John P. Buonaccorsi
Research Subjects
John P. Buonaccorsi has a total of 32 co-authors. Here are the top co-authors by number of shared publications:
- Magne Thoresen
University of Oslo
3 shared publications Giovanni Romeo
Department of Biostatistics, Oslo Centre for Biostatistics and Epidemiology, University of Oslo, Oslo, Norway
2 shared publicationsJohn Staudenmayer
Department of Mathematics and Statistics, University of Massachusetts, Amherst, Massachusetts 01003 USA
2 shared publications- Marit Bragelien Veierød
Oslo Centre for Biostatistics and Epidemiology, Department of Biostatistics, Institute of Basic Medical Sciences, University of Oslo, Oslo, Norway.
2 shared publications Petter Laake
Department of Biostatistics, University of Oslo, Oslo, Norway
2 shared publications- Agnieszka Prochenka
Institute of Computer Science, Polish Academy of Sciences, Warsaw, Poland
1 shared publication
Latest Publications
- Open Access
- 28 March 2023
- Published by Taylor & Francis in Communications in Statistics - Simulation and Computation
- Vol. 53 (11) , 5474-5490
Detecting and correcting for heteroscedasticity in the presence of measurement error
The problem of measurement error in predictors has received a vast amount of attention for both linear and non-linear models. The major focus has generally been on inference for regression coefficients, while residual analysis, variance estimation and heteroscedasticity assessment have received limited attention. In this paper we propose and evaluate new methods, based on reconstructed predictors, to address these problems in the context of linear regression. Methods using moment reconstructed predictors are found to perform well, including a new model-based bootstrapping technique for testing heteroscedasticity. Motivated by an example where the predictor is methylation rate and the measurement error model is based on the binomial distribution, particular attention is given to being able to handle the situation where both the measurement error variances and the variance of the error in the equation may be changing as a quadratic function of the unobserved predictor.
- 28 March 2023
- Published by Taylor & Francis in Communications in Statistics - Simulation and Computation
- Vol. 53 (11) , 5474-5490
Open AccessDetecting and correcting for heteroscedasticity in the presence of measurement error
The problem of measurement error in predictors has received a vast amount of attention for both linear and non-linear models. The major focus has generally been on inference for regression coefficients, while residual analysis, variance estimation and heteroscedasticity assessment have received limited attention. In this paper we propose and evaluate new methods, based on reconstructed predictors, to address these problems in the context of linear regression. Methods using moment reconstructed predictors are found to perform well, including a new model-based bootstrapping technique for testing heteroscedasticity. Motivated by an example where the predictor is methylation rate and the measurement error model is based on the binomial distribution, particular attention is given to being able to handle the situation where both the measurement error variances and the variance of the error in the equation may be changing as a quadratic function of the unobserved predictor.
- Open Access
- 1 July 2020
- Published by SAGE Publications in Multiple Sclerosis Journal – Experimental, Translational and Clinical
- Vol. 6 (3)
Sensorimotor function in progressive multiple sclerosis
Background: A sensitive test reflecting subtle sensorimotor changes throughout disease progression independent of mobility impairment is currently lacking in progressive multiple sclerosis. Objectives: We examined non-ambulatory measures of upper and lower extremity sensorimotor function that may reveal differences between relapsing–remitting and progressive forms of multiple sclerosis. Methods: Cutaneous sensitivity, proprioception, central motor function and mobility were assessed in 32 relapsing–remitting and 31 progressive multiple sclerosis patients and 30 non-multiple sclerosis controls. Results: Cutaneous sensation differed between relapsing–remitting and progressive multiple sclerosis at the foot and to a lesser extent the hand. Proprioception function in the upper but not the lower extremity differed between relapsing–remitting and progressive multiple sclerosis, but was different for both upper and lower extremities between multiple sclerosis patients and non-multiple sclerosis controls. Foot-tap but not hand-tap speed was slower in progressive compared to relapsing–remitting multiple sclerosis, suggestive of greater central motor function impairment in the lower extremity in progressive multiple sclerosis. In addition, the non-ambulatory sensorimotor measures were more sensitive in detecting differences between relapsing–remitting and progressive multiple sclerosis than mobility assessed with the 25-foot walk test. Conclusion: This study provides novel information about changes in sensorimotor function in progressive compared with relapsing–remitting forms of multiple sclerosis, and in particular the importance of assessing both upper and lower extremity function. Importantly, our findings showed loss of proprioceptive function in multiple sclerosis but also in progressive compared to relapsing–remitting multiple sclerosis.
- 1 July 2020
- Published by SAGE Publications in Multiple Sclerosis Journal – Experimental, Translational and Clinical
- Vol. 6 (3)
Open AccessSensorimotor function in progressive multiple sclerosis
Background: A sensitive test reflecting subtle sensorimotor changes throughout disease progression independent of mobility impairment is currently lacking in progressive multiple sclerosis. Objectives: We examined non-ambulatory measures of upper and lower extremity sensorimotor function that may reveal differences between relapsing–remitting and progressive forms of multiple sclerosis. Methods: Cutaneous sensitivity, proprioception, central motor function and mobility were assessed in 32 relapsing–remitting and 31 progressive multiple sclerosis patients and 30 non-multiple sclerosis controls. Results: Cutaneous sensation differed between relapsing–remitting and progressive multiple sclerosis at the foot and to a lesser extent the hand. Proprioception function in the upper but not the lower extremity differed between relapsing–remitting and progressive multiple sclerosis, but was different for both upper and lower extremities between multiple sclerosis patients and non-multiple sclerosis controls. Foot-tap but not hand-tap speed was slower in progressive compared to relapsing–remitting multiple sclerosis, suggestive of greater central motor function impairment in the lower extremity in progressive multiple sclerosis. In addition, the non-ambulatory sensorimotor measures were more sensitive in detecting differences between relapsing–remitting and progressive multiple sclerosis than mobility assessed with the 25-foot walk test. Conclusion: This study provides novel information about changes in sensorimotor function in progressive compared with relapsing–remitting forms of multiple sclerosis, and in particular the importance of assessing both upper and lower extremity function. Importantly, our findings showed loss of proprioceptive function in multiple sclerosis but also in progressive compared to relapsing–remitting multiple sclerosis.
- 30 May 2017
- Published by Oxford University Press (OUP) in Biometrics
- Vol. 74 (1) , 135-144
Model-based Bootstrapping When Correcting for Measurement Error with Application to Logistic Regression
Summary: When fitting regression models, measurement error in any of the predictors typically leads to biased coefficients and incorrect inferences. A plethora of methods have been proposed to correct for this. Obtaining standard errors and confidence intervals using the corrected estimators can be challenging and, in addition, there is concern about remaining bias in the corrected estimators. The bootstrap, which is one option to address these problems, has received limited attention in this context. It has usually been employed by simply resampling observations, which, while suitable in some situations, is not always formally justified. In addition, the simple bootstrap does not allow for estimating bias in non-linear models, including logistic regression. Model-based bootstrapping, which can potentially estimate bias in addition to being robust to the original sampling or whether the measurement error variance is constant or not, has received limited attention. However, it faces challenges that are not present in handling regression models with no measurement error. This article develops new methods for model-based bootstrapping when correcting for measurement error in logistic regression with replicate measures. The methodology is illustrated using two examples, and a series of simulations are carried out to assess and compare the simple and model-based bootstrap methods, as well as other standard methods. While not always perfect, the model-based approaches offer some distinct improvements over the other methods.
- 30 May 2017
- Published by Oxford University Press (OUP) in Biometrics
- Vol. 74 (1) , 135-144
Model-based Bootstrapping When Correcting for Measurement Error with Application to Logistic Regression
Summary: When fitting regression models, measurement error in any of the predictors typically leads to biased coefficients and incorrect inferences. A plethora of methods have been proposed to correct for this. Obtaining standard errors and confidence intervals using the corrected estimators can be challenging and, in addition, there is concern about remaining bias in the corrected estimators. The bootstrap, which is one option to address these problems, has received limited attention in this context. It has usually been employed by simply resampling observations, which, while suitable in some situations, is not always formally justified. In addition, the simple bootstrap does not allow for estimating bias in non-linear models, including logistic regression. Model-based bootstrapping, which can potentially estimate bias in addition to being robust to the original sampling or whether the measurement error variance is constant or not, has received limited attention. However, it faces challenges that are not present in handling regression models with no measurement error. This article develops new methods for model-based bootstrapping when correcting for measurement error in logistic regression with replicate measures. The methodology is illustrated using two examples, and a series of simulations are carried out to assess and compare the simple and model-based bootstrap methods, as well as other standard methods. While not always perfect, the model-based approaches offer some distinct improvements over the other methods.
- 6 June 2016
- Published by Wiley in Statistics in Medicine
- Vol. 35 (22) , 3987-4007
Correcting for binomial measurement error in predictors in regression with application to analysis of DNA methylation rates by bisulfite sequencing
Motivated by a genetic application, this paper addresses the problem of fitting regression models when the predictor is a proportion measured with error. While the problem of dealing with additive measurement error in fitting regression models has been extensively studied, the problem where the additive error is of a binomial nature has not been addressed. The measurement errors here are heteroscedastic for two reasons; dependence on the underlying true value and changing sampling effort over observations. While some of the previously developed methods for treating additive measurement error with heteroscedasticity can be used in this setting, other methods need modification. A new version of simulation extrapolation is developed, and we also explore a variation on the standard regression calibration method that uses a beta-binomial model based on the fact that the true value is a proportion. Although most of the methods introduced here can be used for fitting non-linear models, this paper will focus primarily on their use in fitting a linear model. While previous work has focused mainly on estimation of the coefficients, we will, with motivation from our example, also examine estimation of the variance around the regression line. In addressing these problems, we also discuss the appropriate manner in which to bootstrap for both inferences and bias assessment. The various methods are compared via simulation, and the results are illustrated using our motivating data, for which the goal is to relate the methylation rate of a blood sample to the age of the individual providing the sample. Copyright © 2016 John Wiley & Sons, Ltd.
- 6 June 2016
- Published by Wiley in Statistics in Medicine
- Vol. 35 (22) , 3987-4007
Correcting for binomial measurement error in predictors in regression with application to analysis of DNA methylation rates by bisulfite sequencing
Motivated by a genetic application, this paper addresses the problem of fitting regression models when the predictor is a proportion measured with error. While the problem of dealing with additive measurement error in fitting regression models has been extensively studied, the problem where the additive error is of a binomial nature has not been addressed. The measurement errors here are heteroscedastic for two reasons; dependence on the underlying true value and changing sampling effort over observations. While some of the previously developed methods for treating additive measurement error with heteroscedasticity can be used in this setting, other methods need modification. A new version of simulation extrapolation is developed, and we also explore a variation on the standard regression calibration method that uses a beta-binomial model based on the fact that the true value is a proportion. Although most of the methods introduced here can be used for fitting non-linear models, this paper will focus primarily on their use in fitting a linear model. While previous work has focused mainly on estimation of the coefficients, we will, with motivation from our example, also examine estimation of the variance around the regression line. In addressing these problems, we also discuss the appropriate manner in which to bootstrap for both inferences and bias assessment. The various methods are compared via simulation, and the results are illustrated using our motivating data, for which the goal is to relate the methylation rate of a blood sample to the age of the individual providing the sample. Copyright © 2016 John Wiley & Sons, Ltd.
- 30 August 2011
- Published by SAGE Publications in Statistical Methods in Medical Research
- Vol. 23 (3) , 218-243
On the power of the Cochran–Armitage test for trend in the presence of misclassification
The Cochran–Armitage (CA) test is commonly used in both epidemiology and genetics to test for linear trend in two-way tables with a binary outcome. There has been increasing interest in the power and size of the test and in determination of sample size, especially when there is potential misclassification in the ‘exposure’ category. This article provides a unified approach to determination of the power function over different sampling strategies (fixed overall sample size or fixed marginal sample sizes) and allowing for misclassification in one or both variables. The misclassification may be either differential or non-differential. In addition to the standard CA test, results are also given which provide some insight into the performance of the modified CA test, which utilizes a standard error obtained without invoking the null hypothesis. Even without misclassification, some new expressions are also obtained for determining power with a fixed overall sample size. Numerical illustrations are presented with an emphasis on the more commonly occurring problem of misclassification in the exposure category.
- 30 August 2011
- Published by SAGE Publications in Statistical Methods in Medical Research
- Vol. 23 (3) , 218-243
On the power of the Cochran–Armitage test for trend in the presence of misclassification
The Cochran–Armitage (CA) test is commonly used in both epidemiology and genetics to test for linear trend in two-way tables with a binary outcome. There has been increasing interest in the power and size of the test and in determination of sample size, especially when there is potential misclassification in the ‘exposure’ category. This article provides a unified approach to determination of the power function over different sampling strategies (fixed overall sample size or fixed marginal sample sizes) and allowing for misclassification in one or both variables. The misclassification may be either differential or non-differential. In addition to the standard CA test, results are also given which provide some insight into the performance of the modified CA test, which utilizes a standard error obtained without invoking the null hypothesis. Even without misclassification, some new expressions are also obtained for determining power with a fixed overall sample size. Numerical illustrations are presented with an emphasis on the more commonly occurring problem of misclassification in the exposure category.
- 1 May 2009
- Published by Wiley in Ecological Monographs
- Vol. 79 (2) , 299-324
Statistical methods to correct for observation error in a density‐independent population model
The problem of observation error in assessing the dynamics of populations over time has received increasing attention of late. Of particular interest has been a density‐independent dynamic model, which allows a trend and is commonly employed in population viability analysis (PVA). Most of the recent work in this area has focused on assessing the impact of the observation error and on finding corrected estimators, primarily under normal models with the observation errors assumed to have a constant variance. This paper provides a comprehensive overview of statistical methods for this problem and evaluates them through simulations. This includes the development and assessment of simple and practical ways to obtain standard errors and confidence intervals for the basic parameters in the model and functions of them, such as the intrinsic rate of increase or the probability of eventual extinction. We allow for unequally spaced data and possibly changing observation error variances, and we also discuss how to employ standard errors that often accompany the estimated abundances. Both likelihood techniques under normality and methods allowing non‐normal observation errors are discussed, and we describe how the various likelihood‐based techniques can be implement using standard mixed‐model software, such as PROC MIXED in SAS. The projection technique we use to handle functions of the basic parameters is also valuable when there is no observation error. The methods are motivated and illustrated using data for grizzly bears, Whooping Cranes, California Condors, and Puerto Rican Parrots. We present a simulation experiment to evaluate the performance of the some of the methods.
- 1 May 2009
- Published by Wiley in Ecological Monographs
- Vol. 79 (2) , 299-324
Statistical methods to correct for observation error in a density‐independent population model
The problem of observation error in assessing the dynamics of populations over time has received increasing attention of late. Of particular interest has been a density‐independent dynamic model, which allows a trend and is commonly employed in population viability analysis (PVA). Most of the recent work in this area has focused on assessing the impact of the observation error and on finding corrected estimators, primarily under normal models with the observation errors assumed to have a constant variance. This paper provides a comprehensive overview of statistical methods for this problem and evaluates them through simulations. This includes the development and assessment of simple and practical ways to obtain standard errors and confidence intervals for the basic parameters in the model and functions of them, such as the intrinsic rate of increase or the probability of eventual extinction. We allow for unequally spaced data and possibly changing observation error variances, and we also discuss how to employ standard errors that often accompany the estimated abundances. Both likelihood techniques under normality and methods allowing non‐normal observation errors are discussed, and we describe how the various likelihood‐based techniques can be implement using standard mixed‐model software, such as PROC MIXED in SAS. The projection technique we use to handle functions of the basic parameters is also valuable when there is no observation error. The methods are motivated and illustrated using data for grizzly bears, Whooping Cranes, California Condors, and Puerto Rican Parrots. We present a simulation experiment to evaluate the performance of the some of the methods.
- 15 September 2007
- Published by Springer Nature in Biodiversity and Conservation
- Vol. 17 (11) , 2691-2700
Avian population trends in the vulnerable montane forests of the Northern Appalachians, USA
- 15 September 2007
- Published by Springer Nature in Biodiversity and Conservation
- Vol. 17 (11) , 2691-2700
Avian population trends in the vulnerable montane forests of the Northern Appalachians, USA
- 1 November 2006
- Published by Elsevier in Theoretical Population Biology
- Vol. 70 (3) , 322-335
Modeling observation error and its effects in a random walk/extinction model
- 1 November 2006
- Published by Elsevier in Theoretical Population Biology
- Vol. 70 (3) , 322-335
Modeling observation error and its effects in a random walk/extinction model
- 15 September 2006
- Published by Wiley
F ieller's Theorem