ESRA 2019 Programme at a Glance
|Session Organiser|| Dr Caroline Roberts (University of Lausanne)
|Time||Wednesday 17th July, 15:00 - 16:30|
|Room||Aula (ground floor)|
If you are interested in presenting a poster at ESRA 2019, please submit your abstract to this session.
Effects of Web Survey Design on Data Quality: A Randomised Trial
Mr Knut-Petter Leinan (Regional Center for Child and Adolescent Mental Health, Eastern and Southern Norway) - Presenting Author
Mr Ole-Martin Vangen (Regional Center for Child and Adolescent Mental Health, Eastern and Southern Norway)
Dr Tore Wentzel-Larsen (Regional Center for Child and Adolescent Mental Health, Eastern and Southern Norway)
Dr Annette Jeneson (Regional Center for Child and Adolescent Mental Health, Eastern and Southern Norway)
In web surveys, as in other types of surveys, questions are typically presented in a grid or matrix, with multiple questions in a row and radio buttons to indicate response options in columns. Despite their widespread use, grids are associated with undesirable outcomes, such as a higher rate of missing data, straight-lining or non-differentiation, and higher breakoff rates compared to single items or grids containing fewer items. In addition, some studies indicate that respondents prefer a format with a single item per screen even when that format requires more time.
To compare the effects of two formats of web survey design on data quality.
The formats will be evaluated in a randomised controlled trial on the following outcomes: rate of missing data, level of differentiation, and break-off rate. We are also measuring the effects of these different layouts on time spent per questionnaire and their overall rating of their experience taking the survey. In the trial, the respondents are randomly assigned to a grid versus carousel (single question per screen) layout of questions. The experiment is conducted as part of a randomised trial including parents of young children in daycare centers across Norway.
Based on the responses so far, we expect approximately 2300 answered surveys, of which approximately 900 will be answered on a PC and randomly assigned to one of the two layout conditions. Results will be analysed winter 2019.
There is little data on the effect of layout of questions on data quality in our study population. The goal of this experiment, and related experiments, is to guide web survey design to yield good data quality as well as good user experiences.
Nonadherence in the Item Count Technique
Miss Beatriz Cobo Rodríguez (University of Granada) - Presenting Author
Mr Pier Francesco Perri (University of Calabria)
Mrs María del Mar Rueda García (University of Granada)
Nowadays, in many areas of applied research, surveys focus on sensitive or confidential aspects of survey participants.
Due to the sensitive nature of the investigated topics, the typical problem that arises is the social desirability, the tendency of the respondents to answer according to what is socially acceptable. For this reason, many interviewees refuse to participate in the survey or provide false answers. As a consequence, the reliability of the analyses, including the accuracy of the estimates, turns out to be altered in an important way.
The randomized response technique (RRT), by protecting the anonymity of the interviewees, is a possible solution for reducing nonsampling errors due to nonresponse and/or untruthful.
The RRT originated a huge literature and has been applied in many areas. Nonetheless, it suffers from difficulties and limitations, such as the reproducibility of the results, the correct understanding of the method to use to randomize the response, the increasing complexity of the questions. Consequently, other indirect questioning techniques (IQTs) emerged as an alternative to the RRT.
Anyway, although privacy protection is preserved and respondents’ cooperation enhanced, the IQTs are not immune to cheating in the sense that respondents might deliberately not follow the instructions prescribed by the techniques, or might release “self-protection” answers that avoid them being identified with a sensitive behavior. This drawback generated a series of new models with the aim of obtaining more reliable estimates. In this direction, we focus on the ICT and, by simulating different situations, we investigate the sensitivity of the estimator of the prevalence of a sensitive trait. In particular, using different probabilities for the sensitive and innocuous responses and a different number of innocuous items, we discuss the variation of the statistical power when introducing liars and nonadherence in the survey. We also analyze the impact of this setting on the distribution function of the estimator.
Consent to Data Linkage: A Meta-Analysis.
Miss Anne Elevelt (Utrecht University) - Presenting Author
Dr Vera Toepoel (Utrecht University)
Dr Peter Lugtig (Utrecht University)
Linking survey data to all kind of other data sources becomes increasingly important and the widespread adoption of digital technologies is even further expanding opportunities for survey researchers to enhance survey research. This can include all kind of data, but most researchers currently use sensor data (e.g. location, accelerometer), biomarker data (e.g. blood, saliva), social media data (e.g. Facebook, Twitter) or administrative records (e.g. employment data, health records, social security records).There are several advantages of linking data, but the main reasons are the decrease of respondent burden (by reducing the number of questions) and improving data quality (by collecting more accurate and detailed data than respondents can provide).
However, a crucial element of being able to do research successfully with linked datasets is to get respondents to consent in order to prevent non-consent bias. Ineffective ways of asking for consent can lead to low consent-rates, as various aspects of the consent question (wording, sponsor, placement, etc) seem to influence the consent rates.
We will conduct a meta-analysis, in which the results of multiple experimental studies on consent for data linkage are combined and re-analyzed. The main contribution of this paper is to systematically present the state of the art of asking for consent. Based on the results of this study, guidelines will be developed for researchers so that we can give an advice on effective ways to ask for consent to link survey data to other types of data.
Sunny Mood on a Sunny Day? - How Weather Influences Well-Being
Dr Corinna Frodermann (Institute for Employment Research (IAB) ) - Presenting Author
Empirical research in social and political science is fundamentally based on survey data. In these surveys respondents are often asked for assessment on different dimensions of satisfaction. This poster argues that subjective assessments of cognitive life satisfaction are strongly influenced by the affective state of the respondent on the day of the interview.
Previous research has shown that weather conditions affect a broad range of thoughts, feelings, and behaviors. The current study examines whether these effects extend to well-being judgments. In order to test the association between daily weather conditions and well-being empirically, the German Panel Study "Labor Market and Social Security" (PASS), which is established by the Institute for Employment Research, is linked to region-specific weather data. “Sunny” and “rainy” survey days are identified and their influence on people’s assessment on their life satisfaction and mental health is examined.
In PASS approximately 12.000 persons in more than 8.000 households were interviewed every year since 2006. Based on the panel data, not only the correlation between the daily survey weather and life satisfaction can be analyzed, but individual consistencies can also be investigated. These findings introduce interesting and important implications for scientific survey research.
Understanding Perspective on Causes and Effects of School Dropouts in West Java Indonesia Using Small Area Hierarchical Bayesian Neural Network
Dr Toni Toharudin (Universitas Padjadjaran) - Presenting Author
Dr Rezzy Eko Caraka (Institute of Mathematical Research, Universiti Putra Malaysia)
Many reasons behind a person are not / have never been to school or do not continue their education at a higher level. Just because of costs, marriage, or taking care of the household, distant school or infrastructure, not being accepted, disability and so on. The reason is that costs are usually related to poverty or economic difficulties. The reason schools are far related to the availability of a minimum number of schools or the geographical conditions of an area causes challenging to access. In the development of statistics, there are many methods to obtain data, including surveys. Surveys play an essential role in the decision-making process based on data. The survey itself is often conducted regularly both in private and public research institutions. The primary purpose of the survey is to get information about the parameters of the population by making effective the available costs. So far, regular surveys conducted by the government of a country have only been designed to obtain information on national scale data. Problems arise when from surveys such as this one want to collect information for smaller areas, such as provincial, regency/city level information and perhaps sub-district and village levels. In this study, we used small area hierarchical bayesian neural network in estimating regency/city dropout rates in Bandung, West Java which can be used as an alternative to mapping areas that have high dropout rates
Modelling the Generation of Life Courses
Miss Yu Han (Netherlands Interdisciplinary Demographic Institute) - Presenting Author
Professor Cees Elzinga (VU vrij universiteit Amsterdam)
Abstract Life course research has been dominated by methods and models that focus on the description of life course patterns and on the causal patterns between agency- and structure-related variables on the one hand and, on the other hand, outcomes in later life. Little attention has been paid to modelling the driving force, the mechanism, that generates the chain of successive events and stages of the life course: the sequences of individual decisions pertaining to all facets of the life course. This paper presents the minimal requirements that models should satisfy in order to be considered as life course generating models. The paper then proposes Hidden Markov Models as one of the main building blocks life course generating models and discusses a few applications of these models in the domains of family formation, school-to-work transition and their interaction.
Re-Identification Risk Evaluation in Implementing Privacy-Preserving Analytics
Miss Sapphire Han (Statistics Netherlands) - Presenting Author
Dr Peter-Paul de Wolf (Statistics Netherlands)
Advanced IT technologies such as blockchain and secure multi-party computation make complex data-driven analytics such as privacy-preserving analytics (PPA) possible. Researches on preserving data privacy without affecting data utility and model learning are emerging. PPA can enable parties having conflict interests achieving data analytics without sharing data from these parties. These analytics can analyze personal data at an unprecedented scale and depth. However, due to the recent General Data Protection Regulation (GDPR), PAA in Europe must comply with data protection rules. Therefore, PAA bring both opportunities as well as challenges for statistical disclosure control (SDC). When adopting proper encryption procedures in PAA, data utility can be largely enhanced without traditional SDC techniques such as cell suppression and aggregation. At the same time, these encryption procedures in PAA must demonstrate robustness against re-identification attacks from both inside the involved parties as well as outside intruders.
Previous work done by our colleagues demonstrated the application of blockchain and secure multi-party computation help to achieve data queries to improve patient care using data from a national statistical office, patient data from a hospital and insurance data from a health insurance provider with PAA. This study aims to examine the extent to which analytics such as statistical summarization, supervised learning and unsupervised learning influencing re-identification risks in PAA procedures. For this real world testing, we matched socio-economic status (SES) data from Statistics Netherlands, with health record from a hospital and medical cost declaration from a health insurance provider. Re-identification risks of different analytics will be discussed considering inside and outside intruders.
Contact via Dating Apps - Comparison of Two Ways of Recruitment of Men Who Have Sex With Men
Mr Stefan Albrecht (Robert Koch Institue) - Presenting Author
Dr Ulrich Marcus (Robert Koch Institute)
Dr Klaus Jansen (Robert Koch Institute)
Dr Barbara Gunsenheimer-Bartmeyer (Robert Koch Institute)
Dr Viviane Bremer (Robert Koch Institute)
Dr Uwe Koppe (Robert Koch Institute)
Background: HIV pre-exposure prophylaxis (PrEP) is a method to prevent HIV-infection and is currently not covered by health insurances in Germany. Therefore it is only available through private prescriptions and non-prescription sources. We conducted a survey about the use and the sources of PrEP in German PrEP users. The following recruiting strategies were used: dating apps for men who have sex with men, a PrEP community website, anonymous community testing checkpoints, and peer-to-peer recruitment through friends. In this analysis we investigate the effects of different recruitment methods.
Methods: Data was collected with an online-survey in 2018. Participants were asked to identify through which sources they were recruited to our survey. To evaluate the different recruitment strategies and their outcome on sample composition, we used Chi-square tests and uni- and multivariate logistic regression.
Results: Of 4,018 people that consented to participate, 1,745 (43.4%) respondents completed the survey. Most people were recruited solely via Dating Apps (1,385; 79.4%), a significantly smaller part solely via one of the other options (296; 16.7%) and very few people stated to have been informed via both ways (64, 3.7%). Respondents that have been recruited via Dating Apps are statistically significantly younger (mean 37.5; 39.6 years) and more likely to take PrEP intermittently or not anymore (33.7%) than respondents recruited by one of the other ways (24.1%). Meanwhile they do not differ significantly in their income and the number of sex partners in the last six months. When controlled for age and country of origin in a multivariate analysis (logistic model) the effect of different PrEP usage prevalence and age still are statistically significant.
The different ways of recruitment result in different sample compositions. It is therefore highly recommended to use different recruitment strategies in these kinds of studies.
Pre-Checked Forced-Choice Questions as an Alternative to Check-all-that-Apply
Dr Naoki Tomita (Department of Geriatrics and Gerontology, Institute of Development, Aging and Cancer, Tohoku University) - Presenting Author
Dr Takahiro Tsuchiya (Yokohama City University)
When asking respondents whether each item in a list is applicable to them, questionnaire designers apply two types of question formats: check-all-that-apply and forced-choice questions. In the former format, the respondents are requested to select an unspecified number of applicable items; in the latter format, they are asked to select whether each item "applies" or "does not apply." The check-all-that-apply question imposes less burden on respondents than the forced-choice question, but has several drawbacks: non-selection of the item does not necessarily mean "does not apply," and the question often produces lower response proportions than the forced-choice question.
The present paper proposes a new question format named pre-checked forced-choice for surveys conducted on computers, in which all items are pre-checked with "applies" or "does not apply," and the respondents are requested to change the checks if the pre-checked options are inappropriate. The proposed method combines the advantages of both check-all-that-apply and forced-choice questions: it lessens the burden on respondents while providing data that can be selected as "applies" or "does not apply."
In a comparative survey, the proportions yielded by the proposed pre-checked forced-choice questions were higher than those of check-all-that-apply questions, but lower than those of traditional forced-choice questions. However, when the proportions of the "no-to-yes" and "yes-to-no" types of pre-checked forced-choice questions were combined, they were comparable to those of the forced-choice question, although the questions incurred less burden on the respondents.
Conceptualizing the Potential Visitors: The Distinctiveness and Variety of Respondents of Visitor Surveys in Comparison with Nationally Representative Surveys
Dr Naoko Kato-Nitta (The Institute of Statistical Mathematics) - Presenting Author
Dr Tadahiko Maeda (The Institute of Statistical Mathematics)
This study aims to conceptualize the potential visitors to scientific attractions as the target population by examining the attitudes and behavior of the respondents of visitor surveys at open house events at a scientific research institution in comparison with nationally representative surveys. Given increasing global competition, raising public support for scientific research has become an important focus among many countries. The Japanese government also have promoted scientific organizations and universities to disseminate scientific information to the visitors by arranging public outreach activities. Nevertheless, those who do not participate in such events will not receive the information. There are various reasons for non-participation, but if the findings from this study provide insights into the potential visitors, we may have a key to reach out to the wider public and present a step forward in methodology on visitor studies as well as sociology of science.
For this purpose, we propose an approach of statistically analyzing the contrasting data from different groups when a sample from the target population is not available. We utilized a simplified model that distinguish the levels of survey population: general public, potential visitors (target population), visitors, and participants (highly engaged visitors who responded questionnaires). Then we drew data from the three out of the above four levels of survey population and discussed the distinctiveness and similarity among them: general public (questionnaire items of nationally representative surveys), visitors (behavior log data from visitor surveys), and participants (visitor questionnaire items identical to nationally representative surveys). The statistical analyses based on cultural capital and value toward various aspects of culture as lenses showed that there were not only differences in terms of attitudes and behavior between visitors to the scientific exhibitions and the general public but also differences between visitors who participated by completing the questionnaire and those who did not. The implications derived from the findings will be discussed.
Repeated Cross-Sectional Surveys on Japanese National Character: Results of Nationwide Surveys from 1953 to 2018
Mr Tadahiko Maeda (The Institute of Statistical Mathematics) - Presenting Author
Dr Takashi Nakamura (The Institute of Statistical Mathematics)
Dr Yoosung Park (The Institute of Statistical Mathematics)
Dr Yusuke Inagaki (The Institute of Statistical Mathematics)
The surveys on the Japanese National Character is a repeated survey project by the Institute of Statistical Mathematics since 1953. To be precise, it is a series of cross-sectional surveys, in which nationwide survey has been conducted every five years based on probability samples selected for each occasion, by the same survey mode (face-to-face interviewing), and with basically the same questions. This project has the longest history among longitudinal nationwide surveys in Japan, and has maintained three basic aims: (1) Elucidation of the Japanese "national character", (2) Research on survey methodology, and (3) Development of techniques of statistical analysis. Most recent is the fourteenth survey carried out in the autumn of 2018.
The questionnaire consists of items covering broad topics in daily life, which intend to measure values, attitudes and opinions that might characterize “Japanese ways of thinking”. Since the fifth survey carried out in 1973, we have used two types of questionnaires administered to half of the sample respectively. This presentation intends to introduce the outline of the project, by exemplifying a few typical items which record long-term changes of Japanese opinions, and emphasizes the importance of continuity of the project. While the most dominant trend for more than sixty years of survey history is the decline of the traditional values on several aspects, some items showed interesting “swing-over phenomena” during the period. One of the advantages of this type of repeated surveys is that we can examine the age, period, and cohort effects on each item. We will also demonstrate a few typical results of the Bayesian logit cohort analysis, which was developed by the one of the authors.
What Predicts the Validity of Self-Reported Paradata? Results from the German HISBUS Online Access Panel
Dr Nadin Kastirke (German Centre for Higher Education Research and Science Studies (DZHW))
Mrs Swetlana Sudheimer (German Centre for Higher Education Research and Science Studies (DZHW)) - Presenting Author
Mrs Gritt Fehring (German Centre for Higher Education Research and Science Studies (DZHW))
Mrs Ulrike Schwabe (German Centre for Higher Education Research and Science Studies (DZHW))
Paradata, such as user agent strings (UAS), provide us with client-side information about the technical conditions of web surveys. If UAS are not available, one may directly ask survey participants for the required data. However, it is unclear whether the validity of self-reported paradata is determined by participants’ general attitudes towards surveys, their willingness to participate and – regarding technically demanding questions – distraction while answering.
To find out what predicts the consistency between UAS and self-reported paradata, we used the HISBUS Online Access Panel. The sample comprised 3.137 members with UAS known to us that were asked for used device (DEV), operating system (OS) and web browser (WB). Additionally, data on survey attitude (SAS; de Leeuw et al. 2010), survey participation evaluation (SPE; Struminskaya et al. 2015) and multitasking (MT; Zwarun/Hall 2014) were collected. The Big Five personality traits (BF; Rammstedt et al. 2013) served as covariates in our applied logistic and ordinal regression analyses for DEV as well as OS/WB, respectively. Predictors with p<.05 were included in the final models.
UAS and self-reported paradata were highly consistent (kappa: DEV=.95, OS=.95, WB=.87). Agreement regarding DEV depends on SAS subscale value (OR=1.31) and SPE (OR=0.80). The OS/WB agreement was predicted by electronic and non-electronic MT (OR=1.32; OR=0.72).
Directly asking web survey participants is a promising way to get valid information about their technical equipment if UAS are not available. The chance to get valid DEV data is higher if surveys are generally considered valuable, but lower if the evaluation of willingness to participate is considered solid. The chance for valid OS/WB data is higher with electronic MT present that may indicate technical skills. Non-electronic MT seems to be rather distracting and predicts lower chances for valid OS/WB data.
Research Support and Training Offer for Social Scientists. Activities of CESSDA
Mrs Irena Vipavc Brvar (ADP - Slovenian Social Science Data Archives) - Presenting Author
Mrs Ellen Leenarts (DANS - Data Archiving and Networked Services)
Dr Peter Doorn (DANS - Data Archiving and Networked Services)
The Consortium of European Social Science Data Archives (CESSDA) is an ERIC in the field of social sciences. CESSDA currently has 17 members and 1 observer and continues to grow. One of the focus areas within the CESSDA ERIC is providing training and knowledge sharing among social science data archives. In addition, CESSDA also provides training to future depositors of data and to (social) scientists who are looking for data to reuse. In this poster, we will give an overview of the CESSDA training offer. Specifically, we will provide information on the Research Data Management (RDM) and data discovery workshops that CESSDA is organizing. In addition, a quick introduction will be given to the online CESSDA Data Management Expert Guide (https://www.cessda.eu/dmeg) and we will present an overview of our webinar offers and train the trainer events. The workshops for researchers which are organised by CESSDA members typically focus on general issues concerning RDM, but CESSDA also provides information about specific topics such as sensitive data, metadata and documentation, preparing data for depositing, guidelines on licensing and citing data, and privacy (GDPR-compliant) in practice. For archivists within CESSDA, we will organise training days on curation of data, controlled vocabularies, metadata and tools. Last but not least there are thematic webinars organized for researchers on data discovery, focusing for instance on data on migration and on poverty.
The Gender Gap and the Costs of Voting: A Comparative Analysis
Professor Susana Aguilar (Universidad Complutense de Madrid) - Presenting Author
Dr Andrés Santana (Universidad Autónoma de Madrid)
This paper focuses on the effect of gender upon the costs of voting (C), the latter being one of the key determinants in the calculus of voting equation. With the aid of comparative data from the Making Electoral Democracy Work (MEDW) project, which includes the same explicit measure of C for Switzerland, France, Germany, Canada and Spain, it unveils that there is a gender gap whereby women have consistently higher C than men. This gender gap holds for the whole sample and also for each individual country. It also holds after controlling for a wealth of variables that may affect C, and is robust to several changes in the specification of the models. This result poses important analytical, political and normative questions, as well as challenges for public policies aimed at redressing inequality in the political realm.
The Effects of Response Burden When Collecting Life-History Data in a Self-Administered Mixed-Device Survey
Mrs Fine Cordua (German Centre for Higher Education Research and Science Studies (DZHW)) - Presenting Author
Mr Johann Carstensen (German Centre for Higher Education Research and Science Studies (DZHW))
Response burden, which may be defined as the effort required to respond to a survey, is suspected to be a major source of different kinds of error in survey data. As a result of high response burden, respondents may become frustrated, which finds expression in a number of adverse effects such as poor data quality due to satisficing, dropout from the survey, or panel attrition.
This is especially important if the information of interest is complex and detail-rich as is the case for life-history data. In interviewer-administered survey modes, interviewers are able to reduce the impact of response burden by providing help and motivation. Collecting life-history data in self-administered surveys, however, offers little to no such possibilities. Hence, focussing on this aspect is eminently important in the presence of self-administered data collection.
We argue that response burden is not only driven by questionnaire length but also by the complexity and difficulty of the recollection process, determining the cognitive load to answer the questionnaire. When collecting life-history data, the richness of details asked for every event, and the time span for which information is requested, are core factors in reducing response burden.
We conducted an experimental variation of response burden in a self-administered mixed-device survey, aiming to collect life-history data among secondary school leavers, undergraduates and university graduates. Therefore we varied the retrospective reference period between two and five years. Additionally, we also varied the amount of information collected on each episode. The poster reports results concerning the effect of these variations on dropout, data quality and willingness to participate in further surveys, considering the usage of mobile and non-mobile devices.
Family Formation as a Sensitive Topic: Do Gender and Social Norms Matter?
Dr Irina Gewinner (Leibniz Universität Hannover) - Presenting Author
Family formation represents a methodological challenge to investigate in a twofold sense. On one hand, people might face health problems and obstacles to conceive. On the other hand, they might be convinced childfree representatives and renounce having children. While in some populations, this subject is deemed a usual topic of discussion, in other cultural environments people react with refusal and/or shame and often give untrue answers in order not to be judged in a negative way. This might especially be the case in societies that communicate motherhood as a social norm and a primary role of women and expose childless women to public shaming. A negative connotation of childlessness on a macro level of a single society may, thus, result in secondary practices of individuals, such as interview refusal rate or deliberately untrue answers pertinent to family formation and a desire to have children.
These peculiarities described above, can be detected not only in respective countries themselves, but also in the answers of migrants who originate from these countries. This statement might hold true for both low and highly skilled, since motherhood is a gender role attitude and is, therefore, rather stable over time. Migrant family formation is of particular interest for several reasons: (1) Itruns simultaneously to the stage of professional establishment. (2) Mainly individuals in highly skilled professions show large gaps between desired and actual numbers of children. (3) The societal, cultural and political context is crucial.
This contribution aims at discussing challenges in studying family formation behaviour as a sensitive topic for Russian-speaking migrants by showcasing the German case. This contribution aims at discussing challenges in studying family formation behaviour as a sensitive topic for Russian-speaking migrants by showcasing the German case. I outline barriers in interviewing migrants on this sensitive topic and describe experiences from face-to-face vs telephone interviews.
Event History Calendar: Are There Positive Effects in a Self-Administered Mixed-Device Survey, Too?
Dr Johann Carstensen (German Centre for Higher Education Research and Science Studies) - Presenting Author
Mr Sebastian Lang (German Centre for Higher Education Research and Science Studies)
Research questions in the area of the sociology of education often require life-history data to address issues of causality, e.g. regarding social inequality in education or educational returns. Retrospective collection of life-history data is always highly demanding for the respondents and therefore prone to error since respondents have to retrieve and provide complex and extensive data. Thus, such data is usually collected through interviewer-administered survey modes, where the interviewer is able to provide help or motivation. Although interviewer-administered surveys are very costly, so far there have been very little attempts to collect life-history data using a more cost-effective self-administered survey. We conducted a study to evaluate the possibilities of collecting life-history data in an online mixed-device survey.
Following a life course perspective highlights the complexity of individual biographies and the interconnectedness of different biographical spheres. As a standard questionnaire does not fit the structure of autobiographical memory well, it is supposed that cognitive processes can be facilitated and response burden reduced by using an event history calendar, thereby improving the efficiency and usability of the questionnaire. It is, however, not only unclear if this method provides the desired effects in a self-administered survey mode, but especially in a mixed-device survey, permitting the completion of the questionnaire on a mobile device.
We developed a questionnaire to collect life-history data especially on educational and occupational trajectories among secondary school graduates, undergraduates and university graduates. We conducted an experimental variation of the use of a responsive event history calendar prior to detailed questions about event-episodes in a self-administered mixed-device survey. The poster reports results concerning the effect of using an event history calendar on dropout, data quality and willingness to participate in further interviews, considering the usage of mobile and non-mobile devices.
Italian Doctorate Holders in Political and Social Sciences: Results from Two National Surveys
Professor Alessandra Decataldo (University of Milano-Bicocca)
Professor Brunella Fiore (University of Milano-Bicocca) - Presenting Author
We develop a secondary analysis of the data provided by the Italian National Statistical Institute (Istat) by means of two surveys on the PhD graduates who obtained their doctoral degree from Italian universities in 2004 and 2006 (the first survey, 12,964 interviews completed); and in 2008 and 2010 (the second survey, 16,322 interviews completed).We focused on Doctorate holders in Political and Social sciences. After a descriptive social make-up of the four cohorts, we carried out an analysis to verify the mid-term occupational outcomes of PhD graduates in Political and Social Science in a period of academic reforms and economic crisis in Italy. The process of merging of two databases was complex due to a different wording of the questionnare and consequent definition of some variables. A process of harmonization of some variables was necessary, for instance, for the employment condition and for the amount of monthly salary our analysis underlines relevant differences between the two cohorts graduated before the economic crisis and the other two. One one hand, the latter has more chance in obtaining fixed-term contracts and low salaries; on the other hand, PhD graduated in the second wave experiment more frequently the chance/risk of being non-academic researchers, self-employed or non-researcher. The economic crisis fragmented the contracts’ scenario increasing, in particular, fixed-term contracts, unemployment and, furthermore, polarizing the salaries (very high salary or very low salary). We find an increasing chance in having non-academic career after the economic crisis; further studies are necessary in order to understand if this new condition is a costrain or an opportunity for PhDs in Political and Social Science.
Web Archiving and Social Sciences: Challenges, Problems and Solutions
Mr Matous Pilnacek (Institute of Sociology of the Czech Academy of Sciences) - Presenting Author
Mrs Paulina Tabery (Institute of Sociology of the Czech Academy of Sciences)
Research problem: Over the past 25 years the Internet and the World Wide Web have completely transformed the way people communicate. Many people and institutions now communicate via the internet and the reasons why they communicate in this environment are very different. This means that there is considerable variability and volatility in the content of communication and raises the question of whether and how this type of content can be maintained. Worldwide initiatives for archiving the web have begun, and the next step is to reflect on the use of this archived data by social scientists (Brügger & Schroeder 2017).
Objectives: The aim of our poster is to present data archived in the Czech Webarchiv, which collects and preserves Czech web resources. We will outline possible ways of working with the data and draw attention to the legal and ethical limitations and possible solutions. We will also introduce our work on the software that is being developed in collaboration of social scientists, archivists, and IT specialists in three partner organisations: the Institute of Sociology of the Czech Academy of Sciences, the National Library of the Czech Republic, and the Department of Cybernetics, Faculty of Applied Sciences, University of West Bohemia. This software should make data of this type accessible to social scientists to process and analyse.
LGBT in Northern Ireland: Equality, Attitudes and Policy
Dr Paula Devine (ARK, Queen's University Belfast) - Presenting Author
Dr Martina McKnight (ARK, Queen's University Belfast)
The 1998 Good Friday/Belfast Agreement was an unsteady political settlement, but meant that Northern Ireland gained the strongest equality legislation in the United Kingdom (UK). Under Section 75 of that Agreement, statutory bodies are required to promote equality of opportunity for people within nine defined groups, one being sexual orientation. However, twenty years later, Northern Ireland is the only UK region where same-sex marriage is not legal, and attempts to legalise it have been vetoed by politicians. Furthermore, in a 2015 referendum, voters in the Republic of Ireland (a country historically regarded as socially conservative) voted in support of same-sex marriage.
McAlister, Carr and Neill (2014) argue that the religious and moral discourse that is often reflected in public and policy debates in Northern Ireland has consistently impacted on the implementation of policy and legislative change in relation to family, sex and sexuality.
Since 1998, the Northern Ireland Life and Times (NILT) survey (www.ark.ac.uk/nilt) has recorded public attitudes to key social policy issues. In 2013, this cross-sectional, annual survey included questions on LGB issues to provide a base-line of opinion before the publication of a proposed Sexual Orientation Strategy in that year. The survey findings provide evidence of public support for same-sex marriage, and, while the data was used extensively by lobbying groups to call for legislative change it has not been forthcoming. However, Northern Ireland’s government has been suspended since January 2017.
The 2018 NILT survey revisited the same-sex marriage. It also explored attitudes to transgender people, which have been the subject of recent public and policy debate. Drawing on 2013 and 2018 data, this paper discusses if and/or how attitudes have changed over time, and whether government policy is in tune with public opinion.
Students' Health Monitoring at Saint-Petersburg State University: E-Day Method as a New Way of Data Collection
Mr Andrey Govorov (Saint-Petersburg State University ) - Presenting Author
Mrs Anastasiia Aleksandrova (Saint-Petersburg State University )
Dr Maia Rusakova (Saint-Petersburg State University )
Every four years Saint-Petersburg State University conducts the monitoring of students’ health by circulating questionnaires among them. The first survey was held in 2002. In 2018, the fifth wave of monitoring was conducted. The studies were done in different ways: the first one used a paper questionnaire, and the last one – an online form.
The main task of modern surveys is to find the most optimal form of data collection, which would improve the quality of obtained data by inducing students to be open and sincere when answering questions on the sensitive topics of health.
For this purpose, by the year 2018 we had developed author's method of data collection (the working title – ‘E-day method’), which we tested at the University. The E-day method resembles those applied at political elections: for a period of time data collection polls are opened at each department. To assess the potential of the E-day method for researching closed communities and sensitive topics, we conducted expert interviews with methodologists and focus groups with students. We discovered that E-day method can improve the quality of received information, as it allows to:
1. Monitor the formation of sampling by registering of those who come;
2. Preserve the anonymity of respondents, voluntary participation;
3. Increase the honesty of answers to sensitive questions, as the interactive and closed format of the event creates a comfortable environment for completing the questionnaire;
4. Provide for the flexibility and transparency of data collection, as participants can express their suggestions directly at the information Desk;
5. Increase the involvement of respondents through the opportunity to participate in an unique event.
Moreover, the E-day method assumes students’ participation in the organization of data collection polls. Such format inspires students to trust the system, as they become part of it, but also involves students in practice-oriented activities.
The Effect of Existence of Interviewers: Comparative Analyses of CAPI to CASI and CAPI to Web
Dr Aki Kaeriyama (Gunma Prefectural Women's University) - Presenting Author
Dr Daisuke Kobayashi (Kanazawa University)
Mr Isamu Sugino (Ochanomizu University)
There has been a great discussion about the survey mode difference between interviewer-administered and self-administered questionnaire. However, little study has been done in Japan, because of the absence of the dataset whose data is obtained through experimental design. In this point, Kaeriyama et al. (2015) is one of the few studies to investigate the effects of the modes of data collection on the data quality, using the survey data from an experimental design where probability samples were randomly assigned to three modes, PAPI, CAPI and CASI(hereafter “three modes survey”). Compared the means of attitudes and behaviors among modes using t-tests, they show that there was no significant difference in the comparison of PAPI and CAPI and that there were significant differences in some attitudes and behaviors in the comparison of CAPI and CASI. These outcomes suggest that the presence of interviewer has influenced the respondents’ answers also in computer-assisted modes. Nevertheless, it should be noted that the differences between CAPI and CASI were not large as they assumed. Why so?
One possibility is as follows: respondents were influenced to no small degree by interviewers also in CASI in this survey, since interviewers waited standing nearby the respondents during their answering the questionnaire. The present study was undertaken in order to examine this possibility, comparing the three modes survey with the CAWI dataset randomly sampled from same population as three modes survey in 2015 and examining whether the difference between CAWI and CAPI is larger than the difference between CASI and CAPI.
Analysis on Survey Data Quality and Respondents' Attitude toward Surveys: A Case Study with 5 Nationwide Surveys on Japanese National Character
Professor Yoosung Park (The Institute of Statistical Mathematics) - Presenting Author
Professor Tadahiko Maeda (The Institute of Statistical Mathematics)
The extent of nonresponse is an important factor for the quality of survey data. Although many studies have examined those factors relevant for the decision to take part in face-to-face doorstep interviews, it is unclear why respondents to decide to answer ‘don’t know’ responses. According to Beatty and Herrmann (2002), the respondents’ decision to answer a survey question is assumed to depend on three factors; the difficulty of questions, the sensitivity of questions, and the motivation for answering questions. However, the respondents’ the cognitive extent of three factors might be not always consistent, varying substantially depending on the unit of analysis with respect to change of the times. Therefore, this study analyzes the relationship between the respondents’ attitudes toward surveys and item nonresponse using the repeated survey data on the Japanese National Character.
The Surveys on the Japanese National Character have been carried out every five years since1953 with face-to-face doorstep interviews. Specifically, we will analyze most recent five surveys since 1998, which include a few items about respondents’ attitude toward surveys, focusing on the relationship the number of “Don’t Know” answers and respondents’ characteristics and attitudes toward surveys.
Changes in New Parents’ Employment Trajectories. Institutional Reform or Change in Social Norms?
Dr Ariane Bertogg (University of Konstanz) - Presenting Author
Ms Anna-Theresa Saile (University of Konstanz)
The recent decades have seen a rising share of mothers participating in the labor market in Germany. Two developments at the macro-level accompany this development: institutional reforms in the domain of parental leave on the one hand, change in gender norms on the other hand. For the former, Germany has experienced a comprehensive reform of the parental leave legislation in 2007. We argue that this has set incentives for women to return to the labor market more quickly while encouraging fathers to take parental leave. Second, as gender norms have become less strict, couples’ division of labor may have become more equal. Here, we argue that weakening gender norms enable particularly couples with egalitarian attitudes to choose a model according to their preferences.
We draw on data from the German Socio-Economic Panel (GSOEP), using observations from 1991 until 2015. First, we identify five typical groups of new parents’ employment trajectories by means of sequence and cluster analysis. As our results show, egalitarian patterns have become more frequent as the “classic” male breadwinner pattern has lost ground. In a second step, we trace these changes back to the 2007 reform and changes in gender norms. To measure the latter, we draw on data from the German General Social Survey (Allbus 1991-2012), using the aggregated percentage to the statement that “mothers should stay home to care for the children and fathers should work to provide for the family”. The lower the agreement to this statement in the year their first child was born, the more likely couples are to choose an egalitarian pattern over the “classic” male breadwinner model. Implementing interaction effects, we are able to show that the 2007 parental leave reform has sped up the effect of loosening gender norms, which points at the effects of institutions and norms being intertwined.
Analyzing the Impact of School Routes and Educational Decisions - Challenges of Data Protection in Linking Survey and Geospatial Data
Dr Michaela Sixt (LIfBi Leibniz Institute for Educational Trajectories) - Presenting Author
Not only since the adoption of the General Data Protection Regulation (GDPR) contact and address data of survey participants are under special protection. They had to be stored separate from survey data and there were further limitations to access before. Especially if they should be combined with other addresses, e.g. the addresses of schools, further data protection mechanisms must be considered. The paper at hand presents a data protection concept realized for a project in the German context. In this project the impact of infrastructure (in the sense of the distance to offerings) on educational decisions is investigated. To answer this question, distances are generated (offline) between survey participants’ geocoded residence address and the geocoded address of the relevant institution. Furthermore the additional knowledge gain when using distances instead of or additional to aggregated information will be shown.
Surveying Mental Health in the Military: Definitely a Sensitive Topic! Definitely?
Mr Wolfgang H Prinz (National Defence Academy) - Presenting Author
Mr Stefan Rakowsky (National Defence Academy)
Mr Stefan Lackinger (National Defence Academy)
Mr Stephan Heisig (National Defence Academy)
Mental health in the military has become an issue of growing concern in societies and research. However, investigating military health is associated with major challenges. Screening instruments, for an example, which have been determined to be reliable and valid in civilian populations, frequently lack psychometric properties in the military. One reason for this phenomenon can be found in military culture/socialization, which can promote stigma and therefore lead to social desirability (SD) biases.
In the Austrian Armed Forces, mental health is monitored in the context of the annual postal "Internal and Social State"-Survey. In this recent study, we re-analyzed data which had been obtained in 2017 from 1.046 professional soldiers and 442 civilian employees to investigate sensitivity and possible SD-biases effecting the validity of our data on mental health.
We hypothesized that (1) questions about mental health would be more sensitive than items on general working conditions and (2) personnel status (professional soldier vs. civilian employee) would moderate a possible association of SD-tendency with mental health scores.
Non-response rates (NRR) were analyzed as indicators for sensitivity and scores of a SD-scale were tested for associations with mental health scores. Results showed that a particularly high NRR only occured for one item adressing alcohol consumption whereas average NRR for other mental health symptoms did not differ substantially from items on working conditions. Civilian employees showed higher NRR and SD-scores. Regression analysis revealed that personnel status moderated the association between SD and alcohol consumption scores.
Our results indicate that (1) mental health topics generally appear to be less sensitive among austrian soldiers than expected and (2) a SD-bias might have primarily occured for items assessing alcohol consumption of soldiers. Possible explanations may be seen in military culture/socialization, methodological issues and occupational norms related to alcohol misuse.
An App for the Correction of Measurement and Sampling Error in the European Social Survey
Dr Jorge Cimentada (Universitat Pompeu Fabra) - Presenting Author
Dr Wiebke Weber (Universitat Pompeu Fabra)
Although the importance of correction for measurement and sampling error is widely acknowledged, researchers face often challenges in applying methods for correction. We present therefore a Shiny app, an open source web application framework for R, which allows users to correct for those errors by navigating menu options while the statistical analysis is conducted in R without them having to know anything about R. For more advanced users, they can also access the R code itself, download it, take it apart, learn from it, and adapt to their own analysis.
Standardizing Sequence Lengths by Using the Relative Duration of Episodes
Dr Gesche Brandt (German Centre for Higher Education Research and Science Studies (DZHW)) - Presenting Author
Mrs Susanne de Vogel (German Centre for Higher Education Research and Science Studies (DZHW))
In Germany, trajectories of doctoral candidates are diverse and often consist of different stations. For example, doctoral candidates start with their PhD within a graduate school or scholarship, but later change into gainful employment inside or outside academia. Subsequently, sequences can take from a few months up to several years. The aim of our analyses is to identify patterns of doctorates, e.g. constant courses in a single institution or very complex courses in many different ones. This can help to identify typical doctoral trajectories in different subject areas, to identify problematic processes and to help optimizing doctoral conditions.
Sequence analysis is a very useful method for investigating types of patterns, including the timing of events or transitions, the duration and the order of life course episodes. However, analyses are complicated if the sequences have large gaps or different lengths. In order to make the doctoral candidates’ trajectories comparable, we use an innovative method for standardizing the sequence lengths by using the relative duration of episodes. For this, the percentages of single elements are calculated on the total length.
Analyses are based on longitudinal data from the DZHW PhD panel study. This representative sample comprises around 5400 doctorate holders of all disciplines who received their degree from a German University in 2014.
First analyses show that using the relative durations of episodes can be a helpful method to compare sequences of different length.