Dawid Gondek PhD, Senior Researcher at FORS

Title: Estimating the causal effects of the Covid-19 pandemic on mental health – a futile task ?

Abstract
There has been an abundance of studies aiming to examine the impact of the Covid-19 pandemic on mental health. Investigating an “impact” of a phenomenon implies that the study objectives are causal. Nonetheless, the vast majority of evidence on the Covid-19 pandemic and mental health has been descriptive. Despite its importance, descriptive studies are often undervalued in social sciences. Is estimating the causal effect of the pandemic feasible? If so, is it even useful?

In this talk, Dawid Gondek discussed the assumptions required to estimate average causal effects of the Covid-19 pandemic on mental health. Researchers typically focus on exchangeability, which refers to a lack of confounding. However, other assumptions in causal inference – positivity and consistency – are often overlooked. Consistency assumption requires that the exposure is defined unambiguously so that it can be acted on by a potential well-defined intervention. Hence, the key question becomes – how do we define the Covid-19 pandemic? Dawid Gondek shared some suggestions going forward based on the work conducted within the CovidGen project.

Robin Benz and Tobias Ackermann (ICER, University of Bern)

Disadvantaged by Chance? Cut-off Dates for School Enrolment and Their Consequences for Educational Outcomes

Abstract:
Pupils who did not start learning at the same level as their peers might subsequently fall behind throughout their educational careers (e.g., Heckman 2006; Passaretta et al. 2022). The modalities of compulsory school admission may contribute to the emergence of early gaps in educational performance. Nearly all education systems have arbitrarily chosen cut-off dates for school enrolment, which create age differences of up to a year within a cohort of pupils. Prior research has shown that the youngest pupils in a cohort fall behind their relatively older peers in educational performance (e.g., Bedard and Dhuey 2006; Peña 2017; Dicks and Lancee 2018). These performance gaps are coined as relative age effects, which can be framed within theories of cumulative (dis)advantages (e.g., DiPrete and Eirich 2006).

Drawing on a comprehensive data set encompassing the entire student population in North Western Switzerland (BR NWCH 2021), the study addresses three research questions. First, it is investigated to what extent pupils’ relative age affects their educational achievement in different subjects and track placement in secondary education. Second, by exploiting the longitudinal structure of the data, it is examined whether the influence of relative age diminishes the course of educational trajectories. Third, the study establishes a record linkage between administrative data and pupils’ test data to investigate whether pupils from disadvantaged backgrounds suffer more strongly from relative age effects.

A pupil’s relative age might be correlated with various unobserved factors. Two strategies are employed to address these endogeneity concerns. First, the study employs an instrumental variable approach using “assigned relative age” (e.g., Bedard and Dhuey 2006) as an instrument for pupils’ actual age. Second, the study uses a regression discontinuity design contrasting pupils born just before and after the cut-off date to estimate relative age effects in Switzerland.

Preliminary results provide evidence that students with a relative age advantage when they entered school achieve significantly higher than their counterparts with a relative age disadvantage during their first years of primary education. However, relative age effects vanish the more students advance in their educational trajectory. Additional analyses shed light on potential effect heterogeneity.

The study illustrates how early disadvantages emerge by chance through arbitrarily chosen cut-off dates for school eligibility. Scholars and policy-makers alike are urged to debate how the modalities of school entry can be designed to ensure equal starting conditions for all.

Katarzyna Wac (University of Geneva)

Title: mQoL Living Lab: Quantifying Quality of Life Beyond Self-Reporting and Incorporating Daily Life into Medicine

Abstract:
Current Quality of Life assessment methods and tools are mostly based on self-reports, which are infrequent, subjective, memory-based, and context-poor. Personal technologies can enable quality-of-life quantification. But how? The Quality of Life Technologies (QoL Lab) lab has been established in 2010 and since then is researching how mobile and emerging sensor-based technologies can be leveraged for an accurate, longitudinal personalized assessment of the individual’s behavior and life quality, as they unfold naturally over time and in context, and the improvement of the latter. The QoL lab research draws on new emerging models from computer science incorporating examination, diagnosis, and treatment of daily life as an “organ” – much like a cardiologist examines heart. The mQoL Living Lab is a software infrastructure designed and maintained by the QoL Lab that leverages personal, longitudinal, real-world data approaches acquired via mixed-methods along human subject studies spanning observational to interventional designs (including RCT and N-of-1 designs). More at qol.unige.ch.

Dawid Gondek (FORS) presented the following project:

Title: Deriving a measure of wellbeing in the Swiss Household Panel and examining its psychometric properties

Abstract: The Swiss Household Panel is a large, nationally representative survey of households in Switzerland. It has been an invaluable source of knowledge about wellbeing at the population level, often serving for cross-countries comparisons. However, the researchers have been highly inconsistent in their use of wellbeing indicators, making comparability of findings across studies difficult. Hence, we aimed to derive a measure of wellbeing and examine its psychometric properties.

We assessed the factorial structure and internal reliability of identified wellbeing indicators and tested their measurement invariance across age groups, periods, genders, languages of questionnaire completion and modes of collection.

We demonstrated that psychometrically robust wellbeing measure can be derived using individual items of the Swiss Household Panel. The measure comprises two subscales: 1) positive affect and life satisfaction, 2) negative affect. An overall score of wellbeing, despite having a satisfactory internal reliability, should be used with caution, as our findings suggests that it is not a unidimensional construct when operationalised in the Swiss Household Panel. Some caution, however, needs to be taken when an overall score of wellbeing is used or the research question is about comparing age or lingual groups.

Our research has potential to save researchers valuable time and inform their decisions about using the measures of wellbeing. Furthermore, using a consistent measure of wellbeing can make research using the SHP more comparable, which will facilitate interpreting effect sizes.

Paulina Pankowska (Assistant Professor at the Sociology department of Utrecht University and currently visiting researcher at LIVES) presented the following:

Accelerating progress in the social sciences: the potential of benchmarks

Abstract: Social scientists aim to create explanations of the world. For each social phenomena, scientists have proposed a myriad of theories to explain its working mechanisms. Traditionally, these theories are tested by translating them into statistical models and assessing the significance of the model coefficients. This approach however is not free of shortcomings. Most importantly, it can result in the specification of a variety of models that represent competing theories, are based on different statistical techniques, or include different predictors of the social phenomena studied. While equally plausible and well-justified, these models often provide contradictory results and lead to inconsistent findings. As things currently stand, there is no framework that allows for the comparison of these models and the question of which model works better under which circumstances remains unanswered. As a result, it is difficult to evaluate conflicting theories, and monitor progress in the social sciences.

We argue that benchmarks can be used as such a standard frame of reference and accelerate progress in the field of social sciences. They have large potential for answering long standing questions in the field and can drive the field forward. We define a benchmark as a standardized validation framework that allows for the direct comparison of the prediction accuracy of various models which address the same research problem. The use of benchmarks has led to progress and breakthroughs in many fields of science including computer and data science, physics, biomedicine, and the humanities.

The slides of the presentation

It was a great pleasure to invite the authors of the following FORS GUIDES to present their latest guides:

Data Citation: How and Why Citing (Your Own) Data (FORS Guide N°19), Christina Bornatici, Nicolas Fedrigo

Qualitative data anonymisation: theoretical and practical considerations for anonymising interview transcripts (FORS Guide N°20), Alexandra Stam, Pablo Diaz

Data Sharing in the Social Sciences (FORS Guide N°21), Marieke Heers

Measuring psychological constructs (FORS Guide N°22), Valérie-Anne Ryser

All the FORS Guides publications can be viewed here: https://forscenter.ch/publications/fors-guides/.

Olga Gorodetskaya (PhD candidate – Sociology and Social Research, University of Trento and currently visiting researcher at LIVES) presented the following project:

The best time to become a parent: The effect of tenure on fertility among academics in Italy

Joint work with

Valentina Tocchioni, University of Florence, Agnese Vitali, University of Trento and Alessandra Minello, University of Padua

 Abstract

Previous literature convincingly showed that women who work in Academia tend to have slower careers compared to their male peers and are less likely to reach top academic positions. Childbearing has been indicated as responsible for a considerable part of the existing gender gap in Academia. While the association between fertility and career progression is well known, little do we know about the reverse link: does career progression foster fertility among academic women? And what about men? The aim of this contribution is to verify whether and how academic position and academic promotion has an impact on three fertility-related outcomes: (1) the transition to parenthood, (2) the transition to the second child, and (3) short-term fertility intentions. We use a unique source of primary-collected survey data on a large sample of Italian female and male academics, including retrospective information on fertility histories and job histories, and their fertility intentions. All stages of career progression of PhD graduates are considered, from postdoctoral researcher up to full professor. We apply discrete time event-history models on the transition to the first- and second-child birth and logistic regression for fertility intentions, separately for men and women. Our preliminary results show that both male and female academics wait until obtaining their first tenure-track position to become a first time-parent; however, the ‘fertility window’ for men is larger than for women. Conversely, no significant difference among academic ranks as well as genders emerges with regards to the transition to a second child, where other factors may play a role.

Maud Reveilhac (Post-doctoral student at the Department of Communication and Media Research, Zurich University) presented the following project:

Integrating surveys and social media to better understand the dynamics of public opinion

Abstract

Our study presents results around the influence of social media on survey results, especially by looking at the variations of topic and tweeter have on that influence. To do so, we propose two study cases of Swiss politics about social media effect on opinion change during election and about the similarity of arguments in direct democracy votes. We rely on a two-year longitudinal data collection of tweets emitted by more than 50’000 identified Swiss seed users and their most active followers. In a first step, the article investigates the extent to which the topics most important to the public correlate with communication available on Twitter for different users groups (notably, politicians, media actors and users interested in politics). We also benchmark the results against external data about mass media content and a survey of political candidates. In a second step, and based on open-ended responses from representative survey conducted during direct democracy campaigns, we assess the congruence between supportive and opposing voting arguments from a representative sample of citizens and communication from several Twitter groups (notably, politicians, associations, activists, etc). These results have the potential to inform researchers of the circumstances in which social media data are most likely to be available and a useful complement to surveys for understanding the dynamics of public opinion.

During the seminar the winners of the FORS Data Re-use Award 2022 Rita Schmutz (first prize) and Mengling Cheng (second prize) presented their work and received their awards.

Rita Schmutz (LIVES, University of Lausanne) presented the following project:

Inequality of Educational Opportunity in Switzerland: Exploring Regional Differences and Institutional Factors

Abstract

This study provides estimates of inequality of educational opportunity (IEOp) in compulsory education in Switzerland using the 2016 Swiss large-scale assessments of basic competencies (ÜGK – COFO – VeCoF) dataset. IEOp measures the share of inequality in the distribution of educational performance that can only be attributed to students’ characteristics beyond their control (circumstances), such as their family socioeconomic status (SES), parental education, gender, and immigrant status. By employing a parametric ex-ante estimation approach, I estimate that IEOp in Switzerland is 21%, with considerable heterogeneity across cantons and the German-speaking region showing higher levels of unfair educational inequalities. The IEOp estimates are decomposed to identify the contributions of different circumstances and to develop a better understanding of the observed inequality. Socioeconomic status, parental education, and occupation are the most relevant circumstances in most cantons. Inequality of educational opportunity is associated with the institutional structure of cantonal education systems. Educational inputs in primary school explained most of the cross-canton variation in IEOp, followed by the level of stratification in secondary school.

Mengling Cheng (LIVES, University of Lausanne) presented the following project:

Evolution of the income-related gap in health with old age: Evidence from 20 countries in European and Chinese panel datasets

(joint work with Nicolas Sommet, Daniela Jopp and Dario Spini)

Abstract

Background: Some studies show that the protective effect of higher income on health weakens with old age (the age-as-leveler pattern), whereas others show that it strengthens with old age (the cumulative advantage/disadvantage pattern). Existing studies are limited in that they use single-country and/or single-timepoint designs. To overcome these limitations, we used cross-national and longitudinal data to clarify how the income-related gap in health evolves with old age.

Methods: We used the longest-running European and Chinese panel datasets, namely, SHARE (2004-2019, 73,407 European participants from 19 countries) and CHARLS (2011-2018, 10,067 Chinese participants). We operationalised health using multimorbidity and three alternative indicators (functional disability, mobility disability, and memory). We performed Poisson growth curve modeling to capture the between-participant effects of age and the within-participant effects of aging.

Results: We obtained three consistent findings for both Europe (the effect was observed in most countries) and China. First, the protective effect of higher income on multimorbidity, functional disability, and mobility disability was weaker for older than for younger adults (between-participant age-as-leveler effects). Second, only the protective effect of higher income on mobility disability weakened over the later life course (within-participant age-as-leveler effects). Third, the protective effect of higher income on memory was stronger for older than for younger adults and strengthened over the later life course of the individual (both between-participant and within-participant cumulative advantage/disadvantage effects).

Conclusion: Our results suggest that the income-related gap in physical health (i.e., multimorbidity, functional disability, and mobility disability)—but not in cognitive health (i.e., memory)—narrows in old age for both Europe and China.

Jonathan Jubin^a, Emilie Antille^b, Vladimir Jolidon^b, Léonard Roth^b and Isabelle Peytremann Bridevaux^b presented the following project:

^a La Source School of Nursing, HES-SO University of Applied Sciences and Arts Western Switzerland, 1004 Lausanne

^b Epidemiology and Health Systems, Center for Primary Care and Public Health, 1010 Lausanne

The Swiss COhort of Healthcare Professionals and Informal CAregivers (SCOHPICA): Career trajectories, intent to stay/leave the profession, wellbeing, and their determinants

 Abstract 

To address the shortage of qualified personnel and prepare for the future, the Swiss healthcare system needs reliable data on its professionals, their employment trajectory and the reasons why they leave or stay in their profession. However, such data is scarce in Switzerland. Therefore, the SCOHPICA open cohort has started surveying in autumn 2022 professionals working with patients from all sectors of healthcare regardless of their status and practice settings. The longitudinal study design is mixed, consisting of a yearly online questionnaire and biennial focus groups. Additionally, SCOHPICA aims to launch in 2024 a cohort of informal caregivers, who are key yet often forgotten actors of the health system. Raw data and results will be made available to decision-makers and researchers through a repository and an interactive online platform.

Maud Wieczorek (Centre LIVES, Swiss National Centre of Competence in Research LIVES – Overcoming vulnerability: Life course perspectives, Lausanne and Geneva, Switzerland) will present the following project:

Association between multiple chronic conditions and insufficient health literacy: cross-sectional evidence from a population-based sample of older adults living in Switzerland

Abstract 

Health literacy is the ability to find, understand, assess, and apply health information. Individuals suffering from multiple chronic conditions have complex healthcare needs that may challenge their health literacy skills. This study aimed to investigate the relationship between multimorbidity, the number of chronic conditions, and health literacy levels in a sample of adults aged 58+ in Switzerland.

We used data from 1,615 respondents to a paper-and-pencil questionnaire administered as part of wave 8 (2019/2020) of the Survey of Health, Ageing and Retirement in Europe (SHARE) in Switzerland. Health literacy was measured using the short version of the European Health Literacy Survey questionnaire. The final score ranged from 0 to 16 and was categorised into three health literacy levels: inadequate (0–8), problematic (9–12), and sufficient (13–16). The number of chronic conditions was self-reported based on a pre-defined list. Associations were examined using multivariable ordinary least squares and ordered probit regression models, controlling for key socio-demographic characteristics.

Overall, 63.5% of respondents reported having at least one chronic condition. Respondents who reported one, two, and three or more chronic conditions were more likely to have lower health literacy scores compared to respondents who did not report any chronic condition (p<0.05, p<0.01, and p<0.001, respectively). Suffering from two and three or more chronic conditions (vs. no chronic condition) was significantly associated with a higher likelihood of having inadequate or problematic health literacy levels (both p-values <0.01).

Our findings suggest a need to improve health literacy in older adults suffering from chronic conditions. Improved health literacy could constitute a promising lever to empower individuals to better self-manage their health to ultimately reduce the double burden of chronic diseases and insufficient health literacy in this vulnerable population.

Presenters: Prof. Dr. Laura Bernardi, Prof. Dr. Moritz Daum, Dr. Doris Hanappi, Dr. Martin Kindschi, Prof. Dr. Michael Shanahan

Abstract
Rising social inequality and diminishing returns to education make early investment in children’s different living environments (contexte de vie) – their learning, leisure, and school – increasingly important for their development and well-being. BUNAVIA, a project funded by the Jacobs Center for Productive Youth Development (UZH), and conducted in collaboration with University of Lausanne and the Department of Education Planning of the Canton of Zurich, investigates the impact of context on child development by studying social networks and the development of identity. It pays particular attention to the perception of autonomy, self-efficacy, and self-regulation during transitional phases, for example, starting school, changes within and between schools. Since Spring 2022, BUNAVIA is part of the Zurich Learning Progress Study (Zürcher Lernverlaufserhebung). Data will be collected via web-based surveys from parents and in-class assessments from the focal child (age 4-5 entering premier cycle primaire/kindergarten,) in the Canton Zurich in Fall 2023. As BUNAVIA plans to propose a parallel study in the canton Vaud, the presentation will give an overview of the project and a summary of its main concepts, research priorities, and data collection design.

Robert Reinecke¹, Sarah Vilpert^1,2, Gian Domenico Borasio³, Jürgen Maurer²

¹Swiss Centre of Expertise in the Social Sciences (FORS), University of Lausanne, Switzerland
²Faculty of Business and Economics (HEC), University of Lausanne, Switzerland

³Palliative and Supportive Care Service, Geriatric Palliative Care, Lausanne University Hospital and University of Lausanne, Switzerland

Abstract 

Individuals tend to wait until the last moment to plan their end-of-life (EOL) care. Yet decision-making abilities and understanding of medical situations decrease with age, which could compromise the writing of advance directives (ADs). Little is known concerning the association between cognitive competence and attitudes/behaviors towards end-of-life (EOL) care planning in older adults. The present study aims to better understand the association of cognitive competence – measured via verbal fluency, immediate and delayed memory, basic calculation skills, and temporal orientation – with attitudes and behaviors towards EOL care planning in a nationally representative sample of older adults in Switzerland. We analyzed data of 1’936 respondents aged 55+ from a paper-and-pencil self-completion questionnaire that was administered as part of wave 6 (2015) of the Survey of Health, Ageing and Retirement in Europe (SHARE) in Switzerland using logistic regression models. Respondents with reduced global cognitive competence are less likely (i) to have discussed their EOL preferences with others and (ii) to have a living will. Impairment in verbal fluency is negatively associated to attitudes and behaviors towards EOL care planning. Our findings highlight that intact cognitive abilities have a positive impact on EOL care planning.

Abstract

Both inside and outside of academia, there have been concerns regarding individualization; people would become more focused on their own lives and would be less willing to engage in solidarity behaviors, such as volunteering time to help each other. My PhD project focuses on a specific solidarity behavior, namely informal helping. This refers to helping people that do not live in the same household (e.g., friends, relatives and neighbors) without the coordination of formal organizations (Einolf, Prouteau, Nezhina& Ibrayeva, 2016). The project specifically examines additional risk (divorce, COVID-19, inability to reciprocate help) and protective (socialization by parents and partner) factors against this presumed trend of individualization. In the presentation, I will zoom in two studies within my PhD project. The first concerns the impact of the first lockdown of the COVID-19 pandemic in the Netherlands. Based on panel data from before and after the first lockdown, we found that informal helping declined during the first lockdown but not for everyone to the same extent. The second study concerns the impact of reciprocity norms in informal helping, which we investigated with a factorial survey. Results show that people are more willing to help neighbors who are likely to reciprocate their help, have helped in the past or clearly need help. Finally, I will shortly discuss our plans for the paper I will work on during my time at FORS. This paper examines the impact that partners have on each other’s informal helping and which partner is the more influential in this respect.

What are the intergenerational consequences of increasing education? This study estimates the effects of women’s education on their offspring using quasi-experimental evidence from six educational reforms that increased the length of compulsory schooling in Austria, the Czech Republic, Denmark, France, Italy, and the Netherlands. The empirical analysis uses data from the Survey of Health, Ageing and Retirement in Europe (SHARE) and instrumental variable estimation to estimate the effects of female education on fertility and on children’s education. This study provides the first analysis using quasi-experimental variation in education to estimate prospective models of intergenerational effects. These models start with a birth cohort and link information on their fertility and on their children’s outcomes to this cohort. By doing so, these models can account for the effect of female education on the probability that women have children when estimating the effect of female education. The findings show that the direct effect of female education on children’s educational attainment, i.e., the effect conditional on the birth of a child, is positive. In addition, higher female education increases fertility. Therefore, the probability that a woman has a child with a high educational attainment is increased when taking into account the effect of female education on fertility. I conclude that studies that estimate retrospective models of intergenerational effects using reforms in the length of compulsory schooling have underestimated the total effect of female education on child education by not taking into account the impact of female education on fertility.

Abstract

In the last few years, the promotion of advance care planning has become a public health priority in Switzerland. In this context, people are invited to reflect on their wishes concerning the end of life, talk about them to their relatives, and document them in order to optimize therapeutic choices in case they are not able anymore to make decisions for themselves following an accident, a stroke or a heart attack, for example.

However, advance care planning assumes that preferences for end-of-life care remain stable over time. Although the stability of these preferences has been explored in seriously or terminally ill patients, so far, few studies have focused on younger people who should now be involved in this planning process.

The purpose of our study is to close this gap by examining the stability of a wide range of end-of-life preferences in the general population and to identify characteristics that are related to their stability or change. Our analyses are based on data collected among Swiss SHARE participants who completed the two paper-and-pencil questionnaires on end-of-life preferences in waves 6 (2015) and 8 (2019-2020). In this presentation, we will examine whether people who completed advance directives had more stable end-of-life preferences than people who did not.

Recently more and more academic journals are implementing open data policies, posing the question for researchers of where, when, and how to share data as well as other research-related materials. As journals and funders demand the sharing of data used in publications, the way social science research is conducted and disseminated is undergoing a progressive change and the available infrastructure is being adapted. A crucial element of this change is the transparency of research. Related to this, replicability and reproducibility are becoming increasingly important in the social sciences. This includes making research data, analysis code, and study materials freely available.

The first aim of this lunch seminar is to give an overview of the challenges of replication and its contribution to scientific transparency. This second is to present SUBreplica, FORS’ new online replication tool and service dedicated to the sharing of replication materials.