Public service announcement: patch your iPhone *immediately.*

Are the mutations seen in #SARSCoV2 viruses in Danish mink likely to cause problems for people? I asked Marion Koopmans, publications: https://pure.eur.nl, Dr Emma Hodcroft, Carl T. Bergstrom & Prof Francois Balloux. Their answers were reassuring. statnews.com/2020/11/05/spr…

Not A Number strikes again! My favourite line I've seen in an article for some time: 'A module whose job was to ensure numbers stay within expected bounds said “not a number, not my problem!”'

Curious what people think of this.

15,000 people in the Moderna control group got a placebo. 90 got COVID. If we do the same math he did for reinfection, then placebo is 99.4% effective.

If you look at autocorrelated data, pick any random day, and look for a break in linear trend on that day, you will find a statistically significant break a ludicrous amount of the time. In this random AR(1) data with delta = .8, I reject the null at 95% about half the time.

If you haven't seen Arpad Elo's original derivation of his rating rule, here it is. Surprisingly, I couldn't find it anywhere on the internet, but it's in his book.

What’s the best way to catch up on new features in Stan from the past year or two?

Click the link for the amazing headline, stay for the thorough comic about barriers to good science in academia.

A lot of new content from Netflix and Disney+ has an “audio description” feature alongside closed captions. They often say these things out loud. Highly recommend trying them if any of these would be useful. They can also be great unintentional comedy.

The three major segments of the general public, according to the MA vaccination plan: * higher education workers * bottled beverage workers * veterinarians mass.gov/info-details/c…

Adding - big reason I am confident in resilience of vaccine efficacy in the face of new variants is super high neutralizing antibodies after two doses. They are *very* low after 1. This is courting disaster.

Too many broadly useful stats methods are masked in domain-specific language. In my new pair of posts, I discuss formula-free #causalinference design patterns to help data analysts recognize frameworks as they encounter them in everyday work emilyriederer.netlify.app/post/causal-de… 1/3

I did not know this.

I feel like this framing implies that 1) people’e only value is to their employer and 2) any time off the clock is automatically leisure time.

Proposed new term: ill-supervised model. Definition: a model that was trained to optimize for something other than a desired outcome. Example: “This ill-supervised model minimizes distances in an embedding space that no one cares about“

This is very cool!

worth keeping in mind.

TIL: there is a bizarro pipe (->.;) in #rstats win-vector.com/2016/12/13/mag…