All time references are in CEST
Improvement of measurements in health research 1
|Session Organisers|| Mrs Sandra Jaworeck (TU Chemnitz)
Mr Philip Adebahr (TU Chemnitz)
Professor Peter Kriwy (TU Chemnitz)
|Time||Thursday 20 July, 14:00 - 15:30|
All scientific disciplines have an interest in improving both measurement instruments and measurement strategies. This is especially pertinent to empirical social science health research. It may be possible to extend measurement beyond the perception of individuals to a broader basis of content.
Results of quantitative empirical health research shape the everyday life of actors in the health care system. Quantitative empirical measures support indices of, for example, health, health care systems, and socioeconomic status. Knowledge of the background of measurements, assessment of measurement quality, and content validity should be critically examined as well as response behavior in terms of social desirability. Changes in response behavior are also evident with different survey modes. Which means that respondent behavior could make a decisive difference in results.
Health is neither a purely objective nor a strictly subjective concept. But in social science surveys, subjective information about health is usually the central element of health measurements. What thresholds, if any, must be exceeded for individuals to respond to stimuli? Once stimuli have been perceived, they are evaluated by the observer (also unconsciously) and classified into contexts as well as learned structures. These processes are shaped by the functioning of the human organs and, above all, by previous individual experiences.
Measurements are often made at the objective level, as in the case of socioeconomic status, which is determined from objective indicators (education, occupation, income). In addition, subjective social status (SSS) has increasingly been measured in recent years. The subjective level is present in pretty much all areas of health research.
This session deals with improvements of objective and subjective health measurements that will enhance and simplify the field in the future.
Keywords: measurements, methods, health research
Dr Cornelia Neuert (GESIS - Leibniz Institute for the Social Sciences) - Presenting Author
Dr Dorothée Behr (GESIS - Leibniz Institute for the Social Sciences)
Dr Katharina Meitinger (Utrecht University)
Self-rated health is one of the most frequently used health measures. The item is asked in many (cross-)national surveys, however, often in (slightly) different versions or translations. The versions vary in the wording of the item text but also in terms of the set of response options used (“excellent” to “poor”; “very good” to “very bad”) and whether they represent a balanced set of positive and negative response options or not. Selection and labelling of response options might have an impact on respondents’ understanding, and thus, their assessment of health.
In this study, we use qualitative and quantitative methods to examine how respondent behaviour differs regarding response scale labelling, how respondents understand the varying labels, and what are the reasons that drive response behaviour of the self-rated health question. To answer our research questions, we use data from the GESIS panel, which is a probability-based mixed-mode access panel (GESIS, 2022) and from a non-probabilistic online access panel. We varied the wording of the scale labels of the self-rated health item leading to four experimental conditions. We compare the distributions of the different versions and which health factors respondents consider when answering questions about their health.
Ms Camilla Salvatore (Utrecht University)
Professor Joseph Sakshaug (Institute for Employment Research) - Presenting Author
Professor Silvia Biffignandi (University of Bergamo)
Professor Arkadiusz Wisniowski (University of Manchester)
Professor Bella Struminskaya (Utrecht University)
Objective health measurements are often considered the gold standard in health survey research. However, collecting objective health measurements in surveys can be expensive and challenging to administer to large samples. In contrast, subjective (self-reported) health measurements are simpler and less expensive to collect but they are prone to misreporting. In this paper, we propose a Bayesian methodology for combining both types of measurements that overcomes their respective weaknesses and potentially reduces survey costs. We evaluate the method through a simulation study and a real-world application involving regression analyses of common health outcomes. We show that the proposed method reduces the mean-squared error of regression coefficients and reduces survey costs under certain conditions. Limitations and extensions of the method are also discussed.
Dr Talip Kilic (World Bank)
Dr Alemayehu Ambel (World Bank)
Mr Philip Wollburg (World Bank) - Presenting Author
Mr Yannick Markhof (World Bank)
Dr Julia Dayton (World Bank)
Dr Patrick Hoang-Vu Eozenou (World Bank)
Dr Jakub Jan Kakietek (World Bank)
Mr Nicholas Stacey (Independent Consultant)
Initially motivated by the COVID-19 pandemic and starting in April 2020, high-frequency phone surveys have been rolled out rapidly across low- and middle-income countries with otherwise limited experience with phone-based data collection. In view of the slow-down in face-to-face survey data collection, phone surveys, particularly those that have leveraged pre-COVID-19 household surveys as sampling frames, have fulfilled important gaps in evidence and knowledge regarding the impacts of and responses to the pandemic, also as they relate to the health sector. Now in the third year of the pandemic, phone surveys continue to respond to evolving data needs not only regarding theCOVID-19 pandemic but also emerging, large-scale, covariant health and economic shocks. Starting in 2022, thanks to the partnership between the Living Standards Measurement Study and the Global Financing Facility, national phone survey systems across several Sub-Saharan African countries have started eliciting high-frequency information regarding health service needs, foregone care, and out of pocket health expenditures, among other socio-economic data. This paper reports on a series of randomized survey experiments that were piggybacked onto the longitudinal phone survey platforms in Burkina Faso, Ethiopia, Malawi and Uganda and that attempted to gauge the impact of (household- versus individual-level) questionnaire design and (purposive versus random) respondent selection on nationally-representative survey data collection on health, for the purpose of informing the design and implementation of downstream phone surveys that are implemented by national statistical offices in low- and middle-income countries.
Mrs Jacqueline Kroh (Leibniz Institute for Educational Trajectories) - Presenting Author
Mrs Julia Tuppat (University of Leipzig)
Mrs Raffaela Gentile (Leibniz Institute for Educational Trajectories)
Mrs Hanna Reichelt (Leibniz Institute for Educational Trajectories)
In large-scale surveys, self-rated health (SRH) based on questions such as "In general, how would you rate your health?" is a widely used measurement to assess individual's health status. While a large number of studies have investigated the health aspects and assessment strategies of adult respondents when answering this question, it is largely unknown how children assess their general health. In an accompanying sub-study to the National education Panel Study (NEPS), we examine how children rate their health according to this question in a sample of 9- to 12-year-olds. By using techniques of cognitive interviewing and content analysis, we investigate the underlying health dimensions, the specific health factors as well as different assessment strategies that children use. We relate these to the actual outcome of SRH, and explore potential group differences by gender, age, migration background and parental education.
Our results indicate that children in this age group mostly refer to the dimensions of physical health and daily functioning. They take into account a wide range of health aspects, such as temporary illnesses, chronic illnesses, pain and symptoms, injuries and affect, with only slight differences between subgroups. They employ different assessment strategies, relating to severity as well as frequency of health problems. Finally, the majority of children assess their health using one particular health dimension only, but reflect on different health factors and combine different assessment strategies. Thus, we identify a certain degree of multidimensionality of the self-assessment as intended with a global measure such as SRH, even though to a lesser degree than in adult respondents, as previous research demonstrates. Furthermore, children are less likely to consider the mental health dimension. We discuss the implications, especially in light of their relevance for longitudinal studies and comparisons between individuals at different life stages.
Mr Patrick Lazarevič (Vienna Institute of Demography) - Presenting Author
Background: Health is a fundamental aspect of many scientific disciplines and its definition and measurement is the analytical core of many empirical studies. Comprehensive scales of health or objective measures, however, are typically precluded in survey research due to financial and temporal restrictions. Self-rated health (SRH) as a single indicator of health, on the other hand, has been shown to exhibit a lack of measurement invariance by age and being biased due to non-health influences on reporting behavior. In the three-item Minimum European Health Module (MEHM), SRH is complemented with questions on chronic health conditions and activity limitations, thus providing a compromise between single indicators and comprehensive measures.
Data & Methods: Using data from the German Ageing Survey (waves 2008 & 2014; n = 12,037), we investigated the feasibility to combine the MEHM into a generic health indicator and judged its utility in comparison to SRH as a benchmark. Additionally, we explored the option of an extended version of the MEHM by adding information on multimorbidity and the presence and intensity of chronic pain.
Results: Analyses showed that both versions of the MEHM had a good internal consistency and each represented a single latent variable that can be computed using generalized structural equation modeling. The resulting indicator was less affected by biases due to age, education, and optimism while being highly correlated to more comprehensive health measures.
Conclusions: The utility of this approach showed great promise as it significantly reduced age-specific reporting behavior and some non-health biases present in SRH, promising interesting applications for healthy life expectancy estimation. To further attenuate systematic response behavior, this approach could be extended by priming the meaning of health in SRH by changing the question order of the MEHM and the use of MG-MIMIC-modeling.