[Phys-L] Re: What is Scientific Process?

[jbellina at SAINTMARYS.EDU (jbellina)]

On Aug 31, 2005, at 12:14 PM, Hugh Haskell wrote:

> At 07:51  -0500 8/31/05, jbellina wrote:
>
> > Au contraire,
> > It applies as well to experimental science.  The experiment you
> > design, or the observations you make are all done within the context
> > of your belief system, so there are lots of ideas and
> > expectations...I tend to avoid the h word.
>
> The "h word"? Hypothesis?  Yes indeed
>
>
> > When M and M did the speed of light work...actually they were trying
> > to measure the velocity of the earth relative to the ether, the
> > expection was that they would get a result.  When they did not, they
> > looked to experimental error, and worked to improve their
> > experiment.  Twenty five years later, after Einstein et al. the
> > person who was the then president of the APS repeated the experiment,
> > and got a positive result.  But the dogma had changed, so he looked
> > for experimental error.  Had his result been obtained 25 years
> > before, he would probably have not done so.
> Isn't the issue with this particular discussion one of objectivity?
> Experimenters can easily fall into the trap of seeing what they
> expect to see, whether it is there or not. I think Bridgman was
> arguing that experimenters not bring their prejudices into the
> laboratory, not that you cannot have some idea of what results to
> expect.

I don't think that Bridgman's requirement can be met...as I wrote
before, see Pattern's of Discovery on Norwood Hanson for more on
this.  In that sense, objectivity does not exist...any double blind
test has built in assumptions which make it not objective in the
sense you suggest.

> In other words, be prepared to be surprised, and don't be
> surprised if you are ( ;-)). It should go without saying that one
> should look for experimental error even if you get the results you
> expected, and one should not stop taking measurements just because
> the results are what you expected (this rule has to be applied with
> some discretion, however, otherwise no experiment would ever end).

And there is indeed a book entitled How Experiments End.  In the real
situation, how do you decide whether a discrepancy between what you
expect and what you get is because of your ideas or because of your
experiment.  Its easy for those of use who work in the world of stuff
everyone agrees on, less easy when you are on the edge.  But we have
to remember, even though we see ourselves on safe ground, for the
students it is new and perhaps not consistent with the ideas they
bring to class, so for them it is the edge.  For them it is more
difficult to determine if it is the experiment or their ideas.

> And one should take special precautions to make sure that one hasn't
> designed their experiment to guarantee the desired results,
> regardless of what the actual situation may be. Langmuir had some
> interesting things to say about this many years ago.

Langmuir was rightfully concerned about strong conclusions drawn from
small effects.

Read Bob Park for more about this sort of thing.
> I think the desire to eliminate experimenter bias is why
> "double-blind" is the gold standard of experimental science.

but not unbiased.
>
cheers,

joe

Joseph J. Bellina, Jr. Ph.D.
Professor of Physics
Saint Mary's College
Notre Dame, IN 46556