Statistical Inference about Quantile Class Means with Simple and Stratified Random Sampling
The topic discussed in this paper is closely related to Mahalanobis's fractile (graphical) analysis. Suppose that a bivariate population is partitioned into subpopulations of given relative sizes by partitioning one of the two marginal distributions into consecutive parts of given relative size...
| Veröffentlicht in: | Sankhyā: The Indian Journal of Statistics, Series B (1960-2002). - INDIAN STATISTICAL INSTITUTE, 1960. - 47(1985), 2, Seite 272-279 |
|---|---|
| 1. Verfasser: | |
| Format: | Online-Aufsatz |
| Sprache: | English |
| Veröffentlicht: |
1985
|
| Zugriff auf das übergeordnete Werk: | Sankhyā: The Indian Journal of Statistics, Series B (1960-2002) |
| Schlagworte: | Large sampling theory of inference Quantile classes Simple random sampling Stratified sampling Mathematics Behavioral sciences |
| Zusammenfassung: | The topic discussed in this paper is closely related to Mahalanobis's fractile (graphical) analysis. Suppose that a bivariate population is partitioned into subpopulations of given relative sizes by partitioning one of the two marginal distributions into consecutive parts of given relative sizes. The bivariate subpopulations defined in this way are called quantile classes. This paper deals with the problem of making statistical inference about such quantile classes when we have a simple random sample or a stratified random sample from the population. In particular, for each one of these two cases, we show: (i) that the asymptotic (large sample) simultaneous distribution of the natural estimators of the quantile class means is multivariate normal under weak assumptions; (ii) how the covariance matrix of this normal distribution can be estimated consistently. |
|---|---|
| ISSN: | 05815738 |