Link to the University of Pittsburgh Homepage
Link to the University Library System Homepage Link to the Contact Us Form

A Wilcoxon-type Statistic for Repeated Binary Measures with Dropouts and Possible Multiple Outcomes

Elci, Okan Umit (2011) A Wilcoxon-type Statistic for Repeated Binary Measures with Dropouts and Possible Multiple Outcomes. Doctoral Dissertation, University of Pittsburgh. (Unpublished)

Primary Text

Download (722kB) | Preview


In clinical trials, we often compare two treatment groups using repeated binary measures over time. In such trials, we may encounter missing observations, adverse side effects, or non-responsiveness to therapy which for ethical reasons, may result in increased medical intervention beyond the protocol therapy. We developed a family of statistical tests based on the Wilcoxon statistic which orders the vectors of repeated binary observations and events where the ordering is determined by 'clinical relevance'. For some scenarios, clinically meaningful ordering of the vectors may be defined by a natural algorithm, while for other scenarios the ordering is obtained from a group of clinicians. We present the statistical development of the proposed method, effects of the variability of rankings among clinicians, examples of the application of the proposed method using data from a clinical trial on otitis media, and simulation studies comparing the statistical power of the proposed method to more traditional methods of analysis. Our simulation studies indicate that the proposed method is competitive with and, for some scenarios, is preferable to the traditional methods. Although the proposed method is not applicable to every situation, we believe that for some diseases and scenarios, this simple method is noteworthy in the sense that it can be adjusted to extremely complex situations if vectors can be hierarchically ordered in a reasonable fashion, it can be focused on alternatives that have high clinical relevance, and it can be readily adapted to accommodate non-protocol 'outcomes' and missing data. The public health relevance of this study is that clinically meaningful results can be targeted in clinical trials.


Social Networking:
Share |


Item Type: University of Pittsburgh ETD
Status: Unpublished
CreatorsEmailPitt UsernameORCID
Elci, Okan
ETD Committee:
TitleMemberEmail AddressPitt UsernameORCID
Committee ChairRockette, Howard Eherbst@pitt.eduHERBST
Committee MemberMazumdar, Satimaz1@pitt.eduMAZ1
Committee MemberAnderson, Stewart Jsja@pitt.eduSJA
Committee MemberSereika, Susan Sssereika@pitt.eduSSEREIKA
Date: 31 January 2011
Date Type: Completion
Defense Date: 27 May 2010
Approval Date: 31 January 2011
Submission Date: 29 November 2010
Access Restriction: 5 year -- Restrict access to University of Pittsburgh for a period of 5 years.
Institution: University of Pittsburgh
Schools and Programs: School of Public Health > Biostatistics
Degree: PhD - Doctor of Philosophy
Thesis Type: Doctoral Dissertation
Refereed: Yes
Uncontrolled Keywords: binary outcomes; Longitudinal data; natural algorithm; ordering; Wilcoxon test statistic
Other ID:, etd-11292010-102813
Date Deposited: 10 Nov 2011 20:06
Last Modified: 15 Nov 2016 13:52


Monthly Views for the past 3 years

Plum Analytics

Actions (login required)

View Item View Item