Statistical Significance of Healtcare Studies

A post mostly to park this image so I don't have to hunt for it again:

Oregon gave a subset of people access to healthcare by lottery, so you essentially got a randomized population there with and without access.  Massachusetts implemented Romneycare.  Both are obvious things for researchers to study.  Massachusetts study found an effect on outcomes with improved insurance, the Oregon one did not.

So different conclusions from two conflicting studies?  Not at all.  The visualization is a simple illustration show that the studies do not conflict--in fact, they happen to agree embarrassingly well.  However, the Oregon study was at lower power so could not be reported as finding an effect.

The Only TED Talk You'll Ever Need

Except maybe for the Reggie Watts one:

DIY Climate Models

According to this American Scientist piece by Brian Hayes, there are a wealth of climate models available as public code, with different levels of complexity.  Some are designed to be pedagogical, others are professional scale.  Compiling one seems like it would be fun and probably cheaper than setting up a new aquarium.

See also http://bit-player.org/extras/climate/

The simple simulator above, with greenhouse effect halved.  The cold temperatures
are due to not just that change but the increased albedo caused by the year round
snow cover.

Things to Do: Sodium Triacetate

Take a supersaturated solution of sodium triacetate.  Add fingers.  Watch crystals of the trihydrate form around your fingers.

The Jet Sex

The Jet Sex

Per LGM, an excellent account of flight attendants and labor & gender issues around their employment.  The quote that adds it to my "to read" list is:

The supervisor handbook for American Airlines stated, “The first fundamental is appearance. A stewardess must be attractive. We can sometimes pretend a person is attractive, if we admire them for some other reason. This should be avoided."