burgeoning development of computational models of these processes will permit further identification of health decision-making phenotypes.
“The unreasonable effectiveness of one-on-ones”, (2019-12-28):
When I started dating my partner, I quickly noticed that grad school was making her very sad. This was shortly after I’d started leading an engineering team at Wave, and so the “obvious” hypothesis to me was that the management (okay, “management”) one gets in graduate school is totally ineffective.
…One-on-ones are a management tradition at lots of tech companies, perhaps popularized by High Output Management,1 in which a manager regularly schedules time with a direct report to discuss whatever the report wants. At Wave, I’ve had one-on-ones with my manager since the time I joined, and I found them incredibly useful for helping me improve at work…mine often included:
- Personal habits and self-improvements…
- Project management…
…Obviously, the things Eve and I talked about weren’t exactly the same as my Wave one-on-ones, though they did share some common themes. Here are some of the things we talked about that Eve thinks made the biggest difference:
- Figuring out when she should be outlining new parts of her dissertation vs. fleshing out existing parts
- Realizing that she was spending a lot of time reading crappy papers that she didn’t have to
- Noticing when and why she was least productive (for instance, noticing when her procrastination was a coping strategy to avoid executing a plan that she didn’t really believe would succeed)
- Asking for more frequent feedback from her adviser and dissertation committee
- Being able to talk through anything stressful
- Allowing herself space to “stare into the abyss” and confront uncomfortable possibilities (eg. is it actually worth finishing her PhD?)
In general, Eve summarized our one-on-ones as being a forcing function for her to fully decide on longer-term goals and then focus her work on the best way to achieve those goals, rather than getting too bogged down in whatever was right in front of her.
…The last thing this helped me realize is that specialists have a lot of non-specialized problems. In one sense, this is so well known it’s become a cliché—the engineer who just wants to crank out code all day, the philosophy professor with their head in the clouds. But the cliché doesn’t really describe me or most engineers or philosophers I know, who are broad-minded enough to be happy thinking about things outside our assigned specialty. Even for us, though, we can often increase our impact a lot by improving our generalized effectiveness.
Objectives: To determine the overall rate of loss of workplace teaspoons and whether attrition and displacement are correlated with the relative value of the teaspoons or type of tearoom.
Design: Longitudinal cohort study.
Setting: Research institute employing about 140 people.
Subjects: 70 discreetly numbered teaspoons placed in tearooms around the institute and observed weekly over five months.
Main Outcome Measures: Incidence of teaspoon loss per 100 teaspoon years and teaspoon half life.
Results: 56 (80%) of the 70 teaspoons disappeared during the study. The half life of the teaspoons was 81 days. The half life of teaspoons in communal tearooms (42 days) was significantly shorter than for those in rooms associated with particular research groups (77 days). The rate of loss was not influenced by the teaspoons’ value. The incidence of teaspoon loss over the period of observation was 360.62 per 100 teaspoon years. At this rate, an estimated 250 teaspoons would need to be purchased annually to maintain a practical institute-wide population of 70 teaspoons.
Conclusions: The loss of workplace teaspoons was rapid, showing that their availability, and hence office culture in general, is constantly threatened.
2015-bronnenberg.pdf: “Do Pharmacists Buy Bayer? Informed Shoppers and the Brand Premium”, (2015-07-15; ):
We estimate the effect of information and expertise on consumers’ willingness to pay for national brands in physically homogeneous product categories. In a detailed case study of headache remedies, we find that more informed or expert consumers are less likely to pay extra to buy national brands, with pharmacists choosing them over store brands only 9% of the time, compared to 26% of the time for the average consumer. In a similar case study of pantry staples such as salt and sugar, we show that chefs devote 12 percentage points less of their purchases to national brands than demographically similar non-chefs. We extend our analysis to cover 50 retail health categories and 241 food and drink categories. The results suggest that misinformation and related consumer mistakes explain a sizable share of the brand premium for health products, and a much smaller share for most food and drink products. We tie our estimates together using a stylized model of demand and pricing.