Mark W Smith1, Jun Ma, Randall S Stafford. 1. Health Economics Resource Center, U.S. Department of Veterans Affairs, VA Palo Alto HCS, 795 Willow Road (152 MPD), Menlo Park, CA 94025, USA. mark.smith9@va.gov
Abstract
OBJECTIVE: A common form of validation study compares alternative methods for collecting data. The Bland-Altman plot pairs observations across methods and plots their mean values vs. their difference. This method provides only limited information, however, when the range of observed values is small relative to the number of observations. This brief report shows how adding a simple bar chart to a Bland-Altman plot adds essential additional information. STUDY DESIGN AND SETTING: The methodological approach is illustrated using data from a randomized controlled clinical trial of patients in a U.S. county health system. RESULTS: When the number of unique values is small, a Bland-Altman plot alone may provide inadequate information. Adding a bar chart yields new and essential information about agreement, bias, and heteroscedasticity. CONCLUSION: Studies validating one data-collection method against another can be performed successfully even when the number of unique values is small. Published by Elsevier Inc.
OBJECTIVE: A common form of validation study compares alternative methods for collecting data. The Bland-Altman plot pairs observations across methods and plots their mean values vs. their difference. This method provides only limited information, however, when the range of observed values is small relative to the number of observations. This brief report shows how adding a simple bar chart to a Bland-Altman plot adds essential additional information. STUDY DESIGN AND SETTING: The methodological approach is illustrated using data from a randomized controlled clinical trial of patients in a U.S. county health system. RESULTS: When the number of unique values is small, a Bland-Altman plot alone may provide inadequate information. Adding a bar chart yields new and essential information about agreement, bias, and heteroscedasticity. CONCLUSION: Studies validating one data-collection method against another can be performed successfully even when the number of unique values is small. Published by Elsevier Inc.