Standardization of non-aggregated data: theory and practice
| Autor: | Eva Grill, Rolf Weitkunat, Richela Fischer, Klaus Schotten, Alexander Crispin, Nicole Meyer |
|---|---|
| Rok vydání: | 2001 |
| Předmět: |
Male
Models Statistical Multivariate analysis Databases Factual Standardization Computer science Aggregate (data warehouse) Computational Biology Health Informatics Scale (descriptive set theory) Confidence interval Computer Science Applications Weighting Data set Data Interpretation Statistical Statistics Humans Female Statistical dispersion Epidemiologic Methods Software |
| Zdroj: | Computer Methods and Programs in Biomedicine. 65:207-227 |
| ISSN: | 0169-2607 |
| DOI: | 10.1016/s0169-2607(00)00132-2 |
| Popis: | Direct standardization of rates is a classical method in epidemiology by which the relationship between confounders and variables of analysis is balanced between the samples or populations to be compared. While the method is appropriate if single variables of aggregated data sets are to be compared, it is limited with respect to handling many variables of different measurement scale, especially in primary data situations. Therefore the method of direct standardization is adapted to the analysis of non-aggregated data. Computation of not only rates but other statistics of central tendency and dispersion as well as confidence intervals hence becomes feasible. In addition it is possible to re-adjust the primary data weights in order to perform stratified analyses. A procedure to do so, based solely on the available original weights without the need for analyzing the standard population is proposed along with an extensive SAS macro. The utility allows for the rapid and standardized analysis and tabulation of large multi-variable data sets. Both the statistical and technical properties of the macro are discussed and its usage is explained as well as exemplified. |
| Databáze: | OpenAIRE |
| Externí odkaz: |
|
Full Text from ScienceDirect