RESEARCH METHODOLOGY

We have sketched the notions of induction and deduction. But before scientific activity can begin, there is a need to specify the unit of any particular branch of science. We can in general specify “family” as the unit in social anthropology. But, in practice, there are many different types of family such as the nuclear, the joint, and the extended. Therefore, in any survey, we can specify family as a unit by defining it as father, mother, and their unmarried children. Such a unit would be different from another family where a grandparent lived with the family. In the physical and the natural sciences, the question of unit takes precedence in research since it is possible to measure anything by several units, and often it is possible to transform any quantity measured in a particular unit into another quantity measured in a different unit. Wealth can be measured in terms of rupees or the pound. Temperature can be measured in degrees Celsius or the Fahrenheit. We have said earlier that will and passion in the Baconian view affected understanding. But understanding is not central to modern science and research. Consider the views of the statistician William F. Ogburn3 who says that the tests of knowledge are reliability and accuracy, not understanding. A man might very well understand the wrong explanation that the sun moves around the earth. But that is not knowledge. A man staying in a tribal area for a long time sometimes is considered an expert in their affairs. But professional anthropologists until recently had a disdain for such knowledge. It is still a matter of controversy in social anthropology whether to believe such a person or a trained anthropologist who has spent not more than some years’ time among the tribes. The point of Ogburn is that impressionistic knowledge is not to be counted as knowledge since it would not be of “accurate, systematic, transferable kind called science.”4 He gives the example of Scandinavians whom an outsider might consider as tall, blue-eyed bonds with long heads and long faces despite the fact that such description fits no more than fifteen percent of the Scandinavian population. The process of measurement involves two types of error neither of which has anything to do with the popular meaning of error. Systematic error is caused by a defect in the measuring instrument such that every measurement done with the instrument involves the same error. Every time a faulty scale is used to measure length, the same faulty reading persists. This error is minimized with refinement of the measuring instrument. Random error is the difference between the true value and the observed value. The difference may either be positive or negative such that as a whole they cancel out when sufficient numbers of measurements are taken. We have said that systematic error persists. The magnitude of the systematic error is accuracy. It is either low or high. In the measurement of weekly income, an accuracy of one per thousand is superfluous. If we have a number of measurements, and the arithmetic mean is calculated, it will be largely free from the random error. Let another series of measurement on the object be taken and a second arithmetic mean calculated. The second average will be close to the first. In other words, the average is reproducible. Precision is a measure of this reproducibility. In other words, the mean is a reproducible measure.5 “The mean of three independent measurements of a quantity may be taken to be three times as reliable as a single measurement and given a weight 3, compared with weight 1 for a single reading.”6 Notice that present day scientists are willing to work with measures even inaccurate ones, provided that the measure is precise or reproducible. In the measurement of the gross domestic product we may be far from accurate from the true value, but the fact that it is reproducible, and three methods are available for its computation, makes it comparable. The gross domestic products calculated by a particular method become comparable over time owing to this reproducibility. Gross domestic products for a number of years can be deflated and expressed in real terms such that we can know whether the income level is really increasing or not. The deflated GDP figures are the replicates of each other (the role of money and other confounding variables being discounted), and we can know whether they are really different from each other. Scientific method does not end with observation and measurement. Imagination is also a part of scientific activity. The cannons of induction so carefully formulated by John Stuart Mill are no longer considered ways of discovery although in the method of difference we have a precursor of the modern experimental method. An hypothesis has become the central part of science, and the method so obtained is known as the hypothetico-deductive or simply the deductive method. S. F. Barker quotes another author D. C. Williams in his book as follows: The mode of argument is to rest on the principle that “if and only if all the consequents of a hypothesis are true, the hypothesis is true, while if all its tested consequents have been proved true, then probably all its consequents are true and so accordingly is the hypothesis.”7 Barker writes that hypotheses about atoms, electrons, valence bonds, genes, libidos, and so on will play a vital role as science advances.8 This is against the verdict of the scientists who long ago believed the electron to exist or claimed that it did not matter whether the gene was real or fictitious as long as it was there in the chromosome.