Health Estimates Using Survey Raked-Weighting Techniques in an Australian Population Health Surveillance System

Abstract

A challenge for population health surveillance systems using telephone methodologies is to maintain representative estimates as response rates decrease. Raked weighting, rather than conventional poststratification methodologies, has been developed to improve representativeness of estimates produced from telephone-based surveillance systems by incorporating a wider range of sociodemographic variables using an iterative proportional fitting process. This study examines this alternative weighting methodology with the monthly South Australian population health surveillance system report of randomly selected people of all ages in 2013 (n = 7,193) using computer-assisted telephone interviewing. Poststratification weighting used age groups, sex, and area of residence. Raked weights included an additional 6 variables: dwelling status, number of people in household, country of birth, marital status, educational level, and highest employment status. Most prevalence estimates (e.g., diabetes and asthma) did not change when raked weights were applied. Estimates that changed by at least 2 percentage points (e.g., tobacco smoking and mental health conditions) were associated with socioeconomic circumstances, such as dwelling status, which were included in the raked-weighting methodology. Raking methodology has overcome, to some extent, nonresponse bias associated with the sampling methodology by incorporating lower socioeconomic groups and those who are routinely not participating in population surveys into the weighting formula.

Including mobile-only telephone users in a statewide preventive health survey—Differences in the prevalence of health risk factors and impact on trends

“…This indicates that weighting alone does not adequately compensate for non-coverage, a finding that differed from other Australian research (e.g. Dal Grande et al, 2015). This difference may be due to the geographically stratified sampling design in the current study which limited the inclusion of characteristics that influence non-response into post-stratification weights (Baffour et al, 2016b, Battaglia et al, 2008).…”

“…These weighting adjustments were used because they were simple and provided consistent representative results. Additional auxiliary variables, such as employment, education and length of residence are known to be associated with non-response (Baffour et al, 2016a, Dal Grande et al, 2015) but due to data availability were not considered.…”

“…This weighting adjusts the sample to match the socio-demographic composition of the population, but there is potential non-coverage bias due to the exclusion of the mobile-only subpopulation, especially if this group is not uniformly distributed in the population (Blumberg and Luke, 2007). In Australia, weighting landline telephone respondents using the Australian Census of Population and Housing data has corrected for bias in prevalence estimates in several studies (Dal Grande et al, 2015, Dal Grande and Taylor, 2010, Wilson et al, 1999). In particular, Dal Grande et al (2015) used a raking methodology to weight estimates and reduce bias from the SAMSS survey which resulted in estimates comparable to those from the 2011 census for South Australia.…”

“…In Australia, weighting landline telephone respondents using the Australian Census of Population and Housing data has corrected for bias in prevalence estimates in several studies (Dal Grande et al, 2015, Dal Grande and Taylor, 2010, Wilson et al, 1999). In particular, Dal Grande et al (2015) used a raking methodology to weight estimates and reduce bias from the SAMSS survey which resulted in estimates comparable to those from the 2011 census for South Australia. The authors found that raking outperformed traditional post-stratification weighting methods for estimating several health indicators as assessed by comparing their estimates to those obtained from the 2013 SAHOS and the 2011–12 South Australian sub-sample from the Australian NHS.…”

“…At the state level, data are predominantly collected by telephone (Barr et al, 2014, Clemens et al, 2014, DHHS, 2015, Tomlin et al, 2016). Whilst the importance of state-wide surveys to national statistical infrastructure is recognised, inclusion of mobile telephones in these surveys has been slow due to the lack of geographical identifiers for Australian mobile telephone numbers (Dal Grande et al, 2015), and prohibitive costs due to screening numbers from unwanted states and territories. Despite this challenge, New South Wales (NSW), Victoria and Queensland have recently transitioned to frames that include both mobile and landline telephone numbers.…”