Relevance & Research Question

Refugees in first host countries often face adverse conditions including limited legal rights, restricted employment access, and inadequate social services. While resettlement programs offer long-term settlement opportunities for particularly vulnerable refugees, little is known about how these individuals evaluate and prioritize different aspects of potential destination countries. Traditional survey items fall short in capturing refugees' complex decision-making processes regarding resettlement. This study addresses this gap by employing a choice-based conjoint experiment to reveal how refugees eligible for resettlement weight different factors in their settlement decisions.
Methods & Data

The study uses data from an online survey (n = 375) conducted jointly by BAMF-FZ and the University of Bielefeld, targeting particularly vulnerable refugees eligible for resettlement procedures. Respondents complete up to four paired comparisons of hypothetical destination countries that vary across eight key characteristics: family presence, diaspora size, crime rates, language skills, attitudes toward refugees, political systems, labor market access, and lifestyle familiarity. After each choice, respondents evaluate their perceived opportunities for building a future in both countries, enabling analysis of both relative preferences and absolute settlement potential assessments.
Results

The conjoint approach successfully reveals how refugees trade off different aspects of potential host countries when making settlement decisions. Preliminary findings indicate that attitudes toward refugees and political systems emerge as the most influential factors in resettlement preferences. Notably, the relative importance of these factors varies considerably across refugee groups, highlighting the heterogeneity in decision-making priorities shaped by diverse experiences and backgrounds.
Added Value

This study advances the methodological and substantive understanding of refugee migration preferences. By focusing on refugees in their first host countries, the study provides reliable insights into real-world decision-making under precarious conditions. Thus, it offers valuable lessons for researchers targeting vulnerable groups through choice-based conjoint experiments.

Relevance & Research Question
The perception of flight safety represents an exemplary case for studying how digital media mediate fear, risk, and trust. While aviation remains one of the safest modes of transport, emotional reactions to incidents are often disproportionate, amplified by online news and algorithmic visibility. This paper asks: How does exposure to digital media shape fear of flying and perceived aviation risk? and To what extent can online sentiment predict self-reported anxiety in surveys? The study addresses an underexplored link between risk perception, media consumption, and emotion measurement in digital environments.
Methods & Data
I integrate two complementary data sources. First, a national online survey (N = 921) which measures emotional, cognitive, and behavioral components of flight anxiety, including trust in institutions, information sources, and prior flying experience. Second, we analyze a corpus of 43,000 online news articles containing the keyword “aviation” (2022–2023), collected from Romanian mainstream media. Using topic modeling (LDA), sentiment analysis (NRC and BERT-based classifiers), and correlation with survey variables, we identify symbolic frames of fear (“catastrophe,” “loss of control,” “safety reassurance”). All analyses were conducted in R.
Results
Findings reveal that individuals with higher exposure to negatively valenced news content report significantly higher levels of aviation anxiety (p < .01). Topic models show a predominance of emotional frames emphasizing loss, danger, and uncertainty, while trust in technological expertise moderates these effects. The integrated model combining sentiment scores and survey data predicts fear-of-flying.
Added Value
This study advances methodological innovation in online research by combining survey data with large-scale digital trace data. It demonstrates how affective communication and digital media coverage can shape collective perceptions of risk, offering new insights for both computational social science and public communication of safety. The proposed framework can be extended to other domains of perceived technological risk (AI, climate, or health).

Relevance & Research Question

Social surveys often collect information about industry and occupation. The traditional ‘gold-standard’ approach has been to capture this information by asking open questions about job title and duties which are later classified into standardised coding systems by expert coders. The shift towards self-completion surveys makes this more challenging as respondents often provide insufficient information to facilitate accurate coding. In addition, office-based coding is expensive and time consuming.

Closed-list questions, where respondents choose from pre-defined categories, could reduce response burden and coding costs and potentially also remove ambiguities inherent to open-ended responses. However, a challenge is that category labels can be difficult to interpret.

This paper assesses whether closed-list questions for industry and occupation can produce industry and occupation classifications comparable to those derived from manually coded open text responses.

Methods & Data

We use data from two waves of the NatCen Panel survey, a UK probability-based online panel, which employs a mixed-mode design combining online self-administration with telephone interviews. In these two waves, the panel supplemented its standard industry and occupation open-text questions (which are manually coded) with closed-list questions. We compare agreement between the two methods, using both descriptive analysis and regression models.

Results

For industry, the mean agreement rate between the closed-list and 1-digit manual codes was 64%, with variation across industries. Agreement was lower for occupation, at 56% for 1-digit codes and 46% for 2-digit codes, with significant differences across occupation types. Agreement rates were also influenced by sociodemographic and the length of open-text entries, with shorter descriptions generally leading to higher agreement rates.

Added Value

This study provides a first empirical assessment of the quality of occupation and industry data collected using a closed-list approach, offering evidence on the method's potential and limitations. The findings help identify potential improvements for collecting industry and occupation data in online surveys.