Page:Sm all cc.pdf/115

 4) decide how to deal with all relevant variables.
 * List all known variables that might influence the result. Classify each variable as either: (a) controllable, (b) uncontrollable but with an approximately known value or known influence on the key dependent variable, or (c) uncontrollable and unknown in influence.
 * Decide which variables are of greatest interest. Try to minimize effects of all other variables (e.g., by keeping them constant or by randomization).
 * Select one of these tactics: (1) focus on only one variable and vary it systematically; (2) analyze several variables simultaneously through use of a factorial design; or (3) analyze several variables sequentially in a series of experiments.

A later section of this chapter, ‘Control of Variables’, discusses these options in more detail.

5) choose the equipment to be used, if any.
 * Consider the relative advantages of buying, borrowing, and building equipment. Preparations and lead time are greatest for building, less for buying, and least for borrowing. Ability to tailor the equipment to your needs is greatest for building, less for buying, and least for borrowing. Costs are high for both building and buying compared to borrowing. Borrowing is OK for a few experiments but usually impractical for a protracted suite of experiments. Experiments on borrowed equipment tend to be done in binges, with less opportunity for intermediate analyses during experiments and for follow-up experiments.
 * Before using equipment, learn its background theory, operations steps, operational considerations, and potential problems. Obviously, some compromise is needed between the ideal of comprehensive understanding and the reality of time constraints. Generating unreliable data and then troubleshooting can be much more time-consuming than learning how to operate the equipment properly and within its limitations. One need not become an electronics technician to use electronic equipment, but pitfalls abound for those who use equipment that they understand only minimally. For example, the dilettante may omit implied operations steps, use the equipment outside its design range, overlook variables that affect equipment results, and misinterpret results.

6) calibrate equipment, both before and during the experiment.
 * Test the equipment before starting the experimental series. Do not assume that it can be trusted simply because someone else recently used it successfully. Their samples may have been subtly different, or the equipment response may have changed.
 * Choose standards and a calibration procedure appropriate for the equipment, samples, and anticipated data range.
 * Recalibrate after the equipment is repaired, moved, or changed, and after any substantial hiatus.
 * Run calibration samples regularly, preferably in a randomized mixture with the experimental samples. If equipment response changes with time or warm-up, calibrating at the start or end of each day is insufficient.
 * Run blanks if feasible.