The conventional randomized response design is unidimensional in the sense that it measures a single dimension of a sensitive attribute, like its prevalenc
17 January 2023 in Scientific Journal of Sport and Performance
The use of performance enhancing substances among elite athletes is a well-known phenomenon, but data on prevalence are inconsistent. The aim of this study was to investigate (1) the prevalence of doping use in Dutch elite athletes and (2) the attitude of elite athletes with regard to the use of performance enhancing substances. A total of 272 elite athletes completed an online questionnaire about doping use using the randomised response method and about their attitude towards the use of performance enhancing substances. The estimated prevalence of the use of doping among Dutch elite athletes during the last 12 months was 12.5 percent (95% CI 3.0 to 24.7%). The prevalence for using non-prohibited medication (without medical necessity) during the last 12 months was 15.4% (95% CI 7.1 to 23.7%). One in four athletes (23%) accepts the use of medication on prescription without a medical necessity, in order to enhance their performance. Doping prevalence among Dutch elite athletes is estimated at 12.5%, with a confidence interval from 3 to 25%. Acceptance of other performance enhancing substances is relatively high demonstrating that educational and preventive programs are essential in the fight against doping and the protection of athletes.
We consider the linkage of two or more registers in the situation where the registers do not cover the whole target population, and relevant categorical auxiliary variables (unique to one of the registers; although different variables could be present on each register) are available in addition to the usual matching variable(s). The linked registers therefore do not contain full information on either the observations (often individuals) or the variables. By treating this as a missing data problem it is possible to construct a linked data set, adjusted to estimate the part of the population missed by both registers, and containing completed covariate information for all the registers. This is achieved using an Expectation-Maximization (EM)-algorithm. We elucidate the properties of this approach where the model is appropriate and in situations corresponding with real applications in official statistics, and also where the model conditions are violated. The approach is applied to data on road accidents in the Netherlands, where the cause of the accident is denoted by the police and by the hospital. Here the cause of the accident denoted by the police is considered as missing information for the statistical units only registered by the hospital, and the other way around. The method needs to be widely applied to give a better impression of the range of problems where it can be beneficial.
12 February 2024 in Journal of the Royal Statistical Society Series A: Statistics in Society
The Extended Crosswise design is a randomized response design characterized by a sensitive and an innocuous question and two sub-samples with complementary randomization probabilities of the innocuous question. The response categories are ‘One’ with two different answers and ‘Two’ with two answers that are the same. Due to the complementary randomization probabilities, ‘One’ is the incriminating response in one sub-sample, and ‘Two’ in the other. The use of two sub-samples generates a degree of freedom to test for response biases with a goodness-of-fit test, but this test is unable to detect bias resulting from self-protective respondents giving the non-incriminating response when the incriminating response was required. This raises the question what a significant goodness-of-fit test measures? In this paper, we hypothesize that respondents are largely unaware which response is associated with the sensitive characteristic, and intuitively perceive ‘One’ as the safer response. We present empirical evidence for one-saying in six surveys among a total of elite athletes, and present estimates of doping use corrected for it. Furthermore, logistic regression analyses are conducted to test the hypothesis that respondents who complete the survey in a short time are more likely to answer randomly, and therefore are less likely to be one-sayers.