Abstract
We assessed potential non‐response bias in obtaining information on residential extremely low‐frequency power frequency magnetic field (MF) in Taiwan. All households occupied by children aged less than 7 years in two study districts, one in an urban town and the other in a rural town, were visited and solicited for on‐site measurements in late 2003. The initial response rate was only 32% (33/104, urban) and 60% (61/101, rural). In the same season 1 year later, we performed a second survey of those who declined to be measured at the initial survey and successfully measured another 77 residences (50 and 27 for urban and rural districts, respectively). The two districts were selected mainly because the local public health officers were quite willing to assist the initial survey and to inform residents of the second survey. Except for meteorological conditions, the two surveys came up with very similar findings regarding residential characteristics and power facilities surrounding the houses. The mean residential MF for the urban residences was .121 and .140 micro‐Tesla (µT) (P = .620) for the two surveys. The corresponding figures for the rural residences were .119 and .115 µT (P = .802). Although limited in its scope, this study tends to indicate that measurement studies of residential MF are less likely to suffer from serious selection bias if sampling is confined within a small district where people have similar socioeconomic characteristics. Bioelectromagnetics 28:340–348, 2007. © 2007 Wiley‐Liss, Inc.