Analysis of Non-Response Bias in a Mailed Health Survey

{ "0" : "Institute of Social and Preventive Medicine, University of Geneva, CMU, Geneva, Switzerland" , "2" : "Response bias" , "3" : "survey methodology" , "4" : "mail surveys"}
Journal of Clinical Epidemiology (Impact Factor: 3.42). 11/1997; 50(10):1123-1128. DOI: 10.1016/S0895-4356(97)00166-2


The objective of this study was to identify characteristics of non-respondents and late respondents to a mailed health survey. Persons who returned and those who did not return the questionnaire were compared using health insurance data, which indicated their age, sex, and health care expenditures in the previous year. Insurance and questionnaire data were used to compare early and late survey respondents and to compare categories of non-respondents. Questions covered use of health services, health status, and sociodemographic characteristics. Participants were members of health insurance plans in Geneva, Switzerland, 19–45 years old (n = 1822). Respondents (n = 1424) and non-respondents (n = 398) were of similar age and sex. The proportion of persons who had health care expenditures greater than zero Swiss francs (SFr) was higher among respondents (75%) than among non-respondents (69%, p = 0.03). Among non-respondents, expenditures of persons who explicitly refused to participate (2378 SFr) were higher than expenditures of persons who moved out of Geneva (1085 SFr) or who failed to return the questionnaire (1592 SFr, p = 02). Among respondents, being born in a Switzerland, having completed elementary school, having generated health care expenditures, and reporting good physical health were independent predictors of early response. In conclusion, low response rates to mailed health surveys may result in overestimating the utilization of health services. However, non-respondents did not constitute a homogenous group, and the strength and even direction of non-response bias depended on the mechanisms of non-response.

10 Reads
    • "Although there are multiple approaches and techniques available for assessing non-response bias (Johnson & Wislar , 2012; Tourangeau & Plewes, 2013), there is little indication as to what might be most effective and efficient for detection of bias in substance use and misuse surveys. Some non-response bias analysis techniques rely on levelof-effort analyses, which, for example, compare the answers of persons who immediately respond to a survey request to those of persons who are more difficult to obtain cooperation from, with the assumption that the latter are more similar to non-respondents (Etter & Perneger, 1997). Another approach employs para-data, such as interviewer observations (Kreuter & Olson, 2013), or preexisting information available in sample frames, to estimate non-response bias (Lewis, Hardy & Snaith, 2013). "
    [Show abstract] [Hide abstract]
    ABSTRACT: This article reviews unfinished business regarding the assessment of substance use behaviors by using survey research methodologies, a practice that dates back to the earliest years of this journal's publication. Six classes of unfinished business are considered including errors of sampling, coverage, non-response, measurement, processing, and ethics. It may be that there is more now that we do not know than when this work began some 50 years ago.
    Substance Use &amp Misuse 09/2015; 50(8-9):1134-8. DOI:10.3109/10826084.2015.1024025 · 1.23 Impact Factor
  • Source
    • "The literature reveals different trends regarding non-response in relation to sex and age. Some reports have found, as we did, overrepresentation of young men among non-responders, while other studies have found no differences with respect to sex or age [14, 19–22]. Low mental health tended to predict non-response in this study, although there was no statistically significant correlation. "
    [Show abstract] [Hide abstract]
    ABSTRACT: Background: A randomized intervention study, "Preventive consultations for 20- to 40-year-old young adults", investigated whether preventive consultations with a general practitioner could help young adults with multiple psychosocial and lifestyle problems to change health behavior. To optimize the response rate of questionnaires at 1 year post-intervention, the non-responders were reminded by telephone. The aim of this study was to examine potential selection bias induced by non-response by comparing responder and non-responder populations at baseline, and to examine the impact on outcomes by comparing initial respondents to respondents after telephone reminding.
    BMC Research Notes 09/2014; 7(1):632. DOI:10.1186/1756-0500-7-632
  • Source
    • "Acock, 2005). Large surveys have pointed that, although respondents and nonrespondents in patient satisfaction surveys may differ according to several demographic and clinical characteristics, differences in satisfaction between them tend to be relatively small (Lasek, Barkley, Harper, & Rosenthal, 1997) and nonrespondents do not constitute a homogenous group (Etter & Perneger, 1997). Many highly sophisticated statistical methods to handle the problem of missing responses are now available (e.g., Little & Rubin, 2002), and the use of them is becoming a standard. "
    [Show abstract] [Hide abstract]
    ABSTRACT: To some extent, results always depend on the methods used, and the complete picture of the phenomenon of interest can be drawn only by combining results of different data processing techniques. This emphasizes the use of a wide arsenal of methods for processing and analyzing patient satisfaction surveys.The purpose of this study was to introduce the self-organizing map (SOM) to nursing science and to illustrate the use of the SOM with patient satisfaction data. The SOM is a widely used artificial neural network suitable for clustering and exploring all kind of data sets. The study was partly a secondary analysis of data collected for the Attractive and Safe Hospital Study from four Finnish hospitals in 2008 and 2010 using the Revised Humane Caring Scale. The sample consisted of 5,283 adult patients. The SOM was used to cluster the data set according to (a) respondents and (b) questionnaire items. The SOM was also used as a preprocessor for multinomial logistic regression. An analysis of missing data was carried out to improve the data interpretation. Combining results of the two SOMs and the logistic regression revealed associations between the level of satisfaction, different components of satisfaction, and item nonresponse. The common conception that the relationship between patient satisfaction and age is positive may partly be due to positive association between the tendency of item nonresponse and age. The SOM proved to be a useful method for clustering a questionnaire data set even when the data set was low dimensional per se. Inclusion of empty responses in analyses may help to detect possible misleading noncausative relationships.
    Nursing Research 09/2014; 63(5):333-345. DOI:10.1097/NNR.0000000000000054 · 1.36 Impact Factor
Show more