Wanjun Liu1, Runze Li2, Marc A Zimmerman3, Maureen A Walton4, Rebecca M Cunningham5, Anne Buu6. 1. Department of Statistics and the Methodology Center, Pennsylvania State University, 413 Thomas Building University Park, PA 16802-2111, USA. Electronic address: wxl204@psu.edu. 2. Department of Statistics and the Methodology Center, Pennsylvania State University, 413 Thomas Building University Park, PA 16802-2111, USA. Electronic address: rzli@psu.edu. 3. Department of Health Behavior and Health Education & Injury Center, University of Michigan, 1415 Washington Heights, Ann Arbor, MI 48109, USA. Electronic address: marcz@umich.edu. 4. Addiction Center & Injury Center, University of Michigan, 4250 Plymouth Road, Ann Arbor, MI 48109, USA. Electronic address: waltonma@umich.edu. 5. Department of Emergency Medicine & Injury Center, University of Michigan, 2800 Plymouth Rd, Bldg 10-G080, Ann Arbor, MI 48109, USA. Electronic address: stroh@med.umich.edu. 6. Department of Health Behavior and Biological Sciences, University of Michigan, 400 North Ingalls, Ann Arbor, MI 48109, USA. Electronic address: buu@umich.edu.
Abstract
BACKGROUND: Retrospective timeline follow-back (TLFB) data and prospective daily process data have been frequently collected in addiction research to characterize behavioral patterns. Although previous validity studies have demonstrated high correlations between these two types of data, the conventional method adopted in those studies was based on summary measures that may lose critical information and the Pearson's correlation coefficient that has an undesirable property. This study proposes the functional concordance correlation coefficient to address these issues. METHODS: We use real data collected from a randomized experiment to demonstrate the applications of the proposed method and compare its analytical results with those of the conventional method. We also conduct a simulation study based on the real data to evaluate the level of overestimation associated with the conventional method. RESULTS: The results of the real data example indicate that the correlation between these two types of data varies across substances (alcohol vs. marijuana) and assessment schedules (daily vs. weekly). Additionally, the correlations estimated by the conventional method tend to be higher than those estimated by the proposed method. The simulation results further show that the magnitude of overestimation associated with the conventional method is greatest when the true correlation is medium. CONCLUSIONS: The findings of the real data example imply that daily assessments are particularly beneficial for characterizing more variable behaviors like alcohol use, whereas weekly assessments may be sufficient for low variation events such as marijuana use. The proposed method is a better approach for evaluating the validity of TLFB data.
BACKGROUND: Retrospective timeline follow-back (TLFB) data and prospective daily process data have been frequently collected in addiction research to characterize behavioral patterns. Although previous validity studies have demonstrated high correlations between these two types of data, the conventional method adopted in those studies was based on summary measures that may lose critical information and the Pearson's correlation coefficient that has an undesirable property. This study proposes the functional concordance correlation coefficient to address these issues. METHODS: We use real data collected from a randomized experiment to demonstrate the applications of the proposed method and compare its analytical results with those of the conventional method. We also conduct a simulation study based on the real data to evaluate the level of overestimation associated with the conventional method. RESULTS: The results of the real data example indicate that the correlation between these two types of data varies across substances (alcohol vs. marijuana) and assessment schedules (daily vs. weekly). Additionally, the correlations estimated by the conventional method tend to be higher than those estimated by the proposed method. The simulation results further show that the magnitude of overestimation associated with the conventional method is greatest when the true correlation is medium. CONCLUSIONS: The findings of the real data example imply that daily assessments are particularly beneficial for characterizing more variable behaviors like alcohol use, whereas weekly assessments may be sufficient for low variation events such as marijuana use. The proposed method is a better approach for evaluating the validity of TLFB data.
Authors: Anne Buu; Runze Li; Maureen A Walton; Hanyu Yang; Marc A Zimmerman; Rebecca M Cunningham Journal: Subst Use Misuse Date: 2014-03-06 Impact factor: 2.164
Authors: Tracy L Simpson; Christopher Galloway; Christina F Rosenthal; Kristen R Bush; Brittney McBride; Daniel R Kivlahan Journal: Am J Addict Date: 2010-11-08
Authors: Anne Buu; Songshan Yang; Runze Li; Marc A Zimmerman; Rebecca M Cunningham; Maureen A Walton Journal: Addict Behav Date: 2019-11-09 Impact factor: 3.913