~~Statistical Methods in Analytical Chemistry
Statistical Methods in Analytical Chemistry - раздел Образование, Part One THE PREDICATE Statistical Methods, As Used In The Design And Interpretation Of Chcmical Exp...
Statistical methods, as used in the design and interpretation of chcmical experimentation, arc not a substitute for common sense or for what scientists refer to as scientific judgement; they rather constitute an objective aid to judgement. In view of the mcticulousncss exhibited by scientific workers in die purely technical aspects of dicir experiments it seems appropriate to devote some thought to two further and no less important aspccts of experimentation: its design and the final interpretation of its outcomc. Very often the experimenter designs the experiment as it proceeds, acting on a moment's intuition. Similarly, the interpretation or a rudimentary study of data, the latter consisting in many cases in ranking studied clTccts in accordance with their observed disturbances caused by experimental and systcmctic errors. Naturally, even the most experienced worker is subject to an occasional intuition. Therefore natural, and actually borne out by (he facts, is the assumption that many experiments could have resulted in more conspicuous and sharper constructed plan.
In taking cognizance of the unavoidable experimental errors, rather than in ignoring them or dismissing them as negligible, and in attempting a mathematical study of the avoidance or the correction of systematical errors, statisticians have succccdcd, to a ccrtain extent, in offering some objective criteria which are invaluable in these situations. One of the important by-products of these methods is an estimate of the quantity of experimental work that is necessary for obtaining sufficient factual proof for a scientific hypothesis, such an estimate being more objective than the guessing technique often used. But among the most important contributions of statistical methodology to scientific experimentation is the possibility of clearly separating the effects of various variables under study as well as the interactions of these variables with regard to the measured quantities, from the data resulting from a complex experiment. Hardly can it be denied that this requires careful planning. Even before the experiment is started, the various possible types of results must be hypothctically considered from the point of view of the questions under study.
