Predictors of locating women six to eight years after contact: internet resources at recruitment may help to improve response rates in longitudinal research
1 Osteoporosis Research Program, Women's College Hospital, Toronto, Canada
2 Department of Health Policy, Management and Evaluation, University of Toronto, Toronto, Canada
3 Health Care and Outcomes Research, University Health Network, Toronto, Canada
4 Department of Public Health Sciences, University of Toronto, Toronto, Canada
5 Institute for Work and Health, Toronto, Canada
6 Mobility Program Clinical Research Unit, St. Michael's Hospital, Toronto, Canada
7 Department of Physical Therapy, University of Toronto, Toronto, Canada
8 Division of Rheumatology, Women's College Hospital, Toronto, Canada
BMC Medical Research Methodology 2007, 7:22 doi:10.1186/1471-2288-7-22Published: 18 June 2007
The ability to locate those sampled has important implications for response rates and thus the success of survey research. The purpose of this study was to examine predictors of locating women requiring tracing using publicly available methods (primarily Internet searches), and to determine the additional benefit of vital statistics linkages.
Random samples of women aged 65–89 years residing in two regions of Ontario, Canada were selected from a list of those who completed a questionnaire between 1995 and 1997 (n = 1,500). A random sample of 507 of these women had been searched on the Internet as part of a feasibility pilot in 2001. All 1,500 women sampled were mailed a newsletter and information letter prior to recruitment by telephone in 2003 and 2004. Those with returned mail or incorrect telephone number(s) required tracing. Predictors of locating women were examined using logistic regression.
Tracing was required for 372 (25%) of the women sampled, and of these, 181 (49%) were located. Predictors of locating women were: younger age, residing in less densely populated areas, having had a web-search completed in 2001, and listed name identified on the Internet prior to recruitment in 2003. Although vital statistics linkages to death records subsequently identified 41 subjects, these data were incomplete.
Prospective studies may benefit from using Internet resources at recruitment to determine the listed names for telephone numbers thereby facilitating follow-up tracing and improving response rates. Although vital statistics linkages may help to identify deceased individuals, these may be best suited for post hoc response rate adjustment.