### Abstract

Misclassification is a long-standing statistical problem in epidemiology. In many real studies, either an exposure or a response variable or both may be misclassified. As such, potential threats to the validity of the analytic results (e.g., estimates of odds ratios) that stem from misclassification are widely discussed in the literature. Much of the discussion has been restricted to the nondifferential case, in which misclassification rates for a particular variable are assumed not to depend on other variables. However, complex differential misclassification patterns are common in practice, as we illustrate here using bacterial vaginosis and Trichomoniasis data from the HIV Epidemiology Research Study (HERS). Therefore, clear illustrations of valid and accessible methods that deal with complex misclassification are still in high demand. We formulate a maximum likelihood (ML) framework that allows flexible modeling of misclassification in both the response and a key binary exposure variable, while adjusting for other covariates via logistic regression. The approach emphasizes the use of internal validation data in order to evaluate the underlying misclassification mechanisms. Data-driven simulations show that the proposed ML analysis outperforms less flexible approaches that fail to appropriately account for complex misclassification patterns. The value and validity of the method are further demonstrated through a comprehensive analysis of the HERS example data.

Original language | English (US) |
---|---|

Pages (from-to) | 1605-1620 |

Number of pages | 16 |

Journal | Statistics in Medicine |

Volume | 34 |

Issue number | 9 |

DOIs | |

State | Published - Apr 30 2015 |

### Fingerprint

### Keywords

- Likelihood
- Logistic regressions
- Misclassification
- Odds ratio

### ASJC Scopus subject areas

- Epidemiology
- Statistics and Probability

### Cite this

*Statistics in Medicine*,

*34*(9), 1605-1620. https://doi.org/10.1002/sim.6440

**Binary regression with differentially misclassified response and exposure variables.** / Tang, Li; Lyles, Robert H.; King, Caroline C.; Celentano, David D.; Lo, Yungtai.

Research output: Contribution to journal › Article

*Statistics in Medicine*, vol. 34, no. 9, pp. 1605-1620. https://doi.org/10.1002/sim.6440

}

TY - JOUR

T1 - Binary regression with differentially misclassified response and exposure variables

AU - Tang, Li

AU - Lyles, Robert H.

AU - King, Caroline C.

AU - Celentano, David D.

AU - Lo, Yungtai

PY - 2015/4/30

Y1 - 2015/4/30

N2 - Misclassification is a long-standing statistical problem in epidemiology. In many real studies, either an exposure or a response variable or both may be misclassified. As such, potential threats to the validity of the analytic results (e.g., estimates of odds ratios) that stem from misclassification are widely discussed in the literature. Much of the discussion has been restricted to the nondifferential case, in which misclassification rates for a particular variable are assumed not to depend on other variables. However, complex differential misclassification patterns are common in practice, as we illustrate here using bacterial vaginosis and Trichomoniasis data from the HIV Epidemiology Research Study (HERS). Therefore, clear illustrations of valid and accessible methods that deal with complex misclassification are still in high demand. We formulate a maximum likelihood (ML) framework that allows flexible modeling of misclassification in both the response and a key binary exposure variable, while adjusting for other covariates via logistic regression. The approach emphasizes the use of internal validation data in order to evaluate the underlying misclassification mechanisms. Data-driven simulations show that the proposed ML analysis outperforms less flexible approaches that fail to appropriately account for complex misclassification patterns. The value and validity of the method are further demonstrated through a comprehensive analysis of the HERS example data.

AB - Misclassification is a long-standing statistical problem in epidemiology. In many real studies, either an exposure or a response variable or both may be misclassified. As such, potential threats to the validity of the analytic results (e.g., estimates of odds ratios) that stem from misclassification are widely discussed in the literature. Much of the discussion has been restricted to the nondifferential case, in which misclassification rates for a particular variable are assumed not to depend on other variables. However, complex differential misclassification patterns are common in practice, as we illustrate here using bacterial vaginosis and Trichomoniasis data from the HIV Epidemiology Research Study (HERS). Therefore, clear illustrations of valid and accessible methods that deal with complex misclassification are still in high demand. We formulate a maximum likelihood (ML) framework that allows flexible modeling of misclassification in both the response and a key binary exposure variable, while adjusting for other covariates via logistic regression. The approach emphasizes the use of internal validation data in order to evaluate the underlying misclassification mechanisms. Data-driven simulations show that the proposed ML analysis outperforms less flexible approaches that fail to appropriately account for complex misclassification patterns. The value and validity of the method are further demonstrated through a comprehensive analysis of the HERS example data.

KW - Likelihood

KW - Logistic regressions

KW - Misclassification

KW - Odds ratio

UR - http://www.scopus.com/inward/record.url?scp=84926419719&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=84926419719&partnerID=8YFLogxK

U2 - 10.1002/sim.6440

DO - 10.1002/sim.6440

M3 - Article

VL - 34

SP - 1605

EP - 1620

JO - Statistics in Medicine

JF - Statistics in Medicine

SN - 0277-6715

IS - 9

ER -