Skip to content
 

Create your own community (if you need to)

Back in 1991 I went to a conference of Bayesians and I was disappointed that the vast majority seem to not be interested in checking their statistical models. The attitude seemed to be, first, that model checking was not possible in a Bayesian context, and, second, that model checking was illegitimate because models were subjective. No wonder Bayesianism was analogized to a religion.

This all frustrated me, as I’d found model checking to be highly relevant in my Bayesian research in two different research problems, one involving inference for emission tomography (which had various challenges arising from spatial models and positivity constraints), the other involving models for district-level election results.

The good news is that, in the years since our book Bayesian Data Analysis came out, a Bayesian community has developed that is more accepting of checking models by looking at their fit to data. Many challenges remain.

The point of this story is that sometimes you can work with an existing community, sometimes you have to create your own community, and sometimes it’s a mix. In this case, my colleagues and I did not try to create a community on our own; we very clearly piggybacked off the existing Bayesian community, which indeed included lots of people who were interested in checking model fit, once it became clear that this was a theoretically valid step.

P.S. For more on the theoretical status of model checking in Bayesian inference, see this 2003 paper, A Bayesian formulation of exploratory data analysis and goodness-of-fit testing and this 2018 paper, Visualization in Bayesian workflow.

P.P.S. Zad’s cat, pictured above, is doing just fine. He doesn’t need to create his own community.

17 Comments

  1. Good advice. We are doing that here in DC, with clinicians, statisticians, patients, and, generally consumers of statistics & medicine. So many groups are funded by special interests. Independence is important.

    • Ben says:

      > funded by special interests. Independence is important.

      Do you say this as, “so many groups are funded by special interests and it is important to stay independent of these groups”

      or

      “so many groups are funded by special interests instead of all having to get money from generic sources (which makes independence possible)”?

      • Ben, Thanks for raising the two versions. I’ve been wrestling with both interpretations. We witness the operations of both versions in different patient advocacy groups. I exercise my independence on my own Twitter account. But am more limited by the membership to a non-profit which seeks funding. I believe that we have been in overdiagnosis & overtreated mode since the end of World War II. That so skews our health care delivery system. Well-meaning progressive voices can fall in line with the allopathic model which has been facing some serious theoretic and practice related challenges. But the professional incentives are yet not aligned to make the best use of the challenges. There is hope though given that some physicians are improving the health of their patients through nutritional & exercise regimes.

  2. jim says:

    “Coronavirus ‘does not spread easily’ by touching surfaces or objects, CDC says. ” Hey! OK! Now even the CDC is on board with aerosol transmission of the infection!

    So here’s a great idea for a modelling community:

    Model the path and concentration of virus in a plume of air from an infected individual and work on ways that businesses – for example food processors – can modify air flow and rearrange workers and/or furniture and fixtures to reduce infection risk.

    Actually it would be beneficial to have teams for different building configurations:

    – small shops like convenience stores, barber shops, hair salons, hobby shops etc (typically single area small square footage low ceiling)

    – larger factories like food processors or other manufacturers (larger areas, higher ceilings, possibly more crowded conditions – but often with a lot to gain by investing to keep production running)

    – LTCFs or other residential facilities

    I’m sure there are more…

    • Ben says:

      > Coronavirus ‘does not spread easily’ by touching surfaces or objects, CDC says

      Oh, I wasn’t aware of that update. Thanks!

      This was from yesterday apparently:

      “The primary and most important mode of transmission for COVID-19 is through close contact from person-to-person. Based on data from lab studies on COVID-19 and what we know about similar respiratory diseases, it may be possible that a person can get COVID-19 by touching a surface or object that has the virus on it and then touching their own mouth, nose, or possibly their eyes, but this isn’t thought to be the main way the virus spreads.”

      Has anything come out about the lifespan on surfaces out in the sun?

      Like crosswalk buttons and stuff. Would a virus die off on that pretty quick or no?

      • jim says:

        The surface mythology is very persistent. It seems strange that so many people worry about getting a respiratory infection from touching things rather than from breathing, but apparently they do.

        This article quotes the CDC as saying the virus “does not spread easily” from touching surfaces or objects. The CDC has apparently revised it’s wording since that article to say:

        “It may be possible that a person can get COVID-19 by touching a surface or object that has the virus on it and then touching their own mouth, nose, or possibly their eyes. This is not thought to be the main way the virus spreads, but we are still learning more about how this virus spreads.”

        But if you talk to experts who have experience with other respiratory viruses, you’ll find they’re quite confident that these viruses spread mostly through the air. Here’s the transcriptof Michael Osterholm’s inverview with Joe Rogan. Osterholm says:

        “an incubation period, we’re talking about from the time you and I got exposed, meaning I was in a room breathing the air that somebody else who was infected with the virus was expelling out, I breathed it in. How long from that time period till the time period that you get sick and what is that? “

        Note he doesn’t mention touching someone’s keyboard or using a pin pad at a grocery store. “Breathing” is the normal way of obtaining a respiratory virus.

        Here are a couple of reports from the CDC that describe “mass infection” vents.

        In this incident, the CDC confirmed a single person sickened 52 people in a choir practice even though people took care not to touch one another.

        In this incident a single case in call center likely infected 97 additional people.

        The CDC notes in the call center report:

        “Despite considerable interaction between workers on different floors of building X in the elevators and lobby, spread of COVID-19 was limited almost exclusively to the 11th floor, which indicates that the duration of interaction (or contact) was likely the main facilitator for further spreading of SARS-CoV-2.”

        Like LTCFs and other big outbreak centers, both of these incidents have or have in part:

        1) crowded conditions
        2) people frequently talking and vocalizing
        3) extended period of close proximity (hours or days)
        4) in the choir people specifically reported not touching one another; touching occurs in offices but isn’t a frequent practice
        5) in the office probably low ceilings; in the choir the side room where many people practiced probably had low ceilings.

        • I think people are aware of the respiratory droplet thing. the thing that is problematic is that early on the CDC and WHO and soforth repeatedly said they didn’t think it was “airborne”, which is apparently jargon for spread *mostly* by very small droplets (~ less than 1micron?) that tend to fall to the floor slowly over periods of tens of minutes to hours.

          Well, it turns out that the evidence is now pointing towards that airborne droplets being an important component in some circumstances, even if it’s not the primary component.

          However, if you’re someone who’s sheltering in place at home, you’re not exposed to respiratory droplets, so the main concern becomes “can I get this from my delivered groceries/mail/take-out-food/etc?” and of course the answer *is yes* because the world isn’t binary, you CAN get it that way, but how much risk is there? It seems clear that it’s not a huge risk, particularly if you’re smart about it, like you quarantine shelf stable groceries for 2-3 days, or you reheat your take-out in the oven/microwave, and throw away the packaging it came in.

          But, surfaces don’t seem to be the biggest risk, that’s absolutely correct. You should be avoiding small indoor locations where people are talking and touching things a lot… like *restaurants* and *hair salons* and *nail salons* and *telemarketing call centers* and *offices* and *churches*.

          So, basically the kinds of things we’re hearing about everyone opening up.

          • Martha (Smith) says:

            And what about having service personnel come to your house to service or repair your air conditioner, plumbing, water heater, etc.?

            • Yep, that’s another one. Though most of those things have lifetimes of ~ 10-20 years so probably only 5-10% of people need such things in any given year?

              • Adede says:

                “The good news is that, in the years since our book Bayesian Data Analysis came out, a Bayesian community has developed that is more accepting of checking models by looking at their fit to data. “

                In other words, a rudimentary application of the scientific method? No wonder Bayesianism was compared to a religion indeed!

              • Martha (Smith) says:

                Around here, it is pretty standard to get a yearly AC inspection and a yearly heating inspection.

          • Ben says:

            > The surface mythology is very persistent. It seems strange that so many people worry about getting a respiratory infection from touching things rather than from breathing, but apparently they do.

            > I think people are aware of the respiratory droplet thing. the thing that is problematic is that early on the CDC and WHO and soforth repeatedly said they didn’t think it was “airborne”

            Don’t give me too much credit! I guess my internal model for sickness has more bad-spirits covariates than it should.

            Thanks for the case studies Jim. Those are neat.

      • I saw one White House Coronavirus Briefing that featured how long the virus lives on surfaces. Sunlight kills the virus within 2 minutes; ethyl alcohol in 15 seconds; other disinfectant sprays within 5 to 10 minutes roughly.

  3. Anonymous says:

    I’ve thought about this a lot over the past decade. To move the needle in statistics is to start over.

  4. jonathan says:

    i appreciate the links to your framing work. For some reason, the paragraph about the unshrunk estimator captures too much variability stands out.

Leave a Reply