On statistical inference in sample surveys and the underlying role of randomization
- 27 Downloads
Assuming only the existence of a universe and a frame that identifies its members (ultimate sampling units), and a minimum number of necessary but nonrestrictive assumptions, the writer derives a basic proposition which shows that for sample surveys any form of inference about any universe characteristic must depend on the sampling distribution of estimates generated by randomization, and, by direct implication, the sampling design. Unbiasedness is validated by this proposition, but likelihood appears to be in conflict with it, and by implication with randomization. The relation between high or low variance and correspondingly low or high probability for an estimator is also investigated in the paper. Finally it is argued that restrictions on randomization may be injurious to normality.
KeywordsSampling Design Statistical Inference Sample Survey Finite Population Distinct Sample
Unable to display preview. Download preview PDF.
- Fisher, R. A. (1935).The Design of Experiments, Oliver and Boyd, Edinburgh.Google Scholar
- Bowley, A. L. (1926). Measurement of the precision attained in sampling,Bull. Int. Statist. Inst.,22, 6–62.Google Scholar
- Gini, C. and Galvani L. (1929). Di una applicazione del metoda rappresentativo all' ultimo censimento italiano della populazione,Annali di Statistica,4, 1–107.Google Scholar
- Kempthorne, O. (1973). Inference from experiments and randomization,A Survey of Statistical Design and Linear Models (ed. J. N. Srivastava), North Holland, Amersterdam, 303–331.Google Scholar
- Godambe, V. P. (1969). A fiducial argument with applications in survey sampling,J. R. Statist. Soc., B,31, 240–260.Google Scholar
- United Nations (1948).Recommendations for the Preparation of Sample Survey Reports, Series C, No. 1, United Nations, New York.Google Scholar
- Basu, D. (1971). An essay on the logical foundations of survey sampling, part one,Foundations of Statistical Inference (ed. V. P. Godambe and D. A. Sprott), Holt, Rinehart and Winston, Toronto, 203–242.Google Scholar
- Goodman, R. and Kish, L. (1950). Controlled selection—a technique in probability sampling,J. Amer. Statist. Ass.,45, 350–372.Google Scholar
- Smith, H. F. (1955). Variance components, finite populations, and experimental inference,N. C. Inst. Statistics Mimeo Series, 135.Google Scholar
- Koop, J. C. (1957). Contributions to the general theory of sampling finite populations without replacement and with unequal probabilities, Ph.D. Thesis, N. C. State Univ. (Also inN. C. Inst. Statistics Mimeo Series, 296, 1961.)Google Scholar