From: Donald Rubin Sent: Wednesday, December 7, 2016 11:58 PM To: jeffrey E. Subject: Re: Placebo effects, EB, etc. Sorry for the delay. Had meetings all =ay with a former PhD student and a current student. Not a silly question. =nbsp;l think that it has a lot to do with how data are collected and =ho summarizes them. With recorded data there is often a lot of =ounding, truncating, heaping (meaning on nice round numbers, as in age =eaping), and "filtering" meaning by the data collector =o avoid reporting suspicious values that could lead to scorn. In =ome contexts we also have scores that are averages of many values or =ffects, which push toward the gaussian by the CUT; for example, I =sed to work at ETS, and exams are notorious for those effects. =nbsp;Theoretical distributions like the Cauchy are almost always just =pproximations because machines or people cannot record data with an =nfinite number of places. I think your question is relatively =eep with a variety of answers depending on the context and could change =s we get better at recording data. :) On Dec 7, 2016, at 12:52 PM, jeffrey E. [email protected] <mailto:[email protected]> wrote: query, why are so many distributions in nature =o consistant and accurate. doesnt it give you pause. =nbsp;. like centrigal force. not a true force but =ne we sense. many distributions could be describeds as =seudo forces pushing the tails toward the mean.. . could =ravity be one of these pseudo type forces pushing things towards = distribution.?? not as silly a question as it first might =ppear to be . :) On Wed, Dec 7, 2016 at 1:47 PM, Donald Rubin > wrote: Think =0. Where will you be? On Dec =, 2016, at 12:25 PM, jeffrey E. <[email protected]> wrote: ar you around on the =eekend in cambridge.? On Wed, Dec 7, 2016 at 1:10 PM, =onald Rubin l wrote: OK. I'll get back to you after = make contact with Adriaan again, and my plans settle a bit. On Dec 7, 2016, at 10:38 AM, jeffrey E.