DIFFERENT FACETS OF SAMPLING PART 3
Vinod Anand | 16 Oct 2009
DIFFERENT FACETS OF SAMPLING
Let us now look at the limitations of the quantitative economics. These limitations are linked with the quantification of economic variables.As we have said earlier, collection ofdata is basic to economic statistics, and before this is done economic and social variables have to be numerically quantified. There are a large number of variables that can be quantified directly without the use of any substitute or alternative variables, also termed as proxy variables, but there are occasions when, due to either practical difficulties in the collection of data in respect of certain variables, or due to the impossibility of direct quantification of many other variables, we have to rely on indirect quantification through the use of proxy variables. For example, variables like, income, output, and expenditure can be directly quantified, but in certain situations when we believe that the respondents will not correctly indicate their income, output, and expenditure due to say, tax problems, we have to collect the required information on these variables through indirect means by asking indirect questions using the directly measurable proxy variables (like, hours of work and leisure, number of employees, amount of raw material bought, means of entertainment in the household) that would approximately reflect the values of their income, output, and expenditure. On the other hand, there are many other variables, which cannot be quantified directly, and as such, we have to rely only on their proxy variables. For example,
· Standard of living can best be assessed through the use of proxy variables like, income, consumption expenditure, housing and furnishing cost, and many such directly measurable variables;
· Concealed poverty can only be assessed through the proxy variables like, the number of poor involved in debt-trap, and in criminal and unethical activities;
· Extent and degree of corruption in a given system, which is beyond any direct measurement, can be assessed only through the use of
(i) proxy instruments based on written documents (like, press reports, opinion polls, court proceedings and judgments, judicial records, records from anti-corruption agencies), and even television talk shows and inside stories;
(ii) certain indices like the Corruption Perception Index (CPI), as used and published by Transparency International in 1995, and later updated in 1996 and 1997, and even beyond that, and the Business International Index (BII) as used by Business International, a subsidiary of the Economist’s Intelligence Unit, and the Global Competitive Report Index (GCRI) as based on a 1996 survey of firm managers who were queried on the extent of corruption relating to various aspects of business.
There is no dearth of such examples in economic and social studies. Proxy variables can either be close or remote/distant. Larger is the number of proxy variables, and more distant/remote they are, less genuine become the results of the given quantitative assessment.
Fixing the Sample Size is another most important issue. Let us see how?
In any quantitative study, based on sampling, correct sample size is a must to have a specified degree of precision. Theoretically speaking, the sample size depends on the
(a) costliness of errors in the estimate, and
(b) costliness of sampling.
No one would refute the fact that if substantial errors in the estimate will result in large penalties, the optimal sample size will tend to be large because the cost of increased sample size is likely to be outweighed by the resulting reduction in the sampling errors contained in the estimate and if it is relatively inexpensive to increase the sample size, the optimal sample size will tend to be larger than if it is relatively expensive to do so. There has to be, therefore, an optimal trade-off between these two determinants of the sample size.