ABSTRACT
Suppose y, x1, . . . , xk are real variables with values Yi, X j i, j = 1, . . . , k; i = 1, . . . , N , assumed on the units of U = (1, . . . , i, . . . , N ), labeled i = 1, . . . , N . If the survey data d = (s, Yi, X j i|is), provided by a design p, are employed in inference about certain known functions of Yi, X j i, for i = 1, . . . , k; i = 1, . . . , N then we have what is called a descriptive study. For example, we may intend to estimate the totals Y = ∑Ni Yi, X j =∑N
1 X j i, j = 1, . . . , k or corresponding means or ratios along with their variance or mean square error estimators and set up confidence intervals concerning these estimand parameters. Or we may be interested to examine the values of correlation coefficients between pairs of variables or multiple correlation coefficients of one variable on a set of variables, or may like to estimate the regression coefficient of y on x1, . . . , xk, and so on. Then the parameters involved are also defined on the values Yi, X j i for i = 1, . . . , N , and our analysis is descriptive.
