### Abstract

In many medical comparative studies (e.g., comparison of two treatments in an otolaryngological study), subjects may produce either bilateral (e.g., responses from a pair of ears) or unilateral (response from only one ear) data. For bilateral cases, it is meaningful to assume that the information between the two ears from the same subject are generally highly correlated. In this article, we would like to test the equality of the successful cure rates between two treatments with the presence of combined unilateral and bilateral data. Based on the dependence and independence models, we study ten test statistics which utilize both the unilateral and bilateral data. The performance of these statistics will be evaluated with respect to their empirical Type I error rates and powers under different configurations. We find that both Rosner's and Wald-type statistics based on the dependence model and constrained maximum likelihood estimates (under the null hypothesis) perform satisfactorily for small to large samples and are hence recommended. We illustrate our methodologies with a real data set from an otolaryngology study.

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

Pages (from-to) | 1515-1529 |

Number of pages | 15 |

Journal | Communications in Statistics: Simulation and Computation |

Volume | 37 |

Issue number | 8 |

DOIs | |

State | Published - Sep 1 2008 |

### Fingerprint

### All Science Journal Classification (ASJC) codes

- Statistics and Probability
- Modeling and Simulation

### Cite this

*Communications in Statistics: Simulation and Computation*,

*37*(8), 1515-1529. https://doi.org/10.1080/03610910802140232

}

*Communications in Statistics: Simulation and Computation*, vol. 37, no. 8, pp. 1515-1529. https://doi.org/10.1080/03610910802140232

**Testing the equality of two proportions for combined unilateral and bilateral data.** / Pei, Yanbo; Tang, Man Lai; Guo, Jianhua.

Research output: Contribution to journal › Article

TY - JOUR

T1 - Testing the equality of two proportions for combined unilateral and bilateral data

AU - Pei, Yanbo

AU - Tang, Man Lai

AU - Guo, Jianhua

PY - 2008/9/1

Y1 - 2008/9/1

N2 - In many medical comparative studies (e.g., comparison of two treatments in an otolaryngological study), subjects may produce either bilateral (e.g., responses from a pair of ears) or unilateral (response from only one ear) data. For bilateral cases, it is meaningful to assume that the information between the two ears from the same subject are generally highly correlated. In this article, we would like to test the equality of the successful cure rates between two treatments with the presence of combined unilateral and bilateral data. Based on the dependence and independence models, we study ten test statistics which utilize both the unilateral and bilateral data. The performance of these statistics will be evaluated with respect to their empirical Type I error rates and powers under different configurations. We find that both Rosner's and Wald-type statistics based on the dependence model and constrained maximum likelihood estimates (under the null hypothesis) perform satisfactorily for small to large samples and are hence recommended. We illustrate our methodologies with a real data set from an otolaryngology study.

AB - In many medical comparative studies (e.g., comparison of two treatments in an otolaryngological study), subjects may produce either bilateral (e.g., responses from a pair of ears) or unilateral (response from only one ear) data. For bilateral cases, it is meaningful to assume that the information between the two ears from the same subject are generally highly correlated. In this article, we would like to test the equality of the successful cure rates between two treatments with the presence of combined unilateral and bilateral data. Based on the dependence and independence models, we study ten test statistics which utilize both the unilateral and bilateral data. The performance of these statistics will be evaluated with respect to their empirical Type I error rates and powers under different configurations. We find that both Rosner's and Wald-type statistics based on the dependence model and constrained maximum likelihood estimates (under the null hypothesis) perform satisfactorily for small to large samples and are hence recommended. We illustrate our methodologies with a real data set from an otolaryngology study.

UR - http://www.scopus.com/inward/record.url?scp=52649173761&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=52649173761&partnerID=8YFLogxK

U2 - 10.1080/03610910802140232

DO - 10.1080/03610910802140232

M3 - Article

AN - SCOPUS:52649173761

VL - 37

SP - 1515

EP - 1529

JO - Communications in Statistics Part B: Simulation and Computation

JF - Communications in Statistics Part B: Simulation and Computation

SN - 0361-0918

IS - 8

ER -