Testing goodness of fit of parametric models for censored data

Nysen, Ruth; Aerts, Marc; Faes, Christel

doi:10.1002/sim.5368

Cited by 13 publications

(8 citation statements)

References 30 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…The first approach used values generated by the laboratory instrument, and observations that were reported as zero were replaced by one-half the next smallest value (other than zero) for that contaminant. In the second approach, censoring methods were used by applying survival analysis techniques to left-censored data that have been demonstrated by other authors ( Helsel 2012 ; Nysen et al 2012 ) to improve estimation and reduce bias. To account for nondetects, the geometric mean (GM) from a lognormal random variable with censoring was calculated using the maximum likelihood method (MLE) and compared with the empirical median from the Kaplan-Meier approach.…”

Section: Methodsmentioning

confidence: 99%

Exposure to Free and Conjugated Forms of Bisphenol A and Triclosan among Pregnant Women in the MIREC Cohort

Arbuckle

Marro

Davis

et al. 2015

Environ Health Perspect

136

View full text Add to dashboard Cite

Background:Bisphenol A (BPA) and triclosan (TCS) are two nonpersistent chemicals that have been frequently measured in spot urine samples from the general population but less so in pregnant women; however, data are limited on the free (bioactive) and conjugated forms of these phenols.Objectives:The Maternal-Infant Research on Environmental Chemicals (MIREC) Study addressed these data gaps by utilizing stored maternal urine samples from a large multicenter cohort study of Canadian pregnant women.Methods:Concentrations of free and conjugated forms of BPA and TCS were measured in about 1,890 first-trimester urine samples by ultra performance liquid chromatograpy–tandem mass spectrometry using isotope dilution.Results:The glucuronides of BPA and TCS were the predominant forms of these chemicals measured (detected in 95% and 99% of samples, respectively), whereas the free forms were detected in 43% and 80% of samples, respectively. The geometric mean urinary concentrations for glucuronides of BPA and TCS were 0.80 μg/L (95% CI: 0.75, 0.85) and 12.30 μg/L (95% CI: 11.08, 13.65), respectively. Significant predictors of BPA included maternal age < 25 vs. ≥ 35 years, current smoking, low vs. high household income, and low vs. high education. For TCS, urinary concentrations were significantly higher in women ≥ 25 years of age, never vs. current smokers, and women with high household income and high education.Conclusions:The results from this study represent the largest national-level data on urinary concentrations of free and conjugated forms of BPA and TCS in pregnant women and suggest that maternal characteristics predicting elevated urinary concentrations of these phenols largely act in opposite directions.Citation:Arbuckle TE, Marro L, Davis K, Fisher M, Ayotte P, Bélanger P, Dumas P, LeBlanc A, Bérubé R, Gaudreau É, Provencher G, Faustman EM, Vigoren E, Ettinger AS, Dellarco M, MacPherson S, Fraser WD. 2015. Exposure to free and conjugated forms of bisphenol A and triclosan among pregnant women in the MIREC cohort. Environ Health Perspect 123:277–284; http://dx.doi.org/10.1289/ehp.1408187

show abstract

Section: Methodsmentioning

confidence: 99%

Exposure to Free and Conjugated Forms of Bisphenol A and Triclosan among Pregnant Women in the MIREC Cohort

Arbuckle

Marro

Davis

et al. 2015

Environ Health Perspect

136

View full text Add to dashboard Cite

show abstract

“…According to the chi-square and the K-S goodness-of-fit tests, the test statistical values between the theoretical data derived from the probability density function or the cumulative distribution curve and the sample data can be calculated. The distribution type that has a test statistical value smaller than the critical value (significance level is equal to 0.05) is selected (Haan, 2002;Nysen et al, 2012). More than one distribution type can be accepted by the chi-square and K-S goodnessof-fit tests.…”

Section: Data Acquisitionmentioning

confidence: 99%

Size effect of RQD and generalized representative volume elements: A case study on an underground excavation in Baihetan dam, Southwest China

Zhou

Chen

Cao

et al. 2013

Tunnelling and Underground Space Technology

View full text Add to dashboard Cite

“…For more details on the SemiNP series, see Zhang and Davidian (); Nysen et al. (, ). We are interested in the SemiNP representation as a framework to model distributions of data such as the Cadmium data, where censoring is typically located in the left tail of the distribution.…”

Section: Application To Data and Extensionsmentioning

confidence: 99%

“…It is assumed that the true unknown density lies in a broad class whose elements may be approximated by the SemiNP density estimator, tailored to provide an excellent approximation to virtually any plausible density. For more details on the SemiNP series, see Zhang and Davidian (2008); Nysen et al (2012Nysen et al ( , 2014. We are interested in the SemiNP representation as a framework to model distributions of data such as the Cadmium data, where censoring is typically located in the left tail of the distribution.…”

Section: The Family M Smentioning

confidence: 99%

“…For a given dataset,r denotes that value of r that optimizes Akaike's information criterion, resulting in the final SemiNP model fˆr. In the same vein, Nysen et al (2012) constructed a formal goodness of fit test based on a modified AIC criterion applied on the semiNP representation. Claeskens and Carroll (2007) used AIC for model selection in general semiNP problems.…”

Section: The Family M Smentioning

confidence: 99%

See 1 more Smart Citation

Model averaging quantiles from data censored by a limit of detection

et al. 2015

Self Cite

View full text Add to dashboard Cite

In chemical risk assessment, it is important to determine the quantiles of the distribution of concentration data. The selection of an appropriate distribution and the estimation of particular quantiles of interest are largely hindered by the omnipresence of observations below the limit of detection, leading to left-censored data. The log-normal distribution is a common choice, but this distribution is not the only possibility and alternatives should be considered as well. Here, we focus on several distributions that are related to the log-normal distribution or that are seminonparametric extensions of the log-normal distribution. Whereas previous work focused on the estimation of the cumulative distribution function, our interest here goes to the estimation of quantiles, particularly in the left tail of the distribution where most of the left-censored data are located. Two different model averaged quantile estimators are defined and compared for different families of candidate models. The models and methods of selection and averaging are further investigated through simulations and illustrated on data of cadmium concentration in food products. The approach is extended to include covariates and to deal with uncertainty about the values of the limit of detection. These extensions are illustrated with (134) cesium measurements from Fukushima Prefecture, Japan. We can conclude that averaged models do achieve good performance characteristics in case no useful prior knowledge about the true distribution is available; that there is no structural difference in the performance of the direct and indirect method; and that, not surprisingly, only the true or closely approximating model can deal with extremely high percentages of censoring.

show abstract

Testing goodness of fit of parametric models for censored data

Cited by 13 publications

References 30 publications

Exposure to Free and Conjugated Forms of Bisphenol A and Triclosan among Pregnant Women in the MIREC Cohort

Exposure to Free and Conjugated Forms of Bisphenol A and Triclosan among Pregnant Women in the MIREC Cohort

Size effect of RQD and generalized representative volume elements: A case study on an underground excavation in Baihetan dam, Southwest China

Model averaging quantiles from data censored by a limit of detection

Contact Info

Product

Resources

About