Search

There are two major errors in medical research. Random errors, which we named as error by chance, and systematic errors, which we called as bias, affect the internal and external validity of researches. The aim of this technical note is to give some information about different bias causes under two major bias headings as “selection and information bias”. It is possible to decrease the probability of random error by increasing the sample size. But systematic error is not prevented by increasing sample size unless the representativeness is not provided. Also information bias, which will occur during planning and conducting of research, will make us away from the truth. Therefore it is very important to consider probable bias causes during planning and conducting of any research from beginning to the end.

It is generally accepted that interviewers have a considerable effect on survey response. The difference between response success and failure does not only affect the response rate, but can also influence the composition of the realized sample or respondent set, and consequently introduce nonresponse bias. To measure these two different aspects of the obtained sample, response propensities will be used. They have an aggregate mean and variance that can both be used to construct quality indicators for the obtained sample of respondents. As these propensities can also be measured on the interviewer level, this allows evaluation of the interviewer group and of the extent to which individual interviewers contribute to a biased respondent set. In this article, a procedure based on a multilevel model with random intercepts and random slopes is elaborated and illustrated. The results show that the procedure is informative to detect influential interviewers with an impact on nonresponse basis.

Refusal by sample units and failure to contact sample units can both contribute to survey non-response bias. However, the nature of the contribution can be rather different in the two cases. Extended field efforts may reduce either or both components of non-response bias. We examine data from several UK surveys on extended field efforts and the effect that they appear to have on non-response bias and its components. Some consistent patterns are found. We also explore the sensitivity of such analyses to the operational definition of extended efforts. The findings provide some evidence of the relative roles of refusal conversion and repeated contact attempts in reducing non-response bias. The study also has implications for the development of appropriate field strategies to combat non-response bias.

A number of researchers have observed that response biases, defined as when subjects respond to items in research instruments in ways that do not coincide with the intent or content of the instrument, suffuse measurements and assessments of mental disorders. They cautioned that the response bias problem has been neglected in mental health research at the price of substantial error. Have the cautions been heeded? Or does the neglect of response bias continue? Articles published in 1998 in three major psychiatric journals were examined: Archives of General Psychiatry, American Journal of Psychiatry, and the Journal of Nervous and Mental Disease. The articles were examined to determine whether response biases were mentioned and whether systematic efforts were made to attend to their influence on the findings of the study. Each article was assessed twice by independent raters. The examination indicates that a very small minority of the articles reviewed mentioned response bias and that among those mentioning it, a minority attempted to control for bias effects. Cautions offered about response bias have not been heeded. Accordingly, the issue is one of how to incorporate concerns about response bias into the institutional structures that influence the culture of mental health research.

Objective: Physician-specific surveys are a frequently used tool in health services research, but attempts at ensuring adequate response rates are rarely reported. We reviewed literature of survey methodology specific to physician surveys and report those found to be most effective. Data Sources: Studies were identified by searching MEDLINE and PSYCHInfo from 1967 through February 1999. We included all English-language studies that randomized physician survey respondents to an experimental or control group. The authors independently extracted data from 24 studies examining survey methodology of physician-specific surveys. We included Mantel–Haenszel chi-squares comparing treatment groups, if present. If not, these were calculated from study data. Results: Pre-notification of survey recipients, personalizing the survey mailout package, and nonmonetary incentives were not associated with increased response rates. Monetary incentives, the use of stamps on both outgoing and return envelopes, and short questionnaires did increase response rates. Few differences were reported in response rates of phone surveys compared with mail surveys and between the demographics and practice characteristics of early survey respondents and late respondents. Conclusions: We report some simple approaches that may significantly increase response rates of mail surveys. Surprisingly, the response rates of mail surveys of physicians compared favorably with those from telephone and personal interview surveys. Nonresponse bias may be of less concern in physician surveys than in surveys of the general public. Future research steps include specifically testing the more compelling results to allow for better control of confounders.

Do studies of time use interfere too much in the lives of the subjects? As a result are those who agree to participate a biased sample of the population? We examine the characteristics of the Experience Sampling Method (ESM) adolescent sample from the Alfred P. Sloan Study of Youth and Social Development in order to detect and quantify instances of sampling and nonresponse bias. According to available proxies for time use and standard demographic variables, the Sloan ESM sample is nearly representative in terms of teen employment rates, parental employment rates, a student's grade point average, and TV watching. Work hours are slightly undercounted in the study because of slightly higher nonresponse rates by teenagers working long hours. The sample is less representative in terms of the time of week and gender; nonresponse is relatively common on school nights and (to a lesser extent) on weekends, and among boys. We offer some suggestions regarding general implications of our findings for the measurement of time use.

Oral health surveys are undertaken to provide estimates of the dental health and behaviors of populations or population subgroups. However, the integrity of the data from sample surveys may be compromised by one or more sources of sampling and nonsampling error. An important source of nonsampling error is the failure to collect data from some of the individuals comprising the sample. Consequently, the response to a sample survey, and the direction and magnitude of bias induced by nonresponse, need to be taken into account when using estimates derived from sample surveys. Although the response rate to a survey is usually used as an indicator of the quality of the data it provides, nonresponse error is a function of nonresponse and the extent of differences in the characteristics of responders and nonresponders. Nonresponse may be managed in two ways. The first is to reduce nonresponse to a minimum using response-enhancement strategies. The second is the post-survey adjustment of data using weighting or imputation techniques to produce estimates that correct for nonresponse. This paper discusses issues concerning response and nonresponse bias in oral health surveys and provides guidelines on the management and reporting of nonresponse. It describes response-enhancement strategies to reduce noncontacts and refusals, sources of data to facilitate the comparison of responders and nonresponders, methods of assessing the degree of bias induced by nonresponse, techniques for producing adjusted survey estimates, and the assumptions on which these procedures and processes are based.

To assess non-participation bias in a survey of male sexual behavior.A household survey was carried out in 1992-1993 using a probability sampling frame in Mexico City. Demographic variables were available for all eligible men. The extent of non-participation bias was estimated using a version of the Heckman method, which utilizes two equations, one to predict participation and the other to predict reports of same-gender sexual behavior.A total of 8,068 of the 13,713 eligible men completed a face-to-face questionnaire (response rate 59%); 173 men (2.1%) reported bisexual behavior in their lifetime, and 37 (0.4%) reported only male partners. Survey participation was predicted using demographic variables: 67% of the observations were correctly predicted by a probit regression model: 82% of participants and 53% of non-participants (pseudo-r2 = 0.13). Same-gender sexual behavior was predicted by variables indicating attachment to gay/bisexual social networks, history of sexually transmitted diseases, positive attitudes towards gay and bisexual males, and lack of support from male relatives. Ninety-seven per cent of the cases was correctly predicted by the probit model (pseudo-r2 = 0.14). The correlation between these two equations was not statistically significant.These results indicate that prevalence estimates of same-gender sexual behavior among Mexico City men were not biased by selective survey participation. Careful selection and training of household interviewers may have assisted in minimizing potential bias.

Employees (N = 194) from a wide variety of organizations participated in this study aimed at describing the attitudes of individuals who refuse to respond to an employee survey request (noncompliants). Noncompliants, in comparison with those individuals who would comply with the survey request, possessed greater intentions to quit, less organizational commitment, and less satisfaction toward supervisors and their own jobs. Noncompliants also possessed more negative beliefs regarding how their organization handles employee survey data (e.g., does not act on survey data). No significant differences were found for work-related demographic variables, satisfaction with pay, and satisfaction with promotion opportunities. Implications for survey research are discussed along with methods to address nonresponse and noncompliance. Organizational surveys are commonly used to assess employee attitudes and perceptions about relevant work-related issues. The value of the survey as a data-gathering tool, however, is dependent on employees participating in the survey effort. Unfortunately for researchers, response rates appear to be on a decline (Schwarz, Groves, & Schuman, 1998). Although this trend has not been systematically analyzed, oversurveying of employees has been cited as a possible root cause (Saari, 1998). A low rate of response to organizational surveys may lead to a number of problems. Most directly, a low response rate results in less data, which in turn may limit both the researcher's choices of and power for statistical tests. Relatedly, sample size is directly and negatively related to the magnitude of obtained confidence intervals about sample statistics (e.g., means, correlation coefficients). A low response rate may also diminish, in the eyes of management and employees, the perceived credibility of the obtained data. Finally, and perhaps most importantly, a low response rate may result in a biased sampling of organizationa l members. Specifically, if survey respondents and nonrespondents differ systematically on the substantive constructs being assessed by the survey, conclusions based solely on respondent data may not generalize to the original sample, and by extension, to the population under study (Rogelberg & Luong, 1998). Therefore, identifying whether respondents differ substantively from nonrespondents to a survey is key in determining whether nonresponse leads to nonresponse bias. Building on this premise, the present study examined the work-related attitudinal characteristics of individuals

Objectives (1) To characterize pharmacists who work part-time and compare them with pharmacists who work a secondary part-time job (moonlighters) with respect to their motivations for working in a part-time job, and (2) to identify the set of characteristics that best describes pharmacists engaged in each of the part-time work arrangements under study. Design A five-page questionnaire was mailed to 3,002 licensed pharmacists nationwide. Two follow-up surveys were sent to nonrespondents at the fifth and ninth weeks from the original mailout to ensure an adequate response rate. Results The net response rate was 57% (n = 1,637). Assessment of the sample did not detect any potential nonresponse bias. Sixty-eight percent of the respondents reported working full time, 13% were moonlighters, 14% were part-timers, and 5% were part-timers in more than one job. Reasons for working part-time were analyzed for the moonlighter and part-timer subsamples. A principal components analysis revealed that both groups have distinct motivations for selecting their chosen work arrangement. Conclusion This study provides some insight into those factors affecting pharmacists' decisions to engage in two different types of part-time work. Since pharmacists' alternative work schedules can have repercussions on the quality of pharmacy being practiced, further research on pharmacists' work arrangement choices should provide useful information for work force-related endeavors.

Introduction: We have reported that men's lifetime sport-caught Great Lakes fish consumption was associated with lifetime prevalence of conception delay or failure. Those cross-sectional data were based on responses to a postal questionnaire. The present study was conducted to evaluate whether nonresponse bias could explain the cross-sectional findings. Methods: We conducted telephone interviews with 230 men and 38 women who did not respond to the original survey, and compared these individuals to the original responders with respect to key demographic, behavioral, and reproductive characteristics. Results: Nonresponders were approximately 1.5 years older at interview, were more likely to be Caucasian, and reported higher incomes than responders. Among men, nonresponders had fished fewer days in the past year (12% reported no fishing, compared to 4.3% of responders). Almost one half of nonresponders reported no fish consumption in the past year, compared to one quarter of responders. Nonresponders were more likely than responders to have ever conceived a live-born child, had more children, and were less likely to intend to have additional children in the next 5 years. Among both responders and nonresponders there was an increased prevalence of a period of conception failure among men who reported consuming greater quantities of sport-caught Great Lakes fish. Discussion: Our study provides support for the cross-sectional analyses presented previously, insofar as nonresponse bias is unlikely to have a major role in the observed association.

Introduction. Part 1: Policy Instruments. 1. Markets for Tradeable Co2 Emission Quotas: Principles and Practice G. Chilchilnisky, G. Heal. 2. Costs of Uniform and Differentiated Charges on a Polluting Input: An Application to Nitrogen Fertilisers in Sweden R. Brannlund and I.-M. Gren. Part 2: Cost-Benefit Analysis. 3. On the Proper Treatment of Defensive Expenditures in 'Green' NNP Measures T. Aronsson, et al. 4. Preserving Species Without an Endangered Species Act: British Columbia's Forest Practices Code G.C. van Kooten. 5. A Cost-Benefit Analysis of an Alternative Waste Treatment in Northern Norway. Use of Refuse Derived Fuel (RDF) instead of Coal in an Industrial Process. A Preliminary Analysis J.A. Riseth, Y. Torbergsen. Part 3: Cost Efficiency. 6. Emission Constraints, Emission Permits and Marginal Abatement Costs G. Heal. 7. Wetlands as a Nitrogen Sink - Estimation of Costs in The Laholm Bay O. Bystrom. 8. Cost Effective Reductions in the Agricultural Load of Nitrogen to the Baltic Sea K. Elofsson. Part 4: Contingent Valuation - Theory and Applications. 9. Valuing the Wolf in Sweden: Are Benefits Contingent on the Supply? M. Boman, G. Bostedt. 10. A Test of Nonresponse Bias in a Mail Contingent Valuation Survey P. Fredman. 11. Determination of WTP for a Change in Water Quality with Free Selection of Reference Points E. Mantymaa. Part 5: Experimental Economics and the Contingent Valuation Method. 12. Statistical Bias Functions: The Alchemy ofContingent Valuation? P. Frykblom. 13. Environmental Damage Assessment with Hypothetical Surveys: The Calibration Approach G.W. Harrison, et al. List of Contributors. Subject Index.

Most surveys using the contingent valuation method (CVM) will be subject to nonresponses in the willingness to pay (WTP) estimate,1 because a response rate of 100% is either practically impossible or very expensive to obtain. This is of special concern for mail surveys since they have shown to be prone to errors as a result of nonresponses (Mitchell and Carson, 1989). Nonresponse biases in CVM surveys can occur when the number of valid WTP responses is smaller than the total sample. This can arise when a member of the sample does not return the questionnaire (unit nonresponse), or when a respondent fails to answer the economic elicitation question (item nonresponse).

Background: Ommoord District residents of Rotterdam, The Netherlands, age 55 and older, completed a two-stage interview to assess the risk factors for chronic disease and disability. Methods: In the in-home Stage I interview (N = 7,983), demographic data and medical history were collected by a trained lay interviewer. During Stage II, a physician interview and examination were conducted at the study center. Subjects (N = 7,129) were asked about their history of psychiatric disorders and 6,596 responded. Results: The lifetime prevalence of self-reported psychiatric disorders was 5.44% for unipolar and bipolar depressive disorders combined, 0.27% for psychotic disorders, 0.08% for alcoholism, 0.05% for drug addiction, and 3.71% for other diagnoses. Residents reporting a psychiatric diagnosis of depression were more likely to be currently taking an antidepressant medication (p < .001) and an antipsychotic medication (p < .0001), to be in current outpatient treatment (p < .001), to have been hospitalized for psychiatric illness (p < .001), to have undergone electroconvulsive treatments (p < .001), and to report a history of having made suicide attempts (p < .001). Conclusion: The self-reported lifetime prevalence of affective disorders was similar to the rates found in the Stirling County Study (Canada) and the Epidemiologic Catchment Area Survey (United States). The lower prevalence rates of the other psychiatric disorders may reflect underreporting or an age cohort effect, or may be due to the nonresponse bias.

Bias is defined as any process that leads to the systematic deviation of study results or inferences from the truth.1 The term bias does not mean that the investigator’s conscious or unconscious views led to a particular subjective interpretation, as in conventional usage of the word. Bias in evidence-based medicine refers to deviations from the truth that generally result from limitations in study design or reporting rather than from personal prejudices. Some forms of bias are inevitable in research, but need to be recognized and accounted for by investigators and readers. Many distinct forms of bias have been recognized. In 1979, Sackett identified 57 types.2 In 2005, Choi and Pak noted 48 varieties of bias solely in studies using questionnaires.3 Dorak described 108 bias categories in 2008.4 I will describe a few …

Due to a lack of pertinent data, little is known about nonresponse in substantive, generally "dependent" variables and its consequences. However, in a study on policy performance of Dutch municipalities, we were fortunately able to gather performance data fur respondents and nonrespondents from independent sources. Our study shows that very successful and very unsuccessful municipalities respond least, while intermediately successful municipalities respond most. In our opinion, this "curvillnear" relationship between performance and nonresponse can be interpreted as an interaction effect between interest in the topic of research and evaluation apprehension. As a consequence, the relationships between variables are also biased. To correct for this bias we made several weighting attempts. Weighting adjustment is based on the: assumption of equal response probabilities. Our study, however, shows that this assumption is not met. Furthermore, our results indicate that weighting by one or two independent variables does not lead to better results, no more than weighting by policy performance alone. Interestingly, only weighting by a dependent and independent variable in combination leads to acceptable results. but only if weighting factors are not trimmed down too strongly and combination of weighting classes is limited. (C) 1998 Academic Press.

In observational studies where average levels or percentiles in a population distribution of nutritional components are to be assessed it is essential that the participating subjects are representative of the population to which the results are to be generalised. The optimum way of achieving this is by random selection from the target population in which all members have equal probability (or at least a known probability) of being chosen and also that all selected subjects participate in the study. The effects of selection bias and non-response can be large and are in general difficult to estimate. In the presentation of results it should always be stated how selection bias and non-response could have influenced the results. In this paper a general selection procedure and certain problems with it is presented. A method for correction of nonresponse bias is also presented. The effect of measurement errors on correlation coefficients between nutritional components or between a nutritional component and other variables is discussed.

OBJECTIVE: To compare the predictive accuracy of two validated indices, one that uses self-reported variables and a second that uses variables derived from administrative data sources, to predict future hospitalization. To compare the predictive accuracy of these same two indices for predicting future functional decline. DESIGN: A longitudinal cohort study with 4 years of follow-up. SETTING: A large staff model HMO in western Washington State. PARTICIPANTS: HMO Enrollees 65 years and older (n = 2174) selected at random to participate in a health promotion trial and who completed a baseline questionnaire. MEASUREMENT: Predicted probabilities from the two indices were determined for study participants for each of two outcomes: hospitalization two or more times in 4 years and functional decline in 4 years, measured by Restricted Activity Days. The two indices included similar demographic characteristics, diagnoses, and utilization predictors. The probabilities from each index were entered into a Receiver Operating Characteristic (ROC) curve program to obtain the Area Under the Curve (AUC) for comparison of predictive accuracy. RESULTS: For hospitalization, the AUC of the self-report and administrative indices were .696 and .694, respectively (difference between curves, P = .828). For functional decline, the AUC of the two indices were .714 and .691, respectively (difference between curves, P = .144). CONCLUSIONS: Compared with a self-report index, the administrative index affords wider population coverage, freedom from nonresponse bias, lower cost, and similar predictive accuracy. A screening strategy utilizing administrative data sources may thus prove more valuable for identifying high risk older health plan enrollees for population-based interventions designed to improve their health status.

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 homogeneous group, and the strength and even direction of non-response bias depended on the mechanisms of non-response.

The aim was to investigate late response and nonresponse bias in a survey related to HIV and infection control. Questionnaires with ID numbers were mailed to a stratified random sample of dentists in Canada with additional mailings 4 and 7 weeks later (n = 6444). We compared responses received after4 weeks, 4-7 weeks,7 weeks. Extrapolation was used to estimate nonresponse bias. Univariate analyses showed significant differences between responses received4 weeks after initial mailing and those received later for items on sociodemographics, knowledge, infection control practices and attitudes: late responders were more likely to report that they would refuse to treat any patients with HIV (P0.01). Multiple logistic regression indicated that the best predictors of responses receivedor = 4 weeks were disagreement that HBV is more infectious than HIV (OR = 1.7); unwillingness to attend a dentist who treats HIV/AIDS patients (OR = 1.3); incorrect perception of the risk of HIV infection after an HIV-contaminated needlestick injury (OR = 1.2); and sometimes or never heat-sterilizing handpieces after each patient (OR = 1.2). Extrapolation indicated that the percentage of all respondents who reported refusal to treat (15.2%) would have been 17.1% if a 100% response rate had been obtained. We found significant evidence of late response and nonresponse bias primarily in knowledge and fears related to HIV infectivity; however, the impact on the final results was small and we concluded that additional follow-up to improve response rates would not be worthwhile.

Background Monitoring health targets is often undertaken using questionnaire surveys of lifestyle risk factors. Non-response bias is recognized but rarely quantified. Methods Following a questionnaire survey on a random sample of 6009 residents of Somerset with a response rate of 57.6 per cent, a telephone survey was undertaken on a random sample of 400 non-responders. A small number of the more important questions from the questionnaire were put to the non-responders over the phone. Results Fifty-nine per cent of the sample were contacted and agreed to participate. Statistically significant differences between responders and non-responders to the original questionnaire were detected for current smoking, hazardous alcohol consumption and lack of moderate or vigorous activity. Conclusions Lifestyle questionnaire surveys need to include an assessment of the non-response bias.

Background Telephone surveys based on samples of landline telephone numbers are widely used to measure the prevalence of health risk behaviours such as smoking, drug use and alcohol consumption. An increasing number of households are relying solely on mobile telephones, creating a potential bias for population estimates derived from landline-based sampling frames which do not incorporate mobile phone numbers. Studies in the US have identified significant differences between landline and mobile telephone users in smoking and alcohol consumption, but there has been little work in other settings or focussed on illicit drugs. Methods This study examined Australian prevalence estimates of cannabis use, tobacco smoking and risky alcohol consumption based on samples selected using a dual-frame (mobile and landline) approach. Respondents from the landline sample were compared both to the overall mobile sample (including respondents who had access to a landline) and specifically to respondents who lived in mobile-only households. Bivariate comparisons were complemented with multivariate logistic regression models, controlling for the effects of basic demographic variables. Results The landline sample reported much lower prevalence of tobacco use, cannabis use and alcohol consumption than the mobile samples. Once demographic variables were adjusted for, there were no significant differences between the landline and mobile respondents on any of the alcohol measures examined. In contrast, the mobile samples had significantly higher rates of cannabis and tobacco use, even after adjustment. Weighted estimates from the dual-frame sample were generally higher than the landline sample across all substances, but only significantly higher for tobacco use. Conclusions Landline telephone surveys in Australia are likely to substantially underestimate the prevalence of tobacco smoking by excluding potential respondents who live in mobile-only households. In contrast, estimates of alcohol consumption and cannabis use from landline surveys are likely to be broadly accurate, once basic demographic weighting is undertaken.

Publisher's version (útgefin grein), Background: This study examines whether alcohol-related hospitalization predicts survey non-response, and evaluates whether this missing data result in biased estimates of the prevalence of hazardous alcohol use and abstinence. Methods: Registry data on alcohol-related hospitalizations during the preceding ten years were linked to two representative surveys. Population data corresponding to the surveys were derived from the Stockholm County registry. The alcohol-related hospitalization rates for survey responders were compared with the population data, and corresponding rates for non-responders were based on the differences between the two estimates. The proportions with hazardous alcohol use and abstinence were calculated separately for previously hospitalized and non-hospitalized responders, and non-responders were assumed to be similar to responders in this respect. Results: Persons with previous alcohol-related admissions were more likely currently to abstain from alcohol (RR=1.58, p, The study was made possible by Stockholm County Council and the Swedish National Institute of Public Health.

Introduction The participation rates in epidemiologic studies have declined in recent decades. Missing data reduce sample size, statistical power, and scientific quality. Objective To study use of home interviews in recruiting individuals with a psychosis. Aims To evaluate effect of home-recruitment on non-response bias. Methods In the Northern Finland 1966 Birth Cohort, field surveys on psychosis were conducted in 1999-2001 and 2008-2010. In order to increase participation in the follow-up sample, cases were offered to be interviewed at home. We studied symptoms, illness severity, functioning, cognition, antipsychotics use, and grey matter (GM) volume between home-recruited and regular participants (RP), and non-participants (NP). Effect sizes (d) were calculated to compare the differences. Results Altogether 18 (33%) out of the follow-up sample (n=54) were home-recruited, 27 did not participate. Home-recruited had more symptoms, lower functioning, cognition and GM volume, and they had used more antipsychotics compared to RP and NP in baseline. NP did not differ from RP. The same differences occurred when home-recruited were compared with RP in the follow-up study. [Selected information from baseline study] Homerecruited participants Regular participants Nonparticipants ANOVA Mean (SD) Mean (SD) Mean (SD) P d1 d2 PANSS, negative 19 (11) 12 (6) 14 (10) 0.028 0.89 0.48 PANSS, disorganization 21 (9) 15 (7) 16 (9) 0.033 0.81 0.60 PANSS, excitement 13 (5) 10 (2) 12 (5) 0.025 0.91 0.38 SOFAS 40 (19) 57 (15) 54 (15) 0.002 -1.04 -0.85 CVLT, total recall 39 (14) 51 (13) 44 (16) 0.032 -0.89 -0.30 Dose years 44 (47) 15 (18) 12 (13) 0.002 0.92 1.00 Grey matter volume 593 (60) 642 (64) 623 (61) 0.039 -0.77 -0.49 Conclusion Owing to the home-recruitment we were able to collect data that may be reasonably non-biased in terms of nonresponse bias, which will yield valid estimates in our future studies on change over time in brain and cognitive ability, prevalence and severity of psychotic illnesses, outcome, medication use, and other issues of interest.

Written, parental consent was solicited from parents of 517 seventh grade students in four junior high schools in Bergen, Norway for their children's participation in a longitudinal nutrition education study. After having received several reminders, 97.3% of the parents gave their written consent. Students whose parents gave an immediate written consent reported more positive eating habits and were more likely to plan for college than were students whose parents responded after receiving one or more reminders. The observed differences did not severely bias our baseline estimates of these variables, and the magnitude of the observed intervention impact did not change as a result of excluding students whose parents were slow to respond. Excluding these students did, however, affect our ability to identify important factors associated with healthy eating behavior including socio-economic status. Subject selection bias may represent a threat to the validity of studies using a written parental consent procedure.

Background. Estimation of non-response bias by modelling prevalence as a function of the number of mailings required to achieve a response, or of the cumulative response, has been advocated, but the models have not incorporated age and sex, differential response rates by age and sex, or season of response. Methods. The effect on age-sex standardized prevalence of estimating non-response bias using a variety of models was investigated using data on nine symptom and medication questions from 13 007 subjects in the three English centres of the European Community Respiratory Health Survey. Comparison was made of goodness of fit and the prediction of responses in a 25% follow-up sample with the observed values. Results. Despite low response rates in Cambridge and significant decreases in prevalence with additional mailings or increasing cumulative response in Norwich, there were only small effects on estimated age-sex standardized prevalences. No model was consistently better for any centre or question. Conclusions. The models are useful for exploring the sensitivity of estimated prevalence to non-response bias, but should be used with caution to adjust estimates. Ideally first mailings should be staggered over the whole year so that mailing and season are not confounded, and sufficient mailings or other contacts carried out for the whole sample to ensure a high response rate.

Contingent Valuation, Transport Safety and the Value of Life, edited by Nathalie G. Schwab Christe and Nils C. Soguel (Kluwer Academic Publishers, 1995). Reviewer: William L. Ferguson, University of Southwestern Louisiana Contingent Valuation, Transport Safety and the Value of Life compiles selected research that was presented to a 1994 multinational conference on the "consequences of road accidents" hosted by the Institute for Regional and Economic Research (IRER) of the University of Neuchatel, Switzerland. The eight papers in the volume review damage valuation techniques applied to transportation situations across European Union countries. The book's primary focus is the contingent valuation method (CVM), which historically was connected with attempts to value noneconomic environmental goods (e.g., safety, recreation, reduction in exposure to a fatal injury, etc.). Various issues and problem areas associated with this genre of valuation process, as well as empirical evaluations and proposed solutions, are explored. The selected papers generally present different applications and evaluation of basic CVM techniques in traditional transportation settings, and the final paper extends the application of CVM to nontransport health outcomes (e.g., cancer, nerve disease). Essentially, CVM is a three-step process that incorporates survey techniques wherein respondents are presented with hypothetical markets in which estimates are generated of subjects' willingness to pay (WTP) to obtain a given level of (or reduce their exposure to) the environmental good in question, and/or their willingness to accept (WTA) certain monetary compensation to induce them to forego (acquire) some level of the good. Appropriate design of the questionnaire and survey method to be employed (e.g., telephone, mall intercept, mail, etc.) is the initial, and most crucial, step in CVM. The good to be valued, including initial and final states, must be clearly described to reduce/avoid bias (e.g., amenity misspecification bias, symbolic bias, objective versus subjective bias, etc.). How the hypothetical market works (e.g., as a public versus private good) also must be explained clearly to reduce/avoid other important sources of bias (e.g., provision bias, strategic bias, nonresponse bias, protest bias, etc.). The second step involves actual conduct of the survey (e.g., open- versus closed-ended format, bidding game, multistage/follow-up sampling, etc.), in which key details and appropriate procedures must be recognized to reduce/avoid other sources of bias (e.g., interviewer bias, starting point bias, range bias, selection bias, etc.). Due to the hypothetical nature of both the market and good in question, appropriate or at least satisfying tradeoffs may have to be made and evaluated for effect in both steps. The final step, analysis and interpretation of the results, including evaluation of construct validity, also depends upon the relative complexity of the survey. "Big picture" practical considerations (e.g., "reasonable" survey length, cost factors, difficulty of subjects to assimilate information, etc.) are given adequate discussion throughout the book. Four papers provide design details and empirical results of CVM studies for auto-related injuries in Switzerland, Denmark, Sweden, and France. Each of these papers highlights positive and negative aspects of the CVM process mentioned above. …

Background Fatigue is an underestimated cause of underperformance among physicians. There is evidence that fatigue or other byproducts of production pressure may negatively influence the quality of colonoscopy. Objective To investigate the practices and perceptions of U.S. endoscopists regarding the effect of production pressure on the performance of colonoscopy. Design We conducted a 40-question online survey to assess endoscopists' practices and perceptions concerning production pressure. Setting A total of 5030 U.S. American Society for Gastrointestinal Endoscopy members. Main Outcome Measurements The proportion of endoscopists responding positively to questions pertaining to the impact of production pressure on colonoscopy practice. Results Ninety-two percent of respondents indicated that production pressure influenced one or more aspects of their endoscopic practices. Examples of production pressure included (1) postponing polypectomy for a subsequent session (2.8%), (2) reducing the length of time spent inspecting the colon (7.2%), and (3) proceeding with colonoscopy in a patient with an unfavorable risk/benefit ratio (69.2%). Forty-eight percent of respondents reported witnessing the effects of production pressure on a colleague. Respondents working fee-for-service and those with >10 years since completion of fellowship were more likely to describe their weekly workloads as excessive compared with those who were salaried (81.3% vs 71.3%; P = .01) and P = .01). Limitations Nonresponse bias due to low response rate (22.3%). Conclusion Production pressure influences the conduct of colonoscopy for many endoscopists and could have an adverse effect on the outcome of colorectal cancer screening. (Clinical trial registration number: RE:GIE D 11-01288R1.) The study was an Internet study and did not involve human subjects.

Given concerns about survey nonresponse bias as well as the need to plan resources for participant recruitment, this study tracked each step of the recruitment process (location, response, consent, and completion) of sociodemographically diverse older women for a survey concerning mammography experience. Younger, less educated poor women were likely to be lost due to inability to locate them, while older middle and upper-economic-group women were more likely to be lost due to refusal to participate. Hispanic and Black women were significantly more likely to respond on successive attempts to recruit them than were White, non-Hispanic women. There was no significant difference in refusal rates by minority women over the successive contacts, as contrasted with White women, who refused at significantly higher rates with each attempt.

A review of the last three volumes of AJIC (1990, 1991, and 1992, numbers 1 through 3) revealed a modest use of statistics in infection control research. Discussion of the statistics most commonly used results in the following recommendations: (1) Researchers who use the t test should report the t test statistic, exact significance probabilities, and a discussion of assumptions and requirements underlying the analysis. (2) A 95% confidence interval for estimated quantities is preferred to p values. (3) The potential impact of nonresponse bias should be included in reports of survey data. (4) An adequate description of statistical methods should be included in all reports of survey data. (5) An adequate description of statistical methods should be included in all research articles.

Background Nonresponse bias in a longitudinal study could affect the magnitude and direction of measures of association. We identified sociodemographic, behavioral, military, and health-related predictors of response to the first follow-up questionnaire in a large military cohort and assessed the extent to which nonresponse biased measures of association. Methods Data are from the baseline and first follow-up survey of the Millennium Cohort Study. Seventy-six thousand, seven hundred and seventy-five eligible individuals completed the baseline survey and were presumed alive at the time of follow-up; of these, 54,960 (71.6%) completed the first follow-up survey. Logistic regression models were used to calculate inverse probability weights using propensity scores. Results Characteristics associated with a greater probability of response included female gender, older age, higher education level, officer rank, active-duty status, and a self-reported history of military exposures. Ever smokers, those with a history of chronic alcohol consumption or a major depressive disorder, and those separated from the military at follow-up had a lower probability of response. Nonresponse to the follow-up questionnaire did not result in appreciable bias; bias was greatest in subgroups with small numbers. Conclusions These findings suggest that prospective analyses from this cohort are not substantially biased by non-response at the first follow-up assessment.

STUDY OBJECTIVE--The aim was to examine whether baseline characteristics from a cross sectional survey provided sufficient information regarding non-response bias in a follow up study when compared with information on hospital admissions in the intervening years. DESIGN--This was an 11 year follow up study of a cohort selected in 1974 with register information on hospital admissions during follow up. SETTING--The study was based on a sample of cement workers from a particular Portland cement factory with suitable controls from other occupations. PARTICIPANTS--A total of 1404 men participated in the first survey in 1974, including a questionnaire and lung function tests. In 1985 1070 men were alive and of these, 928 men (87%) responded to a postal questionnaire. MAIN RESULTS--Non-responders in 1985 did not differ markedly from responders when smoking habits, respiratory symptoms, and lung function were examined in 1974. During follow up, non-responders had twice as high rates of hospital admission due to respiratory diseases as responders. These differences remained present after adjusting for minor differences in age and smoking habits. CONCLUSIONS--Equal distributions of baseline characteristics among responders and non-responders in a follow up study do not preclude non-response bias.

Background Increases in reported leisure time physical activity (PA) and obesity have been observed in several countries. One hypothesis for these apparently contradictory trends is differential bias in estimates over time. The purpose of this short report is to examine the potential impact of changes in response rates over time on the prevalence of adequate PA in Canadian adults. Methods Participants were recruited in representative national telephone surveys of PA from 1995-2007. Differences in PA prevalence estimates between participants and those hard to reach were assessed using Student's t tests adjusted for multiple comparisons. Results The number of telephone calls required to reach and speak with someone in the household increased over time, as did the percentage of selected participants who initially refused during the first interview attempt. A higher prevalence of adequate PA was observed with 5-9 attempts to reach anyone in the household in 1999-2002, but this was not significant after adjustment for multiple comparisons. Conclusion No significant impact on PA trend estimates was observed due to differential non response rates. It is important for health policy makers to understand potential biases and how these may affect secular trends in all aspects of the energy balance equation.

Study objective Questions about burnout, career satisfaction, and longevity of emergency physicians have been raised but no studies have examined tolerance for uncertainty as a risk factor for burnout. Primary objectives of this study are to assess the role of uncertainty tolerance in predicting career burnout and to estimate the proportion of emergency physicians who exhibit high levels of career burnout. Methods A mail survey incorporating validated measures of career satisfaction, tolerance for uncertainty, and burnout was sent to a random sample of members of the American College of Emergency Physicians. Best- and worst-case scenarios of point estimates are provided to assess for the effect of nonresponse bias, and multivariable logistic regression was used to predict evidence of career burnout. Results One hundred ninety-three surveys were returned (response rate 43.1%). A high level of career burnout was exhibited in 62 (32.1%; best-worst case 13.8% to 64.1%) respondents. No demographic variables were associated with burnout status. The final model identified that high anxiety caused by concern for bad outcomes (odds ratio=6.35) was the strongest predictor of career burnout, controlling for all other variables. Conclusion A large percentage of emergency physicians in this study, 32.1%, exhibited emotional exhaustion, which is the core symptom of burnout. Emotional exhaustion was not related to age or type of practice and was not mitigated by training in emergency medicine. Physicians studied did not feel anxiety because of general uncertainty, difficulty in disclosing uncertainty to patients, or admitting errors to other physicians. High anxiety caused by concern for bad outcomes was the strongest predictor of burnout. Despite exhibiting emotional exhaustion, the majority of respondents are satisfied with the career of emergency medicine.

This report summarizes the Army Family Research Program (AFRP) sample design, the distribution of AFRP participants, and the scope of the survey population. It includes a discussion of what sampling weights are, why they are needed, and how they should be used. Calculations are provided for adjusting sampling weights to compensate for potential nonresponse bias.

In a recent survey of depressive symptoms among former prisoners of war, longitudinal data were used to estimate nonresponse bias. A predictive model was fitted to the data of current respondents and then was used to predict the scores of nonrespondents who had earlier provided similar convariate data. This analysis showed that, despite differences between respondents and nonrespondents in age, education, and severity of treatment during captivity, differences between the observed scores of respondents and the predicted scores of nonrespondents were small. For an estimation of the overall impact of nonresponse bias, revised estimates for the entire sample were calculated by combining observed data from respondents and predicted data for nonrespondents; the revised estimates differed little from those for respondents alone.

STUDY OBJECTIVE--The aim was to develop an alternative method of investigating non-response bias in postal surveys, including a method of calculating a final full (100%) coverage confidence interval which avoids the wide intervals of existing approaches. DESIGN AND SETTING--As part of a two stage survey of disablement in the community, a first phase postal questionnaire was sent to 25,168 households in Calderdale, West Yorkshire, England. Confidence intervals were calculated to investigate the precision of estimates using a "no bias" model, where the prevalence in non-responders is assumed to be the same as in responders. RESPONDENTS--A total of 21,889 postal questionnaires were returned (87%), representing households containing 42,826 people aged 16 years and over. This was achieved by the original post (1st wave, 57% response); two further postal follow ups (2nd and 3rd waves, taking the response to 73% and 81% respectively), the latter including a small personal call back; and a final postal follow up (the 4th wave). RESULTS--The cumulative estimated prevalence of those with dependence was plotted as the survey progressed. The final wave full coverage estimated prevalence for those aged 16-64 years was 12.8 per 1000 with 95% confidence intervals of 11.3-14.4 per 1000. The integrity of this estimate holds as long as the true prevalence in non-responders is within the calculated non-response confidence interval under the no bias assumption, 9.7-16.0 per 1000 people. This latter interval represents the tolerance of prevalence in non-responders implied by the no bias assumption. CONCLUSIONS--The findings have general implications for monitoring non-response bias in postal screening questionnaires. The confidence interval approach developed in this paper offers an alternative to existing regression based estimates, giving an indication of the range of prevalence amongst non-responders that could be tolerated before the no bias assumption used by the model is breached. It is suggested that this approach can be used to determine both the extent of bias, and to aid decision making about the appropriate juncture to terminate follow up. It highlights the potential, particularly in the context of a computerised survey operation, of methodological investigation occurring simultaneously with survey operation.

BackgroundThere is substantial interest in use of the Internet for surveys, but there have been few health-oriented, large, randomized trials of general population surveys on the Internet. It is unclear whether providing the option to respond via Internet increases the response rate, and to what degree the results will differ. ObjectiveThe aim of the study was to evaluate changes in response rate and outcomes in a postal respiratory health survey by adding an optional Web response alternative. MethodsThis was a randomized trial of a random sample of 4213 permanent residents of Norway, aged 20-40 years. Participants were randomized into a traditional survey arm, where they were asked to return the survey by mail, and an arm where they were also offered the option to respond via a Web form. ResultsA total of 1928/4213 subjects responded, a response rate of 45.8% across both arms. The total response rate was 44.8% (944/2105) in the postal plus optional Internet response arm and 46.7% (984/2108) in the usual postal survey arm, with no statistically significant difference between the randomized groups (P = .24). In the optional Internet arm, 8.3% (175/2105) of the sample responded using the Internet and 36.5% (769/2105) responded by post. Thus, Internet response was chosen by 18.5% (175/944) of those who replied in the optional Internet arm. In the multivariate analysis, Internet response was associated with being male, frequency and type of Internet access (home users more likely to respond by Internet than work users), and smoking habit, with current smokers being more likely to be Internet responders. 57% preferred postal response (1102/1928), 38% preferred Internet response (733/1928), and 3% preferred telephone interview (54/1928), with no difference between randomization arms (P = .56). But among those who indicated that they preferred the Internet response and who were randomized to the optional Internet arm, only 47% actually chose the Internet response. Asthma prevalence was higher among participants choosing the Internet response mode (16.7% vs 12.4%). ConclusionsWe failed to increase survey response rates by adding an optional Internet response. Asthma diagnosis was higher in the Internet response group, suggesting nonresponse bias. Method comparison studies should be carried out before Internet studies are accepted in new populations or new subject matters.

Purpose To examine response rate reporting in current epidemiology literature and nonresponse bias from a recent population-based case-control study on kidney cancer in Cook County, IL. Methods A literature review and summary of 62 population-based case-control studies on cancer from eight epidemiology and public health journals published during 2000–2004 was conducted to examine response rate reporting, trends in response rates and factors associated with nonresponse. Additionally, the association of nonresponse with age, gender, race (CMS and cases only) and community area was examined from a recently completed case-control study on kidney cancer in Cook County. Results The article review showed lower participation rates among controls from RDD samples, mail interview surveys and for studies on more common forms of cancer. Most articles did not present a final disposition of sample or a formula for calculating response rates. However, a study of the response rates reported in the 62 articles selected showed a decrease in response rates of 10%–15% since the late 1980's. Using data from the Kidney Cancer Study and modified AAPOR response rate formula number 5, defined as all completed interviews (I) divided by the sum of completed interviews (I), partially completed interviews (P), refusals (R), and noncontact (NC) response codes, the response rate was 72% for cases and 49% for controls. An analysis of nonresponse bias among cases showed that participation decreased significantly with age, with an OR of 0.97 for a one year increase (95% CI 0.98 to 0.99) and DMV control participation was higher for members of a community area that was more than 70% black, with an OR of 2.08 (95% CI 1.16 to 2.73). Conclusions Response rate formulas or calculations for population-based case-control studies are not being presented in current epidemiology literature. Furthermore, response rates appear to be declining despite an increase in response effort. Evidence of lower then expected response rates and nonresponse bias of age among cases and community area among controls is present in a recently conducted case-control study on kidney cancer in Cook County, IL.

A mail survey was administered to a panel of Colorado consumers to obtain estimates of willingness to pay (WTP) for preserving Union Park. Detailed socioeconomic data for all surveyed individuals was available a priori. We develop and employ the multiple-bounded probit with selection (MBPS) model, to correct for non-response and self-selection bias. Statistical tests indicate estimates of WTP corrected for these biases are significantly higher than uncorrected estimates. The use of selec- tion models requires an additional investment by researchers. These investments may be warranted for policy relevant studies. (JEL Q26)