Page:Sm all cc.pdf/107

 Pilot studies can indicate whether a potential experiment is likely to be fruitful and how one should deal with the relevant variables. Pilot studies cannot substitute for a well designed formal experiment.

Prototypes and pilot studies are modes of scientific troubleshooting. Whether or not we expect problems, these techniques help us to avoid them.

Troubleshooting and Search Procedures
Troubleshooting is a familiar, intimate part of science. The trouble may involve computer hardware or software, malfunctioning equipment, or an experiment that is giving results that are unexpected and possibly unreliable. These and many other problems are solvable with established troubleshooting and search procedures. Yet the techniques are published in few places, and most of us react to encountered problems by thinking of only one or two remedies. Wilson [1952] considers troubleshooting and search techniques in detail, and the completeness of the following discussion owes much to his comprehensive treatment.

The foremost rule of troubleshooting and search is: keep records to avoid duplication of effort and floundering, to reveal any patterns in the troubleshooting results, and to make it easier to identify potential tests that you have overlooked. Keeping records is unnecessary for the first or even second attempted solution. As soon as troubleshooting extends beyond a few minutes, however, one should start jotting down notes of what has been tried and what it yielded.

The frustration implicit in troubleshooting can result in needless damage. Hippocrates was familiar with the problem 2000 years ago. His guideline, which could have supplanted later leechcraft, is still apropos: Primum non nocere; first do no harm. When diagnosing a medical problem, exploratory surgery is an acceptable last resort; autopsy is not.

A subtler manifestation of primum non nocere is the following question: is the object of the quest worth the cost of the search? Cost can take various forms, tangible and intangible. When cost is computed in dollars, this question is the daily quandary faced by NSF and the occasional topic of intense arguments, as exemplified by the debate over star-wars research.

If troubleshooting new equipment:

 * 1) Remember the facetious saying, “If all else fails, read the manual.” Probably something is connected wrong, a setting is incorrect, or a step is being left out. The better manuals even have a section on troubleshooting.
 * 2) If possible, run a standard that you know is supposed to work on this equipment. If the standard works OK, then how does your sample differ from the standard? If the standard doesn’t work either, then go on to the next step.
 * 3) Examine all of the equipment for visible signs of damage.
 * 4) Try to isolate which part of the equipment is malfunctioning. Some of the search procedures discussed later may help. Sometimes it is possible to swap out parts of the equipment. Some parts can be tried in isolation or in conjunction with other working equipment, and some circuits can be tested with a multitester (AC, DC, and resistance).