The ESA RN21 2022 mid-term conference “Quantitative methods and the COVID19-pandemic” will be held at University of Salamanca, Spain from 5th to 7th October 2022. The meetings and conferences will take place at the Hospedería Fonseca (Calle Fonseca 2, 37007-Salamanca, Spain).
The registration process will begin at 18:00, October the 5th at the venue.
Words of welcome.
Unlike what is usual in other sociological investigations, at the first months of the pandemic I wasn't able to design or accede to an unpublished source of information. The main characteristic of the object of study was the instability and speed of change. I had never carried out a research project in such precarious conditions, isolated in my home, surrounded by uncertainty due to the threat of the virus, looking for information both for Spain and internationally and with no other connection than the virtual one. But there is no worse research than the one that is not done, and not even pandemics or the risk of a computer or server failure completely cut off the will to understand and share. The alternative of giving up investigating would have meant giving up out of hand. So, with all the foreseen limitations and with many hours of dedication, these words have finally been born.
Due to contact restrictions in the wake of the COVID-19 pandemic, surveys often could not be conducted in originally planned face-to-face mode. Consequently, many surveys switched to online modes or used different mixed-mode designs, for example to a combination of CATI (telephone interviews) and CAPI (personal interviews). The presenters also had to pursue a combination of CATI and CAPI for the survey of the ISSP-Module on Environment, which is a cross-cultural comparative survey that is representative for the Austrian population and, in the past, had always been conducted in a face-to-face mode.
Mixed-mode surveys facilitate field access in pandemic times and show great potential to improve sampling processes due to reduced non-response- and coverage errors. At the same time, however, the combination of different survey modes comes along with a series of risks such as mode-effects causing bias due measurement effects. In recent years, numerous studies have increasingly focused on the analysis of response rates and sample composition using mixed modes as well as on mode effects with regard to response patterns and the question of the accumulation of measurement errors. So far, results show a mixed picture concerning the causes and consequences of mode effects.
Against this backdrop, we analyze differences in the factorial structure and response distributions of two central constructs within the Austrian ISSP Environment module 2020/2021 (N=1.261) using Bayesian multigroup confirmatory factor analysis and linear regression. These constructs consider survey questions on institutional trust and the willingness to sacrifice for environmental protection. The findings show support for scalar invariance and therefore the absence of CAPI vs. CAPTI mode-effects on the factorial structure for both constructs. Differences in response distributions seem attributable to differences in sample composition. Based on the results, we finally highlight several implications for the interpretation of mixed-mode effects.
Lee’s Coronavirus Anxiety Scale (CAS) is a popular mental health questionnaire emerged during the pandemic. It includes five items describing mental states attributed to the fear of COVID, and participants report how often they experienced each on a five-point scale. We tested cross-national measurement invariance of CAS using convenience samples from 60 countries and examined the convergent validity of CAS scores in relation to fear of COVID and life satisfaction. We used both conventional exact invariance tests, as well as a more recent alignment procedure. Since the response option “Never” was chosen disproportionally more frequently than all the others, the responses were dichotomized.
Results demonstrated that single-factor model fit the data sufficiently well in almost all the countries. The exact partial scalar invariance model on a reduced set of 56 countries was supported, however some parameters had unrealistic values. Due to the large differences in sample sizes across countries the results might have been biased toward the larger samples. To test their robustness, we employed the resampling techniques with and without replacement. The results were more stable with larger samples. Two versions of the alignment procedure demonstrated a high degree of measurement invariance – only 9% of the parameters were non-invariant. We also ran simulations of alignment with parameters estimated in the current model which demonstrated the reliability of the means, but problematic estimation of the latent variances. Overall, different invariance testing methods demonstrated substantial disagreement.
The correlations between CAS and Fear of COVID Scale estimated with three different approaches were significant and positive in all countries but two, supporting the convergent validity of CAS. Correlations of CAS and Satisfaction With Life Scale were much weaker ranging between -.10 and -.13 and significantly differed from zero only in a small fraction of countries. It marginally supported our expectations.
The COVID-19 pandemic is a substantial challenge for established social survey programs, relying on high quality face-to-face data collection. Secure circumstances for personal interviews became uncertain with lockdowns and social distancing in place. Hence, survey researchers had to alternate data collection modes. While planned for 2020, the sixth edition of Social Survey Austria was conducted under such circumstances in mid-2021, after vaccinations were rolled out in Austria and the most severe restrictions were lifted, leaving only basic distance and mask rules in place. The sample is selected from a household-based sampling frame with telephone (approx. n=900) and face-to-face interviews (approx. n=350). A response rate of 52% and matching criteria for representation on age, sex, education, and regional distribution suggest high data quality.
In our study, we employed this quasi-experimental setting to analyze if adherence to distance and masking rules during the interview, fear of COVID-19 and classical interviewer effects influenced responses. Using the telephone interviews as baseline, we use the face-to-face interviews to control for interviewer effects and elaborate on COVID-19 related factors as potential bias for survey response.
Preliminary results suggest a complex situation. Health related COVID-19 fears demonstrate fewer significant effects than economic COVID-19 fears on latent constructs like institutional trust, immigration attitudes or environmental behavior. In addition, interviewer as well as situational effects emerge in some models adding complexity. Despite the heterogeneity of effects, we detect a bias across most models in the responses comparing telephone and face-to-face interviews.
Our results propose an interesting puzzle for all datasets collected in face-to-face mode during the pandemic: Shall we interpret the bias as a crisis effect influencing responses in other topical areas or are those attitudes and behaviors only another manifestation of underlying values and attitudes and the crisis affects these underlying concepts?
The spreading of diseases is driven by the topology of the underlying network, which is formed by interactions [1, 4, 5]. A current example is the ongoing pandemic of SARS-CoV2 in the world-wide social network, consisting of humans and their contacts with each other. The number and diversity of contacts a person has has a direct influence on the spread of viruses [2, 3]. Mathematical models can model the spread to increase understanding and allow the prediction of outbreaks or even epidemics. Informative data about parameters is required for the purpose of model specification. A common obstacle in most work is missing data in age and/or location groups. Mostly, surveys were only approved for adults and social contact data for children were only available, if a parent participated in the survey and provided the information. One possible reason for missing data in locations (e.g. provinces or other administrative divisions) are online surveys without sampling design that yield a convenience sample. The aim of the proposed paper is to investigate applied approaches to handle missing data in age-groups and the effect on the estimated age-stratified contact matrices. A special focus lies on the effect on those results of enforcing reciprocity constraints on the network’s topology. I conduct a simulation study to compare the predictions to pandemic and non-pandemic contact data and evaluate with respect to predictive accuracy of mean number of contacts within and between different age-groups. I compare weighted, sampling-weighted, and the unweighted reciprocal topology case.
Attrition is a key challenge for panel surveys. First, if attrition is systematic, it might bias the data. Second, attrition lowers the number of cases available for analysis. Adaptive survey design offers a solution to this challenge. Before fielding a panel wave, likely attriters can be identified and treated differently compared to the standards protocols. Such a tailored treatment aims to increase retention among cases likely to attrite from the panel. In the present study, we investigate whether increasing incentives for likely attriters can help to reduce attrition. We further investigate whether incentives for the treated should remain increased or be switched back to the standard after attrition was prevented once.
To investigate our research questions, we conducted a survey experiment in the German Family Demography Panel Study (FReDA). FReDA was recruited in 2021 based on a probability-based sample drawn from population registers (N~37,000). Each year, two panel waves are conducted in self-administered modes (web and mail). Respondents who did not participate in the first regular wave of the panel (W1A) were considered likely to attrite and randomly allocated to two experimental groups which received different incentives when being invited to the second regular panel wave (W1B). The first group received a 10€ prepaid incentive instead of the standard 5€ prepaid incentive, whereas the second group received the standard 5€ prepaid incentive. We continued the experiment in the next panel wave (W2A). Respondents of the first group in W1B were divided into two experimental groups. The first group again received the 10€ incentive, while the second group was switched to 5€. To investigate our research questions, we compared retention rates, indicators of sample balance, and data quality between the two groups.
This presentation introduces a newly developed survey scale to measure respondents’ political ideologies (POLID). On a theoretical level, the scale builds on the distinction between economic and social ideologies established in political science and psychology. This way the scale assesses political dispositions in a very detailed manner and remedies issues of previous operationalizations, such as the unidimensional left-right-spectrum often criticized for its inability to capture the complexity of ideological dispositions.
By formulating the items based on qualitative research about the core elements and core concepts of political ideologies and philosophies, the scale’s items are not tied to opinions about concrete issues that might lose their applicability during social changes or sudden events. Instead, the items relate to basic core assumptions of the four central ideologies of western countries (economic libertarianism, socialism, conservatism and liberalism) that are unlikely to change rapidly.
In sum, this scale allows to capture respondent’s ideological dispositions in a detailed manner by allowing them to gain four different ideological scores, one for each ideology. This enables researchers to assess their relationship and relative importance for each individual and/or group. The examination of group differences (e.g. gender or class differences) in ideologies is also possible with the help of the POLID-scale.
This presentation will outline the POLID-scale’s theoretical assumptions, the results of reliability and validity tests (including EFA and CFA), its correlation with established survey scales (e.g. Schwartz’ human values) and other important variables (COVID-conspiracism, educational level, income, age, etc.). For this, I used the Austrian version of the Values in Crisis Survey that measures peoples’ attitudes and values during the pandemic. Lastly, the presentation shall demonstrate its potential for use in future research.
Log-linear models were conceived for the analysis of multidimensional tables and are based on probability products formulas, which constitute an ideal framework for testing hypotheses of statistical independence between two or more variables at the same time (Imrey et al. 1981) with greater rigor than that achieved with the more commonly used contingency tables or correspondence analyses. In the academic literature, the work of Goodman and Haberman from the 1960s onwards in journals of applied statistics stands out, which were later popularized in manuals specialized in quantitative data analysis such as those of Andersen, Agrestri and Upton.
This paper proposes a procedure for the estimation of log-linear models and their graphical presentation through dynamic and interactive networks. It also presents a web page (CARING) for the estimation and representation of the parameters and the adjustments of the models by means of various examples with recent national and international surveys (CIS Barometers and European Social Survey).
What is the reality? What is true? Who decides and by what measures? ICT fundamentally influences the information household and the political structure of societies, how citizens can understand their world and represent their interests and values. The digital age offers promising circumstances for democratic societies to make effective and legitimate decisions, however, the experience shows the opposite picture: increasing democratic deficits along with decreasing quality of information which became more visible and crucial in a crisis.
We argue that the reason behind, this is the inadequate ICT, the web2, and in connection, the outdated democratic models used for public discussions which directly lead to the violation of expressive liberties, and in accordance, to the bad decisions, decreasing the trust between the citizens and the elite. We argue furthermore, that the usage of web3 technology with suitable models of democracy can not just eliminate the mentioned problems, but utilize the advantages which the digital age offers for democratic societies.
To justify this hypothesis, we describe governmental software developed on the technology of web3. This solution implements the cognitive model of democracy, the e-Cognocracy, and incorporates the legal matrix of expressive liberties supposing that the legitimacy and effectiveness of the decisions made can define the quality of public information developed during the democratic processes.
Therefore, we utilize the software to continuously collect real, and up-to-date data from the discussions focusing on the users’ behaviours, evaluations, and the semantics of the debates and their changes in time. Furthermore, we use the EF3 methodology of the e-Cognocracy to process and evaluate data by monitoring the Effectiveness, Efficacy, and Efficiency of the democratic process, hence verifying that the increasing quality of the information household is directly proportional to the legitimacy and the effectiveness of the decisions, which are based on argument-driven debates, where different ideas and opinions of different actors can compete.
The European Social Survey (ESS) is a large scale, biennial, cross-national general social survey which has been providing the academic community with open access data on a range of societal topics since 2002. Prior to the COVID pandemic all data collection took place via face-to-face in-home interviews. Using a single data collection mode was felt to be important for ensuring comparability within and between countries.
The pandemic meant that in some countries face-to-face data collection was not feasible . The ESS therefore piloted and launched a self-completion (web and paper) alternative in a very short period. This self-completion approach was used in around a third of the countries participating in the ESS during the pandemic with the others still using face-to-face.
This presentation will explain the development of the ESS self-completion approach and present initial evidence of its performance in terms of sample composition, measurement and wider data quality. Due to the initial promise of this approach and wider changes in survey infrastructure in many parts of Europe the ESS has decided to move to self-completion in all countries in 2027 or 2029. The presentation will discuss some of the challenges and opportunities of this transition and seek feedback from the audience on the needs of data users in this transition.
Building on our previous work (Ils et al., 2021), we assess how social solidarity towards migrants and refugees has changed before and after the onset of the COVID-19 pandemic, by collecting and analyzing a large, novel longitudinal data set of migrationrelated tweets. To this end, we first annotate above 2k tweets for (anti-)solidarity expressions towards immigrants, utilizing two annotation approaches (experts vs. crowds). On these annotations, we train a BERT model with multiple data augmentation strategies, which performs close to the human upper bound. We use this highquality model to automatically label over 240k tweets between September 2019 and June 2021. We then assess the automatically labeled data for how statements related to migrant (anti-)solidarity developed over time, before and during the COVID-19 crisis. Our findings show that migrant solidarity became increasingly salient and contested during the early stages of the pandemic but declined in importance since late 2020, with tweet numbers falling slightly below pre-pandemic levels in summer 2021. During the same period, the share of anti-solidarity tweets increased in a sub-sample of COVID-19-related tweets. These findings highlight the importance of long-term observation, pre- and post crisis comparison and sampling in research interested in crisis related effects. As one of our main contributions, we outline potential pitfalls of an analysis of social solidarity trends: the ratio of solidarity and anti-solidarity statements depends on the sampling design, i.e. tweet language, Twitter-user accounts’ national identification (country known or unknown) and selection of relevant tweets. In our sample, the share of anti-solidarity tweets is higher in native (German) language tweets and among ‘anonymous’ Twitter users writing in German compared to English-language tweets of users located in Germany.
The pandemic meant that in some countries face-to-face data collection was not feasible . The ESS therefore piloted and launched a self-completion (web and paper) alternative in a very short period. This self-completion approach was used in around a third of the countries participating in the ESS during the pandemic with the others still using face-to-face.
The Corona pandemic has caused fundamental crises in large parts of society. Areas such as education, the labor market and the health care system continue to be the focus of political, media and scientific attention. In contrast, the role of civil society, as a key site of social and political integration, has been insufficiently considered in recent years. In rural areas in particular, voluntary initiatives often play a key role in the supply of everyday services. Where associations and churches are central places of social interaction, the disappearance of these institutions further exacerbates existing difficulties.
This presentation explores the impact of spatial constellations in which volunteerism has been particularly affected by the pandemic. In addition to different spatial typologies, the presentation will also address the potential impact of location-specific issues. Both the hurdles perceived by the voluntary organizations themselves as well as data from official statistics are taken into account.
From what we can observe in our survey, associations have suffered from the pandemic in very different ways. Taking into account the spatial dimension is therefore essential for tailored funding of voluntary work, especially after the difficulties of the pandemic. It is thus not only highly substantial from a policy-making perspective, but also provides a new and fruitful perspective in the area of volunteer and spatial research.
The analysis is based on a German survey of approximately 2.800 randomly selected associations and religious communities in the period from November 2021 to January 2022 as well as on in-depth interviews with selected volunteer organizations which were conducted as a follow up to the survey. The data stem from the project "Between Appstore and Register of Associations - Rural Volunteerism on the Way to the Digital Age (AppVeL)", funded by the German Federal Ministry of Food and Agriculture(BMEL).
This presentation explores the impact of spatial constellations in which volunteerism has been particularly affected by the pandemic. In addition to different spatial typologies, the presentation will also address the potential impact of location-specific issues. Both the hurdles perceived by the voluntary organizations themselves as well as data from official statistics are taken into account.
From what we can observe in our survey, associations have suffered from the pandemic in very different ways. Taking into account the spatial dimension is therefore essential for tailored funding of voluntary work, especially after the difficulties of the pandemic. It is thus not only highly substantial from a policy-making perspective, but also provides a new and fruitful perspective in the area of volunteer and spatial research.
Despite the widespread use of web surveys, Internet coverage bias may still pose a threat to data quality. Using large scale probability-based Eurobarometer data, we put forward a pan-European study on coverage bias (Mohorko, de Leeuw, and Hox 2013) to: i) describe the trend in Internet coverage rate across Europe, ii) investigate demographic and socio-economic differences between the Internet and non-Internet population, iii) explore variation over time and across countries in Internet coverage bias, and iv) assess whether countries’ socio-economic context is associated with Internet coverage bias. We find that, still nowadays, a non-negligible share of the population does not use the Internet and Internet coverage varies widely across Europe. In addition, we document that coverage bias: decreases over time for most of the variables considered; seems more pronounced in age, education, and life satisfaction, and negligible in other variables; and is associated with countries’ socio-economic context.
The German National Educational Panel Study NEPS collects longitudinal data on educational processes, competence development, educational decisions and returns to education. Typically, in the NEPS starting cohort of adults, CAPI waves with competence tests alternate with CATI waves without competence assessment. In CATI waves, until before the COVID-19 pandemic, respondents had the option to switch to CAPI. Due to the pandemic, this was no longer the case in the last two survey waves. Currently we are discussing its re-implementation, considering the survey costs of F2F interviews as well as the fact that the share of F2F participants in CATI waves has continuously decreased. Thus, the question is, can we do without this option or does it still pay off? In the early stages of a panel study, it makes sense to offer different modes to recruit as many and different participants as possible. But what about older, established panel studies like NEPS with more than ten years of data collection? Do target persons who switch to F2F mode differ substantially from CATI participants? To answer these questions, we run selectivity analyses of participants from the last four CATI waves. The NEPS starting cohort of adults is a register-based probability sample of the German population born between 1944 and 1986. The realized mode in a CATI wave is the dependent variable (F2F vs CATI); design information and socio-demographics are explanatory variables. As result, we find that F2F switchers are indeed different from CATI participants in terms of age, education, participation behavior in previous waves etc. Thus, the possibility to switch to CAPI is an important measure to reduce biases, but after several survey waves it is no longer efficient, due to low case numbers. As consequence, we will not offer the option to switch to CAPI in the upcoming CATI waves.
Up to this date, most survey methodology research is still based on source material exclusively from Europe or the USA. While it is not without reason, most of the established surveys are still conducted in the West, ignoring initiatives from other continents results in an incomplete or even biased picture.
However, someone looking for harmonized information about characteristics of comparative survey programs worldwide would encounter a critical issue. While endeavors like Data Documentation Initiative, and similar, for many years, tried to establish an industry standard for survey reporting, their work, unfortunately, did not yield enough. Therefore one, in most cases, must gather and harmonize such data by hand (reporting standards usually do not enable even the usage of machine learning techniques.)
Therefore, in my paper, I will rely on analyses of methodological reports and outcome data from such comparative survey projects as: Afrobarometer, Arab Barometer, Comparative Study of Electoral Systems, East Asia Social Survey, International Social Survey Programme, Latin American Public Opinion Project, Life in Transition Survey, World Values Survey, etc. taking into account data from the early 1990s to 2019. I will present: 1. Distribution of characteristics such as survey mode, frame, sampling design, and various aspects of "fieldwork effort" (material incentives, advance letters, required number of calls, etc.); 2. Data on completion rates, as well as internal and external representativeness criteria - calculated from survey results and UN demographic statistics; 3. An assessment of the completeness of documentation for different years, countries, and projects.
The presentation is expected to provide the audience with a broad overview of the different approaches to survey research done in various parts of the globe, hopefully facilitating future in-depth studies. The additional gain would be to assess how the quality of the survey documentation and survey samples in cross-country projects is changing and what the differences are between various world parts in this matter.
We present changes in various measures of partnership quality based on the still on-going Czech GGS (Generations and Gender Survey) survey. We utilize a preliminary version of the data set covering May through December of 2021. COVID was a major methodological complication for probability household surveys and resulted in ruined interviewer networks, lower participation rates and eventually a much lower efficiency in the data collection exercise. Consequently, Czech GGS had to extend its fieldwork period significantly in order to achieve its planned sample size. In this paper, we turn this methodological complication into a substantive benefit. We split the sample (over 1500 interviews) into bi-weekly subsamples and use them (with covariate adjustment) to describe trends in partnership quality between May and December of last year. We observe a significant decline in relationship quality over time (with 35 % of partners thinking about splitting up in December!), which sped up toward the end of the observation period arguably reflecting not only new COVID cases, but also rising price levels and accumulated economic uncertainty.
In Germany, studies have shown that official coronavirus disease 2019 (COVID-19) vaccination coverage estimated using data collected directly from vaccination centers, hospitals, and physicians is lower than that calculated using surveys of the general population. Public debate has since centered on whether the official statistics are failing to capture the actual vaccination coverage. The authors argue that the topic of one’s COVID-19 vaccination status is sensitive in times of a pandemic and that estimates based on surveys are biased by social desirability. The authors investigate this conjecture using an experimental method called the item count technique, which provides respondents with the opportunity to answer in an anonymous setting. Estimates obtained using the item count technique are compared with those obtained using the conventional method of asking directly. Results show that social desirability bias leads some unvaccinated individuals to claim they are vaccinated. Conventional survey studies thus likely overestimate vaccination coverage because of misreporting by survey respondents.
The interventions that were taken in many countries as a response to COVID-19 pandemic brought into the lives of individuals many interventions (in a form of e.g., curfew, quarantine, and assembly bans) that were aimed at reducing social interactions. The disease and the interventions brought into the lives of the most population a significant change. Restrictions have caused discontinuities in many aspects of daily life and COVID-19 thus represents a significant disruptor of everyday life. Perception of that disruption deserves a significant research attention. There is a high demand from public policy makers to identify when epidemic situation and imposed interventions are becoming unbearable for most of the population. Therefore, it is useful to know what contributes to perceived disruption.
The key research question is how individuals reflect this change, to what extent they perceive such change as disruptive and how difficult it is to handle the new reality. Furthermore, the aim of this paper is to analyze the level of perceived disruption within some sub-segments of the general adult population.
For this purpose, the research based on a robust representative sample derived from the general adult (15–74 years) population was conducted. The sampling technique was a multistage random procedure using address-based sampling. Datafile comprised 1 372 cases; data collection had a form of face-to-face interview.
To identify the associations of perceived disruption, the validated Fear of COVID-19 scale (FC-19S) was used along with other measures representing the key aspects of the quality of life, reported type of family bonds and self-reported mental and physical health. It was observed that selected items differentiate the level of perceived daily life disruption among diverse sub-populations.
The ambition of the project (called “Jagiellonian Panel 2002+”) is to start and test the conditions for carrying a panel study (over longer period of time) in Poland among young citizens at local level (city) especially in time of COVID-19. It begun in 2021 in Cracow with 204 persons born in 2002 (18+) who gave consents to be registered as panelists. The target group were young people who were at the end of their last level of their obligatory education, as then selection procedure could be under control (via schools) with voluntary option of participation in the study. In already two waves of surveys were carried with that group in 2021. We managed to obtain high response rates (I round – 76%, and 95% respectively). The next cohort (born in 2003) have been already invited to participate in the study in 2022. In the questionnaire there are included questions which allow to monitor the respondents’ activity over the life course (labour market, social activities) as well to include ad hoc questions which are typical for certain events in their lives (for their cohorts like fear towards final exams they took). This project currently is using the online way of data collection (Internet via e-mails or/and via mobile phones). The key challenge is to maintain participation and so high response rate in following surveys especially when any financial incentives have been offered for participants. The organization of the study and the ways of increasing participation in surveys will be presented. However, besides of solutions as well problems related to recruiting the young people to such research will be discussed. Our conclusions could give the light how others planning such projects can avoid (un)expected difficulties during the organization such type of research and to carry out their own recruitment more effectively.
Para-data could give important insight into how respondents answer questions in self-report assessments: surveys and questionnaires. Technical capabilities of data collection in computer-based assessments were noticed fairly quickly. However, none of the early proposals on how computer-based paradata should be collected was embraced by researchers, probably because of a rather difficult implementation, requiring basic knowledge of programming and web interface. In this presentation, a framework that allows for easy implementation of log-data collection within LimeSurvey open web-surveying platform is presented (also, possibilities to adapt it to other web-surveying platforms will be shortly discussed). Moreover, an accompanying R package that enables easy data transformations and calculating a wide group of response process indicators (Goldhammer et al., 2021; Kroehne & Goldhammer, 2018) is also presented. Various response time indicators, cursor/mouse moves/trajectories and velocity, and hovering (Horwitz et al., 2017; 2020) are among process indicators that can be obtained by processing log-data in this package.
Web surveys offer new research possibilities, but they also have specific problems. One of them is a higher risk of careless, inattentive, or otherwise invalid responses. Using paradata, which are data collected apart from reactionary data, is one of the potential tools that can help to screen for problematic responses in web-based surveys. One of the most promising forms of paradata is the movement, or trajectory, of the cursor in making a response. This study presented correlations between such indices and constructed components representing their common variance. Moreover, it provided an interpretation and validation of these components (interpreted as indicators of complexity, hesitancy, and regularity), among others, by correlating them with previously known indices of careless responding. Finally, by employing a between-participant experimental design, it tested cursor movement indices during different motivational states induced by experimental instructions.
Log event data of computer-based administered surveys, questionnaires and tests arise from user-interactions and internal system behavior, very specific to the concrete platform and software used for data collections. At the same time, the overarching goal of log data analysis is to create process indicators, for which validation is understood as a process of cumulating theoretical arguments and replicated empirical findings supporting their interpretation regarding the measurement of inter-individual differences. To bridge this gap, Goldhammer et al. (2021) describe different extraction levels for assessments as reasoning from evidence derived from log data. Evidence is brought from the depth of the platform-specific log events via low-level features to the highest level of interpreted process indicators. However, assessments often contain multiple components (i.e., multiple items but also multiple constructs), and besides the vertical dimension, there is also a horizontal perspective on process indicators over the course of an assessment. The relationship of a process indicator for dis-engaged responding across different parts of an assessment is investigated in this presentation. Theoretically, we compare the concepts of rapid guessing (in tests) and rapid responding (in questionnaires) and summarize empirical knowledge about the development of dis-engaged responding over time. On an operational level, we illustrate how the same process indicator can be constructed from log event data of a very different nature. Finally, using empirical data from the PISA 2015 assessment of selected countries, we investigate the dimensionality of indicators for rapid responding within cognitive and non-cognitive domains as well as their relationship. Based on the empirical example, we finally try to answer whether rapid responding in this particular questionnaire results from low reading competence or whether rapid responding in the questionnaire and rapid guessing in the cognitive tests are due to the same phenomenon, namely low test-taking engagement.
The recent switch to computer-based data collection in surveys and questionnaires enables the collection of rich sets of paradata (Couper, 1998/2000). Such paradata can then be used to construct process indicators enabling us to get to know more about the participants’ response behaviour and data quality (Horwitz et al., 2017; 2020; Kroehne & Goldhammer, 2018). However, such process indicators still lack thorough validation. This presentation aims to fill in some of these lacunas by presenting process indicators validation based on comparing data from survey experiments. In a series of web-survey experiments, task difficulty and participants' motivation were manipulated to simulate satisficing (Krosnick, 1991) and to check how formal survey characteristics, e.g. number of response categories, affected process indicators based on response time and mouse moves/trajectories. Furthermore, we compare log-data indices under induced conditions of working memory burden and forced multitasking. Moreover, log-data indices cross-assessment stability is tested as some of the data was collected in a test-retest design. The presentation concludes with a reflection on the utility of currently used process indicators in predicting and modelling survey responding behaviours.
Background: The COVID-19 pandemic and associated social distancing measures such as homeschooling have led to changes in everyday social interactions among adolescents. This has been accompanied by an increasing use of digital media. In this context, there is a need to investigate how different patterns and modes of interaction during the pandemic affect the mood of adolescents.
Objective: The aim is to investigate the relationship between different forms of social contact and daily mood changes in adolescents during different stages of the COVID-19 pandemic. The study also examines the extent to which the relevance of different forms of contact differs between genders.
Method: Using a smartphone-based experience sampling design, German adolescents were questioned up to eight times in four weeks in February (N individuals=290; N observations=1796) and November (N individuals=160; N observations=1061) 2021. A series of fixed-effects regression models were estimated and separate analyses were carried out for boys and girls.
Results: The fixed-effects regression models support that personal contact with friends is positively associated with mood during both the lockdown in February and open schools in November. The influence of chat contact with friends varies. During the hard lockdown and school closures in February 2021, the relationship with mood is slightly positive but non-significant. In November, when schools were open again, there is a significant negative correlation. The models also revealed gender differences, indicating that the contact mode is not relevant for boys. Girls benefit more from face-to-face contact and show a more pronounced negative association between mood and chat contact in November 2021.
The COVID-19 pandemic has affected many aspects of people’s daily lives worldwide. The requirement to socially distance oneself – which might more accurately be described as “physical distancing” – has added more fuel to the debate on the role of social capital in building community resilience and health. Many studies have investigated the effects of social capital on health outcomes, as well as its role in people’s response to disease outbreaks. Few studies to date have focused on the effects of the COVID-19 pandemic on social capital.
The present research contributes to this topic by investigating how institutional trust, a specific dimension of social capital, was affected during the first and second year of the COVID-19 pandemic. To this end, I analysed data collected from Eurobarometer 93.1 (2020) and Eurobarometer 94.3 (2021). For each year, an additive index of institutional trust was created and used as a proxy of social capital. Two multilevel models were applied to identify the effect of the independent variables on institutional trust: one for the first year of the COVID-19 pandemic, and another for the second year. On the individual level, I considered demographic features and frequency of internet use as the predictive variables; whereas on the country (or second) level, I investigated, for example, the percent-positive rate, the COVID-19 mortality rate and the DESI index. The total number of individuals included in the analysis was 26,681 for the year 2020, and 27,409 for the year 2021. All data refer to EU citizens.
The results show that the country of residence has a differential impact on institutional trust, and that some socio-demographic features are associated with higher levels of trust in the institutions; for example, being female and from a high social class increases social capital more than being male and from a low class. Of the contextual variables considered, only the dimension indicating satisfaction with the measures taken to fight the pandemic by the national government had a highly significant effect.
The COVID-19 pandemic crisis has aroused the interest of public opinion on the issue of vaccines, encouraging the dissemination of the views of people who are reluctant or unwilling to vaccinate. These kinds of attitudes can be related to several socio-demographic variables, such as the educational level and the social background. Nonetheless, while these variables are usually stable through time, there are many others which can quickly change. The research is mainly focused on these types of determinants, specifically the concern of being infected, the people’s assessment of government policies and the trust in institutions. The data was collected by means of a web survey administered to a sample of 2787 Italians in April-May 2021 (t0), also including some vaccine hesitancy related questions. In order to observe whether the above determinants have change before and after the starting of the vaccination campaign, they were compared with some findings of a previous questionnaire, which already included them, aimed to the same sample in April-May 2020 (t1). On this way, it is possible to observe and quantify their variations from (t0) to (t1). Moreover, also the quality of the variation is accounted. In fact, sometimes change might lead people to adopt different attitudes, and sometimes it does not. Eventually, a model including this information and the socio-demographic variables is developed throughout a multinomial logistic regression, with the aim to identify the predictors of vaccine hesitancy. The analysis therefore allows to explain how social conditioning affects quantitative and qualitative change of attitudes in one year of pandemic in terms of concern, assessment of government policies, and trust. By proceeding in this way, the wishing is to promote the knowledge about the phenomenon of vaccine hesitancy in a perspective which takes into account the ongoing processes of social mutation, characterized by uncertainty and precariousness.
The aim of the study is to investigate the relationship between migration background and COVID-19 vaccination intentions, exploring multiple mediation paths. We argue that the migration and sociocultural background influence general attitudes toward health and political/public institutions. The effects of these general attitudes on vaccination intentions are mediated by fears of infection. Additionally, we analyze a migrant-only model including acculturation variables (years since migration, foreign and host country media consumption) and region of origin (European vs. Non-European). Design: The data (n = 1027) stem from an online access panel collected between March 15 and March 25, 2021. Quotas for gender and age were set according the online population of Germany. The use of an oversampling framework for first generation migrants resulted in a sample with 50% first generation migrants and 50% native Germans without migration background. Models were calculated using a Structural Equation Modeling approach.
Migration background both increases and decreases antecedents of vaccination intentions. Being a migrant increases positive antecedents like religiosity, which in turn positively influence general attitudes and thus fears of infection and vaccination intentions. But being a migrant has also a significant direct negative association with vaccination intentions, implying missing mediators. Increasing years since migration increase host country (German) media consumption and decrease consumption of media from the country of origin. Both media variables are positively associated with political trust and health consciousness. Additionally, European compared to Non-European migrants have less political trust, fear of personal infection and lower vaccination intentions on the whole.
The study found that vaccination intentions can be understood by applying the proposed hypothetical structure. We found complex associations of the migration and sociocultural background and COVID-19 vaccination intentions, where antecedents of vaccination intentions are both increased and decreased by migration background and migration specific factors.
Air pollution in India has steadily increased in recent decades, and has reached such high levels, especially across the Indo-Gangetic Plain, that it is now receiving more and more attention, both in scientific discourse and among the general public. One aspect that has been studied mainly for individual large cities is the social distribution of this pollution. Currently, there is only one study by Chakraborty and Basu (2021) which also examines environmental inequality for all of India, based on aggregate survey data from the 2011 Census and a model of fine particulate matter (PM2.5) distribution. We take a similar, but enhanced approach, basing our analyses on nationwide survey data at the household level from the National Family Health Surveys 4 and 5 (2015-16 and 2019-21), which we link to an improved updated version of the model of annual average PM2.5 pollution, provided by the Atmospheric Composition Analysis Group (Washington University in St. Louis), and data from SMOG 2015, an India-wide bottom-up registry of local emissions of various air pollutants. This allows us to examine how wealth, education, and membership in marginalized groups affect residential exposure to air pollutants in India. Minorities we consider are the state-recognized ones of Scheduled Castes, Scheduled Tribes and Other Backward Classes, and additionally an indicator for local religious minorities. Some of these minorities are very unevenly distributed between heavily and lightly industrialized areas. In addition, there are also strong differences in air pollution levels between different areas for meteorological reasons. Therefore, it is necessary to consider fixed effects for different spatial units - districts, states, and larger climatically homogeneous regions of India.
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: Our 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.
The aim is to show an example of an outcome-wide longitudinal design approach for empirical studies. This approach is used to assess the effects of a treatment/exposure over numerous outcomes using confounding control. This approach is useful for not only revealing patterns of associations that may not be apparent if a single outcome was examined but also for limiting the risk of cherry-picking only significant results.
Using longitudinal data from 19,821 middle-aged and older adults from 15 countries participating in Survey of Health, Ageing and Retirement in Europe (SHARE), we examined the prospective associations between voluntary and/or charity activities and 21 outcomes from the domains of (1) physical health, (2) emotional well-being, (3) quality of life, and (4) cognitive impairment at a 6-year follow-up. The associations were examined using generalized estimating equations. Adjusting the standard errors and clustering by country were applied to account for the hierarchical nature of data. Three types of estimates were reported: (1) for continuous outcomes - standardized regression estimates, (2) for dichotomous outcomes - odds ratios for rare outcomes and risk ratios for non-rare outcomes. All models were controlled for prior sociodemographic, personality, lifestyle factors, health behaviors, history of health conditions, and pre-baseline values of all outcome variables simultaneously. E-values were calculated to examine sensitivity of associations to unmeasured confounding. Secondary analyses provided further evidence on robustness of the results.
The results indicated that middle-aged and older adults participating in voluntary and/or charity activities at least once a week have lower risk of limitations with daily life functioning (ADL and IADL), lower risk of cognitive impairment, report higher scores on emotional well-being and score lower on the loneliness scale. Participants who engaged in voluntary and/or charity activities had also a substantially lower odds of being diagnosed with Alzheimer’s disease (by 47.7% compared to respondents not participating at all) and lower risk of diabetes (by 16%). No evidence of association with physical health outcomes including heart attack, hypertension, high blood cholesterol, stroke, diabetes, chronic lung disease, and cancer (at the 6–7-year follow-up) was found.
The fertility of immigrant populations and their descendants has played an increasingly important role in the demographic dynamics of European countries in recent decades. However, given that Spain began to receive large migratory flows at the end of the 20th century, little is known about the fertility level of immigrant descendants.
We study the fertility level of Maghrebi and Latin American immigrants and their descendants in Spain. We use a new database linking Natural Movement of the Population records between 2011 and 2015 to the 2011 Spanish Census.
Our sample is made up of women born between 1950 and 1969. We analyze first, 1.5 and second-generation immigrants originating from Latin America and the Maghreb. To account for the diversity of fertility levels in Latin America, we divided Latin American immigrants into those coming from countries with a Total Fertility Rate (TFR) lower or higher than 2.1 in 2011. We compare the complete fertility of these immigrants to that of native women, the dependent variable being the number of children had at the end of 2015. To do so, we implement Poisson Regression models, controlling for relevant factors like educational level and birth cohort, among others.
The main hypothesis is that while first-generation immigrants originating from countries with higher TFRs will have higher fertility than native Spanish women, immigrants descendants will resemble native women closer.
Both Maghrebi first-generation immigrants and Latin American first-generation immigrants originating from countries with higher TFR than 2.1 in 2011 have a much higher fertility level than native Spanish women. Latin American first-generation immigrants originating from countries with lower TFR than 2.1 in 2011 have a very similar fertility level compared to native women. 1.5 and second-generation immigrants resemble native’s fertility level closer. We also observe that among second-generation immigrants, those who have one parent born in Spain and the other abroad, resemble native’s fertility level closer than those whose both parents were born abroad. We conclude than immigrant origin and generation are relevant in explaining the fertility level of immigrant populations.
Report and planning of activities of the research network.
This paper analyzes the results of the ParWelB project, which is an interdisciplinary study that aims to elaborate a pivoted socio-psychological model for assessing preterm parents’ wellbeing during and after the hospitalization in Neonatal Intensive Care Units (NICUs). When we designed the project, the COVID-19 pandemic had already imposed dramatic changes to our habits, including the possibility to enter hospital wards as external researcher. To overcome these issues, we decided to combine mixed methods social research with an e-health approach. The term e-health calls not only for technical development but also for a deeper commitment in raising awareness, promoting peer-to-peer communication, enhancing networking and interdisciplinarity, voicing patients’ experiences and, consequently, improving health care. Drawing upon the idea of Timmermans and Berg (2003), an orientation of “technology-in-practice” allows a critique by scholars in the social sciences regarding the complicated modalities in which the “social” and the “material” intertwine in technologies for health care, in this paper we aim to foster a deeper understanding on the role quantitative methods can play in the implementation of this research field.
The paper will focus on the parents of children born before the 36 + 6 GA admitted to two neonatal intensive and sub-intensive care units in Milan (Italy), and hospitalized for at least ten days (critical threshold of hospitalization length). In both NICUs, we applied standardized questionnaires with internationally validated scales to assess parents’ wellbeing and self-efficacy perception, at the moment of discharge and then — through the use of a web-app purposely designed for the project — during the first year of life of their child. Thus, we investigate the relationship between healthcare staff and parents in a sociological perspective, paying particular attention to the strengths and limits of the collection of quantitative data in e-health care.
The present work builts on the topic of 'disaster studies' which the most significant theoretical aspects are: a) the purpose of studying «meaningful interactions» between the social system and the individuals and their actions; b) the conviction that, in cases of pandemic or natural disasters, pain and tragedy can take on unequalled levels of emergency, generating powerful effects as relates to the rhythms of an ‘ordinary’ life; c) the role of cognitive processes in conditioning daily life and determining individual predispositions towards the future.
Starting from this perspective, the research explores the results obtained from a web-survey conducted in 2021 on all Sapienza University of Rome’ students who, during the last three years, have spent an academic mobility period outside Italy or inside Rome. In detail, analyses explore variations and influences of the cognitive-emotional dimension (beliefs, desires, opportunities) on the (intention-to-) action dimension (attitudes, behaviours, actions), especially considering the physical and mental health aspects. For this purpose, two indexes, based on the mental well-being of the interviewed people, were constructed: the 'distress' and the 'resilience' one, which were analysed in the light of the socio-demographic variables, the physical activity and the lifestyle sections. The results obtained show how the distress index defines women more than men and it tends to be higher among those with a low socio-economic and cultural capital; on the contrary, the resilience index is higher for men, young people and among persons characterized by medium/high socio-economic and cultural capital. In addition to this, looking at the physical activity, the analysis carried out show that those who did not engage in it during the pandemic period had a higher level of distress; vice versa, the resilience index is higher among those who practiced physical activity intensively during the pandemic period, especially among those whose goal was to improve their health. Moreover, looking at the lifestyles, the distress index is higher for those who have experienced an increase in difficulty falling asleep, in smoking and in using alcoholic beverages.
Background: 100 years after the 1918 flu pandemic, the COVID-19 pandemic constituted an unprecedented health crisis on a global scale. Disproportionately affecting the elderly, it caused uncertainty and concern for aging societies in general and the elderly in particular. From a methodological standpoint, this situation offered a unique opportunity to investigate the influence of health crises and threats on health assessments. Building on this idea, we investigated the influence of the pandemic's extent and closeness on health reporting behavior.
Data & Methods: The analyses of this paper are based on data of the German-Austrian Cloister Study collected from October 2020 to March 2021 which were linked to administrative data on each country's number of daily new infections and deaths on the day of the interview. Using generalized structural equation modeling, we investigated the indirect effect of the pandemic's extent and closeness on reporting behavior regarding self-rated health mediated by COVID-19 anxiety and mental health by age group. Reporting behavior was isolated by controlling for physical health prior to the analyses.
Results: The number of new infections or deaths in their country did not significantly affect respondents' COVID-19 anxiety. However, an infection in the respondent's close environment did significantly influence COVID-19 anxiety in respondents younger than 80 while older respondents' worries were present independently of self-reported infections in their social network. In general, COVID-19 anxiety had a negative impact on mental health in both groups which, in turn, negatively influenced reporting behavior with regard to self-rated health.
Conclusions: As shown with the example of COVID-19, health crises and threats indirectly influence health reporting behaviors via mental health and health anxiety. Given that the extent of health anxiety appears to correlate with the (assumed) individual risks of the health threat, systematic reporting biases can arise.
Background: The COVID-19 pandemic increased forms of precarious employment. Precarious employment has been found to deteriorate mental health through health-behaviors and socioeconomic insecurities. Allegedly, the COVID-19 pandemic strengthens the association between precarious employment and low mental health.
Objective: This study aims to (I) investigate pathways of precarious employment on mental health before and during the COVID-19 pandemic. (II) Quantification of the pandemic-related increases in these pathways. (III) Demonstration and application of recent methods for longitudinal mediation analysis.
Method: The German Socioeconomic Panel (GSOEP) and the additional COVID-19 sample (GSOEP-COV) were used to investigate objectives I-II. The GSOEP biannual information on mental health before (2014, 2016, 2018) and during the COVID-19 (2020) pandemic were combined with the SOEP-COV sample (N=3,161), which allowed accounting for pandemic-specific confounders, such as threat perception of COVID-19.
For Investigation of objective III, the mediational g-formula was applied with precarious employment as the predictor (X), mental health (SF-12) as the outcome (Y), and socioeconomic insecurity and health behaviors (heavy drinking) as the mediators (M). The COVID-19 pandemic was modeled for with time-specific direct and indirect effects and time-varying confounding. Further, interactions between the length of precarious employment before the pandemic and the mediators were added to account for different burdens.
Results: The negative relationship between precarious employment and mental health (direct effect) increased during the pandemic. The longer the exposure to precarious employment before the pandemic, the more important the mediators for mental health (indirect effect). Objective III results contrasted the mediational g-formula with traditional methods, such as structural equation modeling. The mediational g-formula was found to be superior to these methods when analyzing dynamic processes that involve (1) time-varying confounding, (2) X-M interactions, and (3) time-dependent non-linearities.
This paper uses an innovative extension of multilevel modeling to examine the extent to which intersectional social identities combine to shape risks of loneliness in a Swiss municipality. In Switzerland, 38% of the adult population experience loneliness, which is more prevalent among older adults and individuals with a migration background. While past interventions have helped to reduce loneliness by fostering social connections, these interventions were often based on unidimensional and broad demographic categorizations (e.g. older adults or foreigners), neglecting the intersectional and multiplicative nature of social identities, thereby limiting the precision of interventions to enhance social inclusion. Using data collected in 2019 from a longitudinal participatory research project (n=1,360), we sought to understand the extent to which intersectional social identities combined to shape risks of loneliness in a local municipality. Employing novel and innovative multilevel techniques from social epidemiology, we found that 56% of the variance between intersectional groups was explained by multiplicative identities (age x gender x nationality x education), above and beyond the additive effects of social identities (age + gender + nationality + education). In addition, we identified that individuals who were simultaneously non-Swiss and aged 65+ and male and have primary educational attainment only were most at risk of loneliness and would be logical intervention targets to reduce loneliness. Methodological and practical implications will be discussed.
In the literature, social class is considered as a “multidimensional concept” defined by a number of socio-economic variables such as income, level of educational attainment, employment status and other lifestyle variables (home ownership, number of children under 18 at home). However, although its theoretical basis is agreed upon, this does not hold true for its measurement as in many instances it is treated as identical to socioeconomic status. As “respondents’ perceptions of social class, while generally shaped by income, are not entirely consistent with strict income-based definitions of social class”, the self-perceived question included in the World Values Survey (WVS) questionnaire serves as a more encompassing measurement classifying respondents into five class categories: upper, upper middle, lower middle, working and lower class. In this respect, the purpose of this presentation is to explore factors impacting social class perceptions extending previous work by including more indicators namely national pride, political interest and religious affiliation. The analysis is based on the 2017-2020 7th wave WVS datasets for Greece. Each of the five social class categories is modeled separately using linear probability models to allow for comparisons among them. The main findings of the regression analyses show that respondents with primary education or less are significantly more likely to feel that they belong to the working of lower classes and those with at least graduate degree to the upper middle class. Respondents with high income are significantly more likely to associate themselves with the upper and upper middle classes and those with low income to the lower class. Nationally proud respondents are significantly more likely to believe they belong to the lower middle class and less to the lower class. Respondents very interested in politics are more likely to classify themselves in the upper and lower classes, whereas those somewhat interested and not at all interested with the lower middle class and lower class, respectively. This study shows that, apart from the socio-economic factors, more indicators have an impact on social class identification and further research is necessary to provide a more coherent interpretation.
The aim of the study is to investigate how attitudinal factors shape the acceptance of emerging sustainable technologies. We try to address two questions: firstly, we investigate how acceptance of sustainable technologies come into existence. For this purpose, we formulate a model where acceptance of sustainable technology is a function of environmental, prosocial and technology attitudes, as well as beliefs about the cost and benefits of the technology. Secondly, we seek to identify structures of similarity regarding technology acceptance in the context of attitudes and value orientations. The study is part of an interdisciplinary research project on vertical air-conditioning and water recycling systems (VertiKKA), funded by the Federal Ministry of Education and Research (BMBF). We use panel data (n=201) from two waves (w1= 2020, w2=2021), randomly drawn from two districts of the city of Cologne, Germany. Interviews were conducted in both PAPI and push2web online CAPI mode. We apply autoregressive and cross-lagged Structural Equation Models (SEM) and k-Means cluster analysis.
Fort the first part, we find stable autoregressive effects for most of the attitudinal factors analyzed. Yet, the main dependent variable, acceptance of sustainable technology, shows more volatility. While prosocial attitudes have no impact in the entire model, environmental and technology attitudes are significant background variables in the emergence of acceptance towards sustainable technologies. Regarding the beliefs about costs and benefits of the technology, it can be shown that benefits have a greater influence on the acceptance than the cost aspects. The second analysis reveals three acceptance-type clusters with two clusters of high and one cluster of low acceptance. High acceptance is associated with high levels of prosocial and environmental attitudes, as well as conservatism, openness and universalism. The low acceptance cluster, which makes up about 30% of the sample, scores lower in most of the attitudinal and value dimensions.
Content analysis is a technique used in the social sciences for the systematic study of the contents of the communication. It is intended to be a valid, systematic, and replicable procedure for condensing words, sentences, or characters into a set of categories based on given rules of coding.
Among the methods of identification of these textual contents, two fundamental forms stand out: manual coding and semi-automatic coding with the use of dictionaries. Numerous software packages are available for this type of analysis: some are qualitative as Nvivo, Atlas-ti or QDA miner, and some are statistical as WordStat TextAnalyst or LIWC.
The TST (Twenty Statements Test), originally developed at the Iowa School by Kuhn and McPartland (1954), is a tool used to investigate personal identity, understood as the way in which individuals conceive themselves as unique and different subjects. It was conceived as a methodological instrument for the empirical study of self-identity and consists of giving twenty open answers to the question "Who am I". As it is an instrument made up of multiple open-ended responses from each subject, it allows the application of natural language processing tools to the textual analysis of open-ended responses with the aim of analyzing the different categories of personal identity. It will be analyzed the answers provided by a sample of more than 5,000 people from Spain, Chile, USA, England, and South Africa surveyed through web.
In a first phase, the responses were coded manually using a multidimensional categorization while in a second phase automatic coding with regular expressions were used using the software WordStat. Both codifications will be compared, and the results will be presented.
Opportunities and challenges of combining survey and digital data.Discussion on the new options to enhance survey data with digital data and other sources (e.g. admin data), and their impacts on data quality. Challenges of using new data sources for methodologists and data users: consent and linkage, specific sources of error….
With the SARS-CoV-2 coronavirus outbreak, the need for high-quality surveys as a basis for pandemic management and for insights into various attitudinal and behavioural aspects of the crisis has increased enormously. At the same time, contact restrictions and the fast-moving nature of regional developments make primary data collection much more difficult, and most surveys have had to adapt their survey programmes accordingly. However, the impact of the pandemic on the quality of the surveys has rarely been addressed. Participation in surveys could be less frequent of illness, sick leave, and caregiving situations.
Using data from five early childhood education and care surveys, this study examines whether county-level COVID-19 incidence affects the likelihood of survey participation in a given week. The surveys were conducted from March to September 2020 among directors and pedagogical staff in day-care centres, their providers, family day-care workers and youth welfare offices. For surveys in the field of early education and care, the closure of day-care centres in particular may have led to a decline in participation.
Using discrete survival analysis, we find a negative effect of the COVID-19 incidence on the likelihood of participating in four of the five surveys (not for youth welfare offices). The negative effects appear to be due to day-care closures and certain time effects of incidence rates (habituation and shocks) in two surveys.
We conclude that the COVID-19 pandemic had a negative effect on participation behaviour in the ERiK-Surveys 2020. It remains to be investigated whether such effects can also be detected in other surveys and whether a selective participation behaviour can be observed.
In this paper we present the main results of empirical research based on a structured questionnaire administered to a sample of 6689 secondary school students in Italy aimed at understanding general attitudes towards the COVID-19 vaccine. The research connects the attitude of mistrust and aversion towards vaccines to forms of social insecurity. The work takes up the conceptual categories of Mannheim's (1929) sociological tradition and in particular the study of the relationship between thinking style, social origin and communicative codes also developed by Basil Bernstein (1971, 1999) and his epigones (Maton 2009; Young 2014). The restricted code, which is used by the most disadvantaged classes, is in fact employed in a family context characterised by precarious material conditions and a consequent state of anxiety, where communication focused on family problems rather than on issues of public interest prevails.
Starting from this theoretical framework, through the construction of multivariate analysis models it can be hypothesised that an individual's level of social insecurity is associated with: a restricted code oriented towards the solution of practical problems and developed within more problematic family environments; a lower level of education, which also affects the degree of trust in scientific knowledge; the propensity to view COVID-19 vaccines with aversion, due also to the effects of social origin, which is mediated by educational status. More specifically, this research is based on the use of hierarchical linear and logistic regression models that allowed us to distinguish the direct effects of social status from those mediated by it through factors relating to family socialisation and thought style.
COVID-19 pandemic has supposed, and still does, many deaths. However, its incidence has been higher in vulnerable groups of the society. Older adults and dependent are population groups with higher probability of dying due to age, chronic conditions, and functional limitations. The aim of this work is to analyse the adjusted mortality rates of dependent people using administrative date of the long-term care system of Castilla-La Mancha.
The data were obtained from the Social Welfare Counselling of Castilla-La Mancha and we merged this information with the National Death Index of the Ministry of Health. We estimated the weekly mortality rate over the whole dependent people and by sex, age, level of dependency, among others. We also estimated the mortality excess and the years of life lost during the first waves of the pandemic. We also studied the sociodemographic factors associated with death during the pandemic compared with a non-pandemic time.
Our preliminary results show that mortality excess is 3,001 people between weeks 16-20 of 2020. On the other hand, being male, having more than 65 years, higher level of dependency and living in a nursing home are positively associated with death in both periods, with an increase in the pandemic time.
These results are valuable for policy makers in healthcare and social care systems.
The COVID-19 pandemic has produced a return towards a more traditional allocation of household and childcare responsibilities, with women disproportionately carrying the burden of closed schools and childcare facilities. In the meantime, these material inequalities have settled back on their pre-crisis level as schools and kindergartens opened again; however, the temporal allocation of household responsibilities in accordance with more traditional gender roles might also be associated with rather enduring value shifts towards less gender-egalitarian attitudes. Moreover, the experience of an existential threat has been found to be related to the desire for a preservation of the (pre-crisis) status quo, and to more conservative political attitudes. In this paper, we investigate the following questions: Firstly, to what extent did perceptions of threat produce more traditional gender-role attitudes among women and men during the first two years of the pandemic? And secondly, did demands for assuming childcare responsibilities lead to more conservative political preferences among women? Furthermore, is this relationship mediated by more traditional gender-role attitudes? We test our hypotheses using panel data from the Values in Crisis (ViC) Austria survey, which has collected data on a range of political and social attitudes in 2020 and 2021. We apply a structural equation modelling (SEM) approach, which allows us to use path analyses to identify the direct and indirect effects of pandemic-related threat perceptions, of working from home, and of assuming childcare responsibilities. Our initial findings suggest that women, who assumed childcare responsibilities during the first wave of the pandemic indeed displayed more conservative political attitudes during the second wave, and this effect is partially mediated by an endorsement of more traditional gender-role attitudes.
Standardization of measurement is a prerequisite for cross-national and/or overtime comparative analyses. However, there are instances in the literature where the validation of constructs resulted in producing scales or subscales defined differently from the proposed theoretical structure and across countries. In this paper, we propose an empirical methodology that provides standardized overall measurements of unidimensional constructs to be used in cross-national and overtime comparative research. Initially, the inclusion of items for further analyses is investigated at country level and overtime. The common items are to define the overall measurements and their structure is validated. Based on the Confirmatory factor analyses results, their psychometric properties are assessed. To demonstrate the implementation of the suggested methodology and facilitate practical applications, we use the human values measurements included in the European Social Survey questionnaire for Southern Europe, 2002-2018. Moreover, in order to show how these measurements may be used in further analyses, their association to subjective life satisfaction, happiness and general health are also presented.
Filter questions are used to administer follow-up questions to eligible respondents while allowing respondents who are not eligible to skip those questions. Filter questions can be asked in either the interleafed or the grouped formats. In the interleafed format, the follow-ups are asked immediately after the filter question; in the grouped format, follow-ups are asked after the filter question block. Underreporting can occur in the interleafed format due to respondents’ desire to reduce the burden of the survey. This phenomenon is called motivated misreporting. Because smartphone surveys are more burdensome than web surveys completed on a computer or laptop, due to the smaller screen size, longer page loading times, and more distraction, we expect that motivated misreporting is more pronounced on smartphones. Furthermore, we expect that misreporting occurs not only in the filter questions themselves but also extends to data quality in the follow-up questions. We randomly assigned 3,517 respondents of a German online access panel to either the PC or the smartphone. Our results show that while both PC and smartphone respondents trigger fewer filter questions in the interleafed format than the grouped format, we did not find differences between PC and smartphone respondents regarding the number of triggered filter questions. However, smartphone respondents provide lower data quality in the follow-up questions, especially in the grouped format. We conclude with recommendations for web survey designers who intend to incorporate smartphone respondents in their surveys.
This work proposes the use of a new multidimensional indicator to measure precariousness among young workers across all EU28 countries between 2009 and 2016, the aftermath of the economic crisis. This indicator measures both the incidence and intensity of precariousness and the contribution of each dimension included in the analysis. The analysis has involved five dimensions: wages, type of contract, type of working day, disempowerment, and job insecurity. Our database is the European Union Labour Force Survey for the period 2009-2016. The main indication of precariousness among young workers is low wages. In terms of the adjusted multidimensional precariousness rate proposed, we find high rates of precariousness for Mediterranean countries (because of low wages and temporary contracts), Denmark (low wages), and the Netherlands (expansion of involuntary part-time jobs). Central European countries have moderate rates, and most Continental and Eastern countries have low rates. We also find that a higher level of education is related to a lower probability of having a precarious job and that the probability of having a precarious job decreases as the number of hours worked increases. Finally, we find a greater probability of having a precarious among women in most countries, non-statistically significant differences by country of birth and a greater probability for younger individuals.
In an era of increasing number and complexity of online surveys, one of the methods to reduce cognitive load and increase the involvement of participants is gamification. Existing studies of gamification effect are limited to marketing research, while academic studies were mostly conducted on schoolchildren. This study is aimed to check whether the use of gamification elements is an effective method for reducing the cognitive load of the respondent measured by pupil diameter dynamics and paradata among respondents aged 18-25 years. And second target is to test multimodal approach to evaluate cognitive load.
We conducted a laboratory experiment in which 124 students were randomly assigned to two groups with traditional and gamified web survey design. We used emojis for ordinal scales and motivating memes between sets of questions as gamification in line with previous research.
Multimodal approach to measure cognitive load included a combination of three types of measurement: (1) objective indicators and paradata (item nonresponse rate, the length of the mouse cursor path, alerts); (2) subjective evaluation of survey burden (NASA-TLX); (3) neurophysiological indicators – pupil diameter dynamics measured by eye tracker.
According to subjective perception (via the NASA-TLX scale), a gamified survey is perceived as longer to complete. Two-way ANOVA comparing average cursor path showed the importance of the interaction factor of the group and the question – different questions are more cognitively demanding in two groups. The average cursor path is lower in the group with gamification. Traditional questionnaire design produced 4 times more warning alerts than gamified. Deviation of the pupil diameter from the baseline was higher in the gamified group which means gamification can increase cognitive load. Different modes of cognitive load assessment were not consistent. Although the results are somewhat inconclusive, we state that gamification has advantages over traditional web survey design among respondents aged 18-25.
With introduction of data protection regulations (e.g., GDPR) each survey collecting and analysing paradata must get a consent from respondents, but it is not clear how such consent and its format may influence the resulting survey data quality. The aim of the study is to assess how informed consent to the collection of paradata and its format can affect the quality of online survey data.
The research is based on vignette experiment with an intrasubject design, in which 461 web survey respondents were randomly assigned to 9 groups depending on the format of informed consent. We varied the following elements of the survey: presence of the consent, place of the consent in a survey (at the beginning, at the end), size of information regarding paradata (short/long), format of the answer to consent (Yes/No or one option confirmation). The survey was conducted using One Click Survey (1ka.si) web survey software which allowed to measure different types of paradata. We used the following indicators to compare data quality between the groups: paradata (mouse movements, browser focus, number of warnings, length of answers to open questions, completion time), test-retest reliability, straightlining, item nonresponse rate and dropout rate.
We found that presence of consent lead to lower completion times and less frequent satiations of leaving the page with the survey during survey completion process with less warnings presented to respondents. While longer information about paradata within consent produced longer answers to open-ended questions and less frequent satiations of leaving the page with the survey during survey completion process. Thus, it is indeed possible to state that the presence and format of consent to the collection of paradata in online surveys can influence the quality of resulting survey data by changing respondents’ level of involvement and attention to the survey completion process.
In recent decades, immigration in Spain has acquired great relevance. With regard to the integration of the immigrant population into the Spanish labour market, this is characterised by the differences in employment between this population group and the native population, which have generated significant inequalities between the two population groups, one of which is occupational segregation. This phenomenon plays a very important role as a source of discrimination, rigidity in labour markets, economic inefficiency and wage inequality.
This research analyses occupational segregation in the Spanish labour market between 2002-2022 from a double perspective, using microdata from the Labour Force Survey. First, an unidimensional analysis is performed taking into account separately segregation by gender and nationality. Secondly, a two-dimensional analysis is carried out considering both manifestations of occupational segregation together. For this purpose, the Karmel-MacLachlan index is used, which, in addition to allowing us to know how this phenomenon has varied over the study period, it has helped us determine which occupations have the greatest influence on it.
Furthermore, this analysis has great relevance in a current context marked by the COVID-19 pandemic, in which economies all over the world have been affected and the labour market has also undergone changes, which have not had the same impact on the native population and the migrant population, nor on men and women, nor with the same severity.
European institutions are increasingly assuming “The Data Governance” approach (European Commission, 2020), which is an organizational model whose core is the informative value of data. Data governance allows institutions to efficiently respond to citizens’ needs, through the Decision Support System: the usage of different types of data for planning public policies. This strategy represents the foundation of new relational structures between citizens and institutions, based on the public value of data (Martire, 2020).
These relational structures are inquired through the case study of Immuni, the contact tracing app adopted by the Italian Government to tackle the spread of Sars-CoV-2. The purpose of this study is to identify the profiles of the people who downloaded the app and those who did not. A structured questionnaire was administered to a probabilistic and representative sample of 800 Italian adult individuals, by using a mixed-mode approach: CATI, CAMI and CAWI. The interviewees were asked to specify whether they downloaded the app and the reasons supporting their choice.
Gender, age, area of residence, education and profession are decisive variables to cluster the 'downloaders' (those who installed Immuni), the ‘doubters’ (those who reported they would do it in the future), and the ‘opponents’ (those who decided not to do it). Distrust in institutions, the fear of sharing private data and the perception of the low efficiency of Immuni are the main reasons that prompted people not to download the app, which has not reached the download threshold (60% of the Italian population), needed to serve its purpose.
Immuni’s failure shows that the Data Governance approach requires that citizens manage the concept of data culture: the participation of citizens shape the awareness of the public utility of private data, through which the individual responsibility is shared with the rest of the community.
The work organization (WO) has played a key role during the COVID-19 pandemic due to its direct impact on workers’ health and living conditions. The COVID-19 experience has shown us that working conditions in a pandemic context can be severely worsened. Also, because certain “new” labour management practices, such as telework, can become key elements to address future pandemics, but at the same time can have detrimental effects in the working and living conditions, that can affect health and quality of life of the working population.
This project aims to establish a panel to obtain longitudinal data to evaluate the WO and its effects on health and living conditions from a gender perspective approach. At the same time, the data obtained will allow the establishment of a monitoring system of WO indicators, thus providing currently non-existent information that may be useful in guiding evidence-based public policies.
Data will be collected every six months by means of a survey, through the self-administration of an online questionnaire. The questionnaire has been thoroughly designed and validated with a group of experts and it has been piloted to assess its comprehensiveness and functionality. It comprises three main sections: 1) Sociodemographic, economic, and occupational characteristics; 2) WO (labour management practices and psychosocial risks); 3) Health (general health, mental health, drug consumption and absenteeism/presenteeism).
n=4000 workers aged over 16, residing in Spain, and who had been working in a salaried job during the week preceding the survey will be recruited from a list of contacts who had previously manifested their will to participate in upcoming studies, ensuring the labour market representativeness. The questionnaire will be filled online after giving consent. Participation will be encouraged by offering an individual report at the questionnaire’s completion, showing the harmful exposures to psychosocial risks and offering information about their mitigation.
A prerequisite for meaningful cross-national comparisons of scale scores is providing evidence of the scale’s measurement invariance. Measurement invariance is essential to ensure that the measurement instrument measures the same construct across all groups. Scaling theory presupposes that the structure of the construct (scale) is determined and its psychometric properties are assessed before their application. This involves splitting the sample of each country randomly into two halves and first performing Exploratory Factor Analysis (EFA) on one half-sample in order to assess the scale’s construct validity. Secondly, the structure is validated by carrying out Confirmatory Factor Analysis (CFA) on the second half-sample. Based on the full sample, the psychometric properties of the resulting scale are assessed. Then, multiple-groups CFA is performed for testing the scale’s across countries measurement invariance. As in any statistical analysis ─ whether univariate, bivariate or multivariate ─ the first and most important consideration is to ascertain the level of measurement of the input variables, in this instance the defining items of the scale. This guides the correct choice of the methods to be used. In this study, the investigation was based on a five-item scale measuring political trust to national institutions from the 2018 European Social Survey (ESS) questionnaire when items were considered as interval. Considering the factor analytic theory, only a unidimensional model could be tested. Therefore, only CFA was applied on the full samples of four European countries: France, Poland, Spain and the UK. Then, based on reliable and valid scales of acceptable or adequate model fit, measurement invariance was explored by performing multiple-groups CFA. Measurement invariance was tested for level of educational attainment groups of a combined sample of those four countries. Based on the mean scores, it is observed that the higher the level of educational attainment, the higher the political trust to national institutions, with the exception of Poland. This work could be extended to cover more countries and also explore overtime measurement invariance of the political trust scale as these items are included in all rounds of the
The global trend of population ageing draws attention to old age pensions. These pensions ensure the livelihood of the growing number of older individuals. At the same time, they need to be monitored and restructured to ensure the financial sustainability of pension schemes. Nowadays, research on pensions abounds. Yet, the self-employed challenge the scientific knowledge generated. The reason is that pensions from self-employment differ considerably from those from employment. While employees are covered by a mandatory pension insurance in many countries, the self-employed are often covered by voluntary pension insurances. While employees are covered by a single set of pension regulations in many countries, the self-employed may be split into sub-groups encountering sub-group specific pension regulations. While employees have a pension contribution that can be easily determined, the self-employed may have fluctuating incomes that misfit conventional ways of calculating pension contributions. Consequently, pensions from self-employment have to be studied separately from pensions from employment. When doing so, pension differences among the self-employed need to be considered. These differences vary across countries. Therefore, the pensions of the self-employed cannot be captured in a simple and straightforward variable – and even less so in country-comparative studies. Instead, a set of variables needs to capture the complexity of the pensions and it needs to be harmonized across countries. Previous research did not yet establish what such a set of variables looks like. The present study fills this gap in knowledge. It does this by comparing Germany and Finland – two countries representing most different cases when it comes to pensions from self-employment. Germany has a fragmented pension insurance system for the self-employed, leaving many of them with a voluntary insurance. In contrast, Finland has a largely uniform pension insurance system for the self-employed with largely mandatory coverage. The harmonization is achieved through an analysis of scientific texts and policy documents. Findings portray fissures in the pensions from self-employment, thereby contributing to our understanding of social inequalities. Moreover, they help researchers decide which groups of the self-employed can be directly compared across both countries, and which groups cannot.
The New Environmental Paradigm (NEP) scale, originally proposed by Dunlap et al. (2000), has been used in environmental sociology for decades and is by far the most frequently used measure of generalized environmental concern. NEP covers a wide range of environmental attitudes, including a coherent ecological belief system and ecological values (Xiao et al. 2019). Yet, questions of measurement invariance – cross-culturally as well as over time – have been raised and are yet to be studied. In this paper we use data from the German Gesis-Panel and estimate longitudinal Confirmatory Factor Analyses as well as Latent Growth Curve Models to assess measurement invariance over time. Preliminary results indicate that metric invariance can be established, but some items need to be excluded to achieve satisfactory fit (cross-sectional as well as over time).
Violence is exacerbated in small close-knit groups and through social identi cation processes. Relying on a new database on the relationships among US Islamist extremists, we explore the structure of this highly violent social network. Social identity theory postulates that individuals tend to relate together because of shared identities. We analyze whether social, cultural, and ideological identities increase the probability to make connections in Islamist extremist networks. We nd that the US Islamist network is highly clustered, modular, and includes only a few larger communities. Furthermore, results from the exponential random graph modeling (ERGM) point to the roles of friends, ideological a nity, and shared culture in driving connections between US Islamist extremists. Overall, these ndings suggest that different social identification processes are at play and that social identities shape violent social networks.
Unemployment has severely affected young people in Spain (and in other countries in Europe) during the latter years, especially since the last economic recession. The global pandemic context did not ease this situation, but it rather made it worse. Personal contacts are highly used in Europe by young job seekers, and even more in situations of job scarcity (economic recession periods). In this way, social capital has proven to be highly important for accessing a job, although there is an important methodological discussion about measuring used and accessed social capital.
The aim of this paper is analysing the labour market trajectory understood as sequential data. This allows us to actually consider the longitudinal component of the concept of ‘trajectory’. The instrument used to collect the (pre-pandemic) data was a hybrid survey (which includes the life history grid technique) in the Metropolitan Area of Barcelona. It provided information on the whole labour market trajectory of 150 individuals (Ego), on the different activities that define the sequence (Events, n=2.143) and on the social contacts (Alteri, n=637) that somehow helped them achieving labour events (n=1.211). Sequence analysis can be used to define a typology of sequences (clustering the computed distances from one sequence to another) and later used as another variable in further analysis, such as Correspondence Analysis, Event History Analysis or Multilevel Regression.
The unprecedented digitization and datafication of various aspects of social life in contemporary societies — such as friendship, love relationships, opinion formation, consumption habits and curriculum vitae… — has impacted social science in many ways in the last decade. Although the fear of an imminent crisis of empirical sociology diagnosed by Savage and Burrows in 2007 is probably behind us today, as more and more insightful research is conducted using Big Data in our discipline, the promises of a social science that, to use a classic formula of computational social science, can do research while we sleep (Salganik, 2019) have obviously not yet been fulfilled.In this presentation we will propose a reflexive return on a recent sociological investigation based on the collection, processing and analysis of data from LinkedIn in France, Brazil and the UK in order to describe and explain how professionals such as journalists make careers. Using such data seemed to us a very good idea given that journalism is a social worlds that is both too small to be well captured by official statistics and too fuzzy to be described only by administrative data provided by the group itself. Our aim is to show why — contrarily to Salganik's promise — we had in fact to remain alert and awake at all stages of this research. The methodological, epistemological and ethical implications of scraping data online in an interconnected world where the sociologist's jurisdiction is uncertain will particularly be discussed in our presentation, as well as the challenges of handling missing data, cleaning, standardizing and interpreting user entries on online platforms in a relevant way for social research.We will argue that Big Data and platform data do not necessarily require the development of very complex computational methods to become useful data for social science. Rather, due to the need to collect them from relational online worlds, they require the mastery — and reflexive use — of sociological skills based on an original combination of quantitative data analysis and participant observation.
The recent studies focusing on the positions expressed by Anti-vaxxer highlight the complexity of the reasons that guide the choice not to get vaccinated (individual risk tendency; consolidated attitudes within the medical care/prevention areas; propensity to conform to the others’ choices; cognitive shortcuts linked to the need to inquire about current events/to have external confirmations of personal beliefs). This contribution aims at discussing the results obtained through a panel web survey, carried out in two research appointments (Spring 2020/2021). The survey made it possible to reconstruct the dynamics of choice linked to the vaccination area, with particular regard to Anti-vaxxer. The recorded responses flows into three main social types: i. the negationist- trivializer, who does not believe in the severity of the disease/in the existence of COVID-19 and who questions the efficacy of vaccines; ii. the distrustful-fearful, who is worried about the possible repercussions of the vaccine on his-her health, in the short and long term; iii. the conformist, who approaches vaccination in an unconvincing way, by a forced and uninformed choice, sometimes because of a social desiderability mechanism. The longitudinal study, with respect to which it is appropriate to make a balance on a methodological level too, has also made it possible to identify the latent dimensions underlying the decision not to vaccinate, and has suggested the construction of synthetic indexes aimed at understanding any changes in status - on a cognitive, emotional, active level - between one survey and the next. These aspects (states of mind; information styles; evaluations of government actions; trust in expert knowledge) lend themselves to further study through a qualitative approach. This research step, planned for the coming months, involves conducting numerous in-depth interviews on the basis of a complex and reasoned case selection strategy.
While the public discourse in Europe revolves only around a few famous cases of ethnoregional mobilization (e.g., independence referendums in Scotland and Catalonia), ethnoregional parties as a party family have faced systematic growth also in less well-known regions throughout Western Europe. Following the pioneering work by Rokkan & Urwin (1983), Keating (1999) introduced a theory of a “New Regionalism” to explain the rise of this party family. This theory claims that the course of globalization lead regions to see themselves as competitors in a global economy rather than complementary parts of sate-wide economies, which in turn fuels ethnoregional sentiments in wealthier regions that no longer want their wealth to be redistributed.
However, despite an emphasis on “configurations” (which conditions need to coincide) of variables within theoretical and qualitative contributions – following Stein Rokkans methodological principles; quantitative models on the electoral success of ethnoregional parties are currently limited to investigating what Rokkan called the “hierarchy” of variables (which variable contributes the most to explaining variance).
To fully understand the rapid growth of ethnoregional parties throughout many Western European regions, it is therefore necessary to emphasize interactions between key variables (culture, economy, political opportunities) rather than adding additional predictors.
I tried to achieve this by adapting Hayes’ (2022) “conditional process analysis” to random effects panel regressions to analyze an original dataset covering all regional elections in Western Europe that took place between 1950 and 2018.
When controlling for cultural differences between regions (language and former statehood), there is apparent evidence that richer regions show higher vote shares achieved by ethnoregional parties. However, this effect only appears in regions with a medium degree of autonomy and only emerges over time. Reflecting the growth of this effect over time, this supports the idea of a competitive “New Regionalism” emerging in the course of globalization.
As a part of the obligatory evaluation at the Jagiellonian University in Cracow, Poland, there are regularly conducted the surveys of graduates (called “Alumni Career Paths Research”) implemented by Careers Service within the Centre for Academic Support. The tradition of such studies comes from 2008, when first attempts of measuring biographies of graduates were carried. The most important is to monitor the employment activity of graduates which is collected via online self-administrated survey, with a direct link sent to those who are registered on a pre-prepared list of e-mail addresses. The research is intended to study the entire population of one year’s graduates instead of selecting a sample. Every graduate is invited to take part in the survey six months after receiving their degree. However, only 3 years ago researchers were able to collect also phone numbers and private email addresses which help for increasing the response rates. Then relatively high response rates were observed in every edition of the survey (from 55% to even 75%). Besides of such challenges, as well the critical is to use the optimal time/period of sending invitation, remind to take part in the study, new ways of dissemination results and promotion of it among future graduates. How to effectively convince to participate in the study will be discussed in the presentation.
Conclusions from the conference,