Chronology Current Month Current Thread Current Date
[Year List] [Month List (current year)] [Date Index] [Thread Index] [Thread Prev] [Thread Next] [Date Prev] [Date Next]

Re: [Phys-L] spinner data -- experimentation 101



Thoreau said
Sometimes circumstantial evidence is very strong,
as when you find a trout in the milk.

The residuals look fishy to me. The period exhibits gradual
smooth changes and then there's a jump. Then some more smooth
changes and another jump:
https://www.av8n.com/physics/img48/fidget-spinner-period-residuals.png

On 09/18/2017 11:19 PM, bernard cleyet wrote:

Note the diff Chisq’s.

There's a saying in the experimentation business:
There are some things you can see but not measure;
There are other things you can measure but not see.

In this case, rather than trying to quantify the chi square,
I suggest eyeballing the residuals. Chi square makes sense
if the residuals result from random noise ... which is not
the case here.

Fitting the data using some drag proportion to v^2
and some proportional to v gives a fit that looks
reasonable if you don't look too closely:
https://www.av8n.com/physics/img48/fidget-spinner-freq.png
https://www.av8n.com/physics/img48/fidget-spinner-period.png

However, you can't trust the fitted parameters if you
don't trust the underlying data.

Because of the jumps, I don't trust this fit farther than
I could throw it. And it's bad manners to throw a fit.

This shows why experimentation is iterative: Do a preliminary
experiment, find out what the problems are, then try again.