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

Semiparametric estimation procedures using local polynomial smoothing for inconsistently measured longitudinal data

Ye, Lei (2015) Semiparametric estimation procedures using local polynomial smoothing for inconsistently measured longitudinal data. Doctoral Dissertation, University of Pittsburgh. (Unpublished)

[img]
Preview
PDF
Submitted Version

Download (883kB)

Abstract

For longitudinal data analyses, existing statistical methods can be used when the independent and dependent variables are measured at the same frequency. In Part 1 of this dissertation, we propose a three-step estimation procedure using local polynomial smoothing for longitudinal data where the variables to be handled are repeatedly measured with different frequencies within the same time period. We first inserted pseudo data for the less frequently measured variable. Then, standard linear regressions were fitted at each time point to obtain raw estimates. Lastly, local polynomial smoothing with analytical weights was applied to smooth the raw estimates. The results showed using analytic weights instead of a kernel function during smoothing is critical when the raw estimates have extreme values, or the association between the dependent and independent variables is nonlinear. In Part 2 of this dissertation, we propose another semiparametric estimation procedure to solve the same problem. After imputing pseudo data for the less frequently measured variable, local polynomial smoothing with analytical weights was applied to smooth the pseudo data for one subject at a time. Then, a suitable parametric mixed-effects model was applied. The results showed that using different types of analytic weights during smoothing produced similar results. Both proposed methods work better when the variances of the repeated measures are small and the within-subjects correlations are high. Statement of Public Health Relevance: The proposed methods are good tools for exploring inconsistently measured longitudinal data. They provide estimation without losing information that has been collected. It is important to biomedical studies especially when many researchers are using diary-based methods to improve the data collection process. For example, paper diaries, personal digital assistants (PDA) and smart phones have been used in the weight loss clinical trials to collect intensive longitudinal data that reflect subjects’ real life experiences and behaviors. The proposed methods can be used when the inconsistent measure problem is present in a longitudinal study.


Share

Citation/Export:
Social Networking:
Share |

Details

Item Type: University of Pittsburgh ETD
Status: Unpublished
Creators/Authors:
CreatorsEmailPitt UsernameORCID
Ye, Leiley9@pitt.eduLEY9
ETD Committee:
TitleMemberEmail AddressPitt UsernameORCID
Committee ChairYouk, Adaayouk@pitt.eduAYOUK
Committee MemberBurke, Lora Elbu100@pitt.eduLBU100
Committee MemberSereika, Susan Mssereika@pitt.eduSSEREIKA
Committee MemberAnderson, Stewart J.sja@pitt.eduSJA
Date: 28 January 2015
Date Type: Publication
Defense Date: 4 December 2014
Approval Date: 28 January 2015
Submission Date: 23 November 2014
Access Restriction: 2 year -- Restrict access to University of Pittsburgh for a period of 2 years.
Number of Pages: 169
Institution: University of Pittsburgh
Schools and Programs: Graduate School of Public Health > Biostatistics
Degree: PhD - Doctor of Philosophy
Thesis Type: Doctoral Dissertation
Refereed: Yes
Uncontrolled Keywords: Analytical weight; Inconsistent sampling; Local polynomial smoothing; Longitudinal data; Repeated measures; Semiparametric
Date Deposited: 28 Jan 2015 17:20
Last Modified: 01 Jan 2017 06:15
URI: http://d-scholarship.pitt.edu/id/eprint/23598

Metrics

Monthly Views for the past 3 years

Plum Analytics


Actions (login required)

View Item View Item