Skip to content

On deck for the first half of 2019

OK, this is what we’ve got for you:

  • “The Book of Why” by Pearl and Mackenzie
  • Reproducibility and Stan
  • MRP (multilevel regression and poststratification; Mister P): Clearing up misunderstandings about

  • Becker on Bohm on the important role of stories in science
  • This is one offer I can refuse
  • How post-hoc power calculation is like a shit sandwich

  • Storytelling: What’s it good for?
  • Freud expert also a Korea expert
  • Data partitioning as an essential element in evaluation of predictive properties of a statistical method

  • A thought on the hot hand in basketball and the relevance of defense
  • A ladder of responses to criticism, from the most responsible to the most destructive
  • “Either the results are completely wrong, or Nasa has confirmed a major breakthrough in space propulsion.”

  • Moneyball for evaluating community colleges
  • Of butterflies and piranhas
  • Science as an intellectual “safe space”?

  • Just when you thought it was safe to go back into the water . . . SHARK ATTACKS in the Journal of Politics
  • One more reason to remove letters of recommendation when evaluating candidates for jobs or scholarships.
  • When doing regression (or matching, or weighting, or whatever), don’t say “control for,” say “adjust for”

  • If this article portrays things accurately, the nutrition literature is in even worse shape than I thought
  • What should JPSP have done with Bem’s ESP paper, back in 2010? Click to find the surprisingly simple answer!
  • The bullshit asymmetry principle

  • “Objective: Generate evidence for the comparative effectiveness for each pairwise comparison of depression treatments for a set of outcomes of interest.”
  • Autodiff! (for the C++ jockeys in the audience)
  • Principal Stratification on a Latent Variable (fitting a multilevel model using Stan)

  • Of multiple comparisons and multilevel models
  • If you want to measure differences between groups, measure differences between groups.
  • New estimates of the effects of public preschool

  • Facial feedback is back
  • The Stan Core Roadmap
  • Fitting big multilevel regressions in Stan?

  • Fitting multilevel models when the number of groups is small
  • Our hypotheses are not just falsifiable; they’re actually false.
  • “Using 26,000 diary entries to show ovulatory changes in sexual desire and behavior”

  • Votes vs. $
  • Should he go to grad school in statistics or computer science?
  • Michael Crichton on science and storytelling

  • Simulation-based statistical testing in journalism
  • More on that horrible statistical significance grid
  • P-hacking in study of “p-hacking”?

  • “Do you have any recommendations for useful priors when datasets are small?”
  • Deterministic thinking meets the fallacy of the one-sided bet
  • Are GWAS studies of IQ/educational attainment problematic?

  • I believe this study because it is consistent with my existing beliefs.
  • Healthier kids: Using Stan to get more information out of pediatric respiratory data
  • “News Release from the JAMA Network”

  • Kevin Lewis has a surefire idea for a project for the high school Science Talent Search
  • HMC step size: How does it scale with dimension?
  • Does diet soda stop cancer? Two Yale Cancer Center docs have diametrically opposite views!

  • Evidence distortion in clinical trials
  • Separated at birth?
  • “Light Privilege? Skin Tone Stratification in Health among African Americans”

  • George Orwell meets statistical significance: “Politics and the English Language” applied to science
  • Good news! Researchers respond to a correction by acknowledging it and not trying to dodge its implications
  • “Yes, not only am I suspicious of the claims in that op-ed, I’m also suspicious of all the individual claims from the links in these two sentences”

  • Journalist seeking scoops is as bad as scientist doing unreplicable research
  • Yes on design analysis, No on “power,” No on sample size calculations
  • (back to basics:) How is statistics relevant to scientific discovery?

  • A corpus in a single survey!
  • The neurostatistical precursors of noise-magnifying statistical procedures in infancy
  • Not Dentists named Dennis, but Physicists named Li studying Li

  • Remember that paper we wrote, The mythical swing voter? About shifts in the polls being explainable by differential nonresponse? Mark Palko beat us to this idea, by 4 years.
  • Political polarization and gender gap
  • Junk science + Legal system = Disaster

  • Yes, you can include prior information on quantities of interest, not just on parameters in your model
  • From the Stan forums: “I’m just very thirsty to learn and this thread has become a fountain of knowledge”
  • “No, cardiac arrests are not more common on Monday mornings, study finds”

  • One more reason I hate letters of recommendation
  • Statistical-significance thinking is not just a bad way to publish, it’s also a bad way to think
  • Estimating treatment effects on rates of rare events using precursor data: Going further with hierarchical models.

  • Are male doctors better for male heart attack patients and female doctors better for female heart attack patients?
  • When and how do politically extreme candidates get punished at the polls?
  • He asks me a question, and I reply with a bunch of links

  • New golf putting data! And a new golf putting model!
  • Balancing rigor with exploration
  • Yes, I really really really like fake-data simulation, and I can’t stop talking about it.

  • Should we talk less about bad social science research and more about bad medical research?
  • Mister P for surveys in epidemiology — using Stan!
  • Here’s a puzzle: Why did the U.S. doctor tell me to drink more wine and the French doctor tell me to drink less?

  • Surgeon promotes fraudulent research that kills people; his employer, a leading hospital, defends him and attacks whistleblowers. Business as usual.
  • A world of Wansinks in medical research: “So I guess what I’m trying to get at is I wonder how common it is for clinicians to rely on med students to do their data analysis for them, and how often this work then gets published”
  • An interview with Tina Fernandes Botts

  • How to approach a social science research problem when you have data and a couple different ways you could proceed?
  • “Boston Globe Columnist Suspended During Investigation Of Marathon Bombing Stories That Don’t Add Up”
  • Here’s an idea for not getting tripped up with default priors . . .

  • Impact of published research on behavior and avoidable fatalities
  • How did our advice about research ethics work out, three years later?
  • What’s a good default prior for regression coefficients? A default Edlin factor of 1/2?

  • “The Long-Run Effects of America’s First Paid Maternity Leave Policy”: I need that trail of breadcrumbs.
  • Question on multilevel modeling reminds me that we need a good modeling workflow (building up your model by including varying intercepts, slopes, etc.) and a good computing workflow
  • “Heckman curve” update: The data don’t seem to support the claim that human capital investments are most effective when targeted at younger ages.

  • Why “bigger sample size” is not usually where it’s at.
  • Emile Bravo and agency
  • “How Sloppy Science Creates Worthless Cures, Crushes Hope, and Wastes Billions” . . . and still stays around even after it’s been retracted

  • Prestigious journal publishes sexy selfie study
  • What sort of identification do you get from panel data if effects are long-term? Air pollution and cognition example.
  • Treatment interactions can be hard to estimate from data.

  • Works of art that are about themselves
  • All statistical conclusions require assumptions.
  • State-space models in Stan

  • The network of models and Bayesian workflow, related to generative grammar for statistical models
  • No, its not correct to say that you can be 95% sure that the true value will be in the confidence interval
  • Differential effects of research trauma on fatigue and functioning of journal editors in chronic sloppy research syndrome

  • Claims about excess road deaths on “4/20” don’t add up
  • Here’s a supercool controversy for ya
  • Wanted: Statistical success stories

  • R-squared for multilevel models
  • “Appendix: Why we are publishing this here instead of as a letter to the editor in the journal”
  • Ballot order update

  • “Incentives to Learn”: How to interpret this estimate of a varying treatment effect?
  • Conditioning on post-treatment variables when you expect self-selection
  • “How many years do we lose to the air we breathe?” Or not.

  • How to think scientifically about scientists’ proposals for fixing science
  • Continuing discussion of status threat and presidential elections, with discussion of challenge of causal inference from survey data
  • “Boosting intelligence analysts’ judgment accuracy: What works, what fails?”

  • “One should always beat a dead horse because the horse is never really dead”
  • Olivia Goldhill and Jesse Singal report on the Implicit Association Test
  • Automation and judgment, from the rational animal to the irrational machine

  • Do regression structures affect research capital? The case of pronoun drop
  • Difference-in-difference estimators are a special case of lagged regression
  • The Arkansas paradox

  • Gremlin time: “distant future, faraway lands, and remote probabilities”
  • That illusion where you think the other side is united and your side is diverse
  • Maintenance cost is quadratic in the number of features

  • Poetry corner
  • Bayesian analysis of data collected sequentially: it’s easy, just include as predictors in the model any variables that go into the stopping rule.
  • BizStat: Modeling performance indicators for deals

  • Scandal! Mister P appears in British tabloid.
  • “We see MRP as a way to combine all the data—pre-election voter file data, early voting, precinct results, county results, polling—into a single framework”
  • “In 1997 Latanya Sweeney dramatically demonstrated that supposedly anonymized data was not anonymous,” but “Over 20 journals turned down her paper . . . and nobody wanted to fund privacy research that might reach uncomfortable conclusions.”

  • On the term “self-appointed” . . .
  • Hey, people are doing the multiverse!
  • Vigorous data-handling tied to publication in top journals among public heath researchers

  • Software for multilevel conjoint analysis in marketing
  • Neural nets vs. statistical models
  • I’m no expert

  • John Le Carre is good at integrating thought and action
  • Donald J. Trump and Robert E. Lee
  • Pushing the guy in front of the trolley

  • Crystallography Corner: The result is difficult to reproduce, but the result is still valid.
  • They’re working for the clampdown
  • What pieces do chess grandmasters move, and when?

  • Let’s publish everything.
  • Why edit a journal?
  • Should we mind if authorship is falsified?

  • Solutions to the 15 questions on our applied regression exam

So, yeah, the usual range of topics.

P.S. I listed the posts in groups of 3 just for easier readability. There’s no connection between the three posts in each batch.


  1. Ecoute Sauvage says:

    An antidote to this tide of insanity:

    I know empirically (from flight controls on airplanes) that analog displays register faster than digital ones – pilot reaction time is of obvious importance in flying.

    So I wonder if you think that psychology / nutrition / climate modeling went off the deep end by trying to digitize data instead of dealing with it the way our brains work, analog. It’s more efficient to simultaneously hold contradictory possibilities as a way to deal with uncertainty.

    • David L. says:

      That article starts out badly.

      “Once it was simple: programmers wrote the instructions that were supplied to the machines. Since the machines were controlled by these instructions, those who wrote the instructions controlled the machines. Two things then happened. As computers proliferated, the humans providing instructions could no longer keep up with the insatiable appetite of the machines. Codes became self-replicating, and machines began supplying instructions to other machines.”

      This makes no sense. Whatsoever. (And it goes downhill from there.) One might infer from this that this is an expert on kayaking writing. You’d be correct.

  2. Martha (Smith) says:

    “When doing regression (or matching, or weighting, or whatever), don’t say “control for,” say “adjust for””

    I go one step further and say “attempt to adjust for”.

  3. Mikhail says:

    Still waiting for the discussion on Mayo’s book

Leave a Reply to David L.