Please use this identifier to cite or link to this item: http://repositorio.unifesp.br/handle/11600/28161
Title: An application of bootstrap resampling method to obtain confidence interval for percentile fatness cutoff points in childhood and adolescence overweight diagnoses
Authors: Colugnati, Fernando Antonio Basile [UNIFESP]
Louzada-Neto, F.
Taddei, Jose Augusto de Aguiar Carrazedo [UNIFESP]
Universidade Federal de São Paulo (UNIFESP)
Keywords: body composition
bootstrap
empirical distribution
confidence intervals
statistical methods
Issue Date: 1-Mar-2005
Publisher: Nature Publishing Group
Citation: International Journal of Obesity. London: Nature Publishing Group, v. 29, n. 3, p. 340-347, 2005.
Abstract: OBJECTIVE: To present a resampling approach to obtain confidence intervals (CIs) and the empirical distributions for the studentized regression residuals percentiles when used as cutoff points for overweight and obesity diagnosis in children and adolescents.METHOD: A tutorial for the nonparametric bootstrap with bias accelerating correction is presented. A classical method, the Binomial interpretation, is used as comparing criterion.SUBJECTS: A case study comprising 418 randomly selected subjects from a private secondary school (age: 10-17 y, boys: 52%).MEASUREMENTS: Body fat percentage (by), age (y) and Tanner criteria.RESULTS: the empirical distributions presented skewness suggesting that the CIs should not be symmetric. CIs obtained by the proposed approach were more realistic than the classical ones.CONCLUSIONS: We propose a simple and efficient way to obtain the interval estimates and the distribution properties of cutoff points for overweight and obese classification using a sample-based method that allows the comparison of cutoffs among many subpopulations.
URI: http://repositorio.unifesp.br/handle/11600/28161
ISSN: 0307-0565
Other Identifiers: http://dx.doi.org/10.1038/sj.ijo.0802866
Appears in Collections:Em verificação - Geral

Files in This Item:
There are no files associated with this item.


Items in DSpace are protected by copyright, with all rights reserved, unless otherwise indicated.