Re: Can we measure it?

[David Kerridge <dfk@mwfree.net>]

Thanks to all contributors.  I hope to reply to all the contributions in time.

 When I sent the original post, I thought that what I was saying was
 non-controversial, and perhaps rather dull. But please remember
 that I am in no sense trying to lay down the law about what should
 be measured, or what should be called measurable. I am simply
 drawing attention to a distinction that WED made,  in the hope that
 between us we can clarify exactly what WED meant.

Dan Swart says:
>
>I can think of "measures" that fit all 6 rules except number 5......

>   << 5    The number has predictive value: in other words, the
>measurement has
>   "temporal spread." >>

>...............................................................................
>..  Later, when
>the system is brought under control, the "measure" has predictive value and
>satisfies rule number 5.  Does this mean a number is just an indicator until
>it reaches statistical control, and then it becomes a measure?
>

That sounds like a useful additional rule 7. I did consider saying something
about variability beyond that implied by rule 5. But most measurements
used by physical scientists are not under statistical control, simply because
the concept is unknown to them. Instead they rely on a measurement
process having a known upper limit to its measurement error - but they often
deceive themselves into assuming greater accuracy than they have, as
Shewhart's data on measurements of the speed of light clearly show.
"Statistical method from the viewpoint of quality control" p 66.

Deming clearly agreed with Dan Swart, saying "A method of measurement
does not exist unless the results show statistical control" (OOC p 280)

I am relieved to find that I am not being stricter than WED.


Best wishes

David

dfk@mwfree.net
======================================================================