Sample allocation balancing overall representativeness and stratum precision.
Purpose: In large-scale surveys, it is often necessary to distribute a preset sample size among a number of strata. Researchers must make a decision between prioritizing overall representativeness or precision of stratum estimates. Hence, I evaluated different sample allocation strategies based on s...
| Publicado en: | Annals of Epidemiology Vol. 28; no. 8; pp. 570 - 576 |
|---|---|
| Autor principal: | |
| Formato: | research Journal Article |
| Publicado: |
Elsevier B.V.
Aug2018
|
| Acceso en línea: | Ver este registro en EBSCOhost |
| fields | @attributes: recordID: 1 pdfLink: plink: https://search.ebscohost.com/login.aspx?direct=true&db=ccm&AN=130837906&site=ehost-live header: @attributes: shortDbName: ccm uiTerm: 130837906 longDbName: CINAHL Complete uiTag: AN controlInfo: bkinfo: dissinfo: jinfo: jid: 10472797 J8R jtl: Annals of Epidemiology issn: 10472797 maglogo: N pubinfo: dt: Aug2018 vid: 28 iid: 8 pid: 467 pub: Elsevier B.V. place: New York, New York artinfo: ui: 130837906 130837906 NLM29784525 130837906 10.1016/j.annepidem.2018.04.011 NLM29784525 130837906 ppf: 570 ppct: 6 formats: tig: atl: Sample allocation balancing overall representativeness and stratum precision. aug: au: Diaz-Quijano, Fredi Alexander affil: Department of Epidemiology, Faculdade de Saúde Pública da Universidade de São Paulo, São Paulo, Brazil sug: subj: Selection Bias Sample Size Human Computer Simulation Validation Studies Comparative Studies Evaluation Research Multicenter Studies Clinical Assessment Tools ab: Purpose: In large-scale surveys, it is often necessary to distribute a preset sample size among a number of strata. Researchers must make a decision between prioritizing overall representativeness or precision of stratum estimates. Hence, I evaluated different sample allocation strategies based on stratum size.Methods: The strategies evaluated herein included allocation proportional to stratum population; equal sample for all strata; and proportional to the natural logarithm, cubic root, and square root of the stratum population. This study considered the fact that, from a preset sample size, the dispersion index of stratum sampling fractions is correlated with the population estimator error and the dispersion index of stratum-specific sampling errors would measure the inequality in precision distribution. Identification of a balanced and efficient strategy was based on comparing those both dispersion indices.Results: Balance and efficiency of the strategies changed depending on overall sample size. As the sample to be distributed increased, the most efficient allocation strategies were equal sample for each stratum; proportional to the logarithm, to the cubic root, to square root; and that proportional to the stratum population, respectively.Conclusions: Depending on sample size, each of the strategies evaluated could be considered in optimizing the sample to keep both overall representativeness and stratum-specific precision. pubtype: Academic Journal doctype: research Journal Article ougenre: Article language: English refInfo: holdings: @attributes: islocal: N |
|---|