D E Neumark1, M Stommel, C W Given, B A Given. 1. Department of Family and Child Ecology, College of Human Ecology, Michigan State University, East Lansing, USA. neumark@msu.edu
Abstract
BACKGROUND: While recognized that loss of subjects over time may introduce bias and complicate statistical analysis in panel studies, it is seldom acknowledged that sampling bias starts with subjects who are eligible but do not participate. OBJECTIVES: Community-based recruiters identified 1,948 subjects as eligible to participate in a descriptive panel survey of older families with cancer. Focusing on the time between identification of eligible subjects until contact for the first interview for consenting subjects, the purpose of this study was to determine if subject or research design factors predicted who was more prone to nonparticipation. METHOD: A multivariate model explored the simultaneous effects of subject and research design characteristics on nonparticipation. Demographic and cancer characteristics, as well as features of the study protocol, were used as predictors in a multinomial logit regression model to enable a three-way comparison between nonconsenters (n = 748), consenters who dropped out prior to data collection (n = 208), and consenters who participated in the intake interview (n = 992). RESULTS: Age and cancer diagnosis played roles in whether consent was obtained, whereas race did not affect consent but raised odds of drop out after consent. Powerful evidence emerged that design features, such as if a caregiver participated, recruitment staff roles, and payment to recruiters, affected the probability of subjects not participating or dropping out before being interviewed. CONCLUSIONS: Findings suggest that both subject and research design characteristics affect the likelihood of nonparticipation in a panel study of older cancer patients and family caregivers. Future research involving testing of strategies addressing access and accrual issues, research staff roles, reimbursements, and responsiveness to the needs of research personnel, ill participants, and family members is warranted.
BACKGROUND: While recognized that loss of subjects over time may introduce bias and complicate statistical analysis in panel studies, it is seldom acknowledged that sampling bias starts with subjects who are eligible but do not participate. OBJECTIVES: Community-based recruiters identified 1,948 subjects as eligible to participate in a descriptive panel survey of older families with cancer. Focusing on the time between identification of eligible subjects until contact for the first interview for consenting subjects, the purpose of this study was to determine if subject or research design factors predicted who was more prone to nonparticipation. METHOD: A multivariate model explored the simultaneous effects of subject and research design characteristics on nonparticipation. Demographic and cancer characteristics, as well as features of the study protocol, were used as predictors in a multinomial logit regression model to enable a three-way comparison between nonconsenters (n = 748), consenters who dropped out prior to data collection (n = 208), and consenters who participated in the intake interview (n = 992). RESULTS: Age and cancer diagnosis played roles in whether consent was obtained, whereas race did not affect consent but raised odds of drop out after consent. Powerful evidence emerged that design features, such as if a caregiver participated, recruitment staff roles, and payment to recruiters, affected the probability of subjects not participating or dropping out before being interviewed. CONCLUSIONS: Findings suggest that both subject and research design characteristics affect the likelihood of nonparticipation in a panel study of older cancerpatients and family caregivers. Future research involving testing of strategies addressing access and accrual issues, research staff roles, reimbursements, and responsiveness to the needs of research personnel, ill participants, and family members is warranted.
Authors: Allison J Applebaum; Wendy G Lichtenthal; Hayley A Pessin; Julia N Radomski; N Simay Gökbayrak; Aviva M Katz; Barry Rosenfeld; William Breitbart Journal: Psychooncology Date: 2011-07-12 Impact factor: 3.894
Authors: Jennifer H Lingler; Karen L Schmidt; Amanda L Gentry; Lu Hu; Lauren A Terhorst Journal: J Empir Res Hum Res Ethics Date: 2014-10 Impact factor: 1.742
Authors: Martine T E Puts; Johanne Monette; Veronique Girre; Christina Wolfson; Michele Monette; Gerald Batist; Howard Bergman Journal: BMC Cancer Date: 2009-08-10 Impact factor: 4.430
Authors: Marloes Oldenkamp; Rafael P M Wittek; Mariët Hagedoorn; Ronald P Stolk; Nynke Smidt Journal: BMC Public Health Date: 2016-06-08 Impact factor: 3.295