Measuring and Explaining Trust (TRUSTME)

• Applicant: Dr. Paul Cornelius Bauer
• Subject Area: Empirical Social Research
• Term: from 2020 to 2024
• Project identifier: Deutsche Forschungsgemeinschaft (DFG) - Project number 449946260

• Final Report Year: 2025

Final Report Abstract

The project "Measurement and Explanation of Trust (TRUSTME)" aimed to investigate two key questions in trust research: How can we measure trust? How can we explain differences in trust? Measurement of Trust. A key challenge in trust research is the lack of consensus on how trust should be measured. The TRUSTME project addressed this challenge by comparing the validity of standard measures of generalized social trust with newer, situation-specific trust measures. We show that survey measures referring to "strangers" in their question wording best reflect the concept of generalized trust, also known as trust in unknown others. While situation-specific trust measures may be linguistically more precise because they define situations more specifically within the question itself, they can also strengthen associations with people one knows personally. This is undesirable when measuring generalized trust. Explanation of Differences in Trust. The TRUSTME project also investigated factors that contribute to differences in trust levels between individuals. One focus was the use of different trust measures. The results show that generally lower trust levels are measured using current specific measures. Additionally, innovative methods, such as audio probing, were applied to measure the thought processes and emotions underlying trust judgments. Using these methods, we found that emotions influence trust judgments. Methodological Contributions. Over the course of the project, it became clear that various methodological studies are necessary to answer the aforementioned research questions and to promote innovation in the field of trust research (and beyond). These were conducted as part of the project. Landesvatter et al. (2023) compares various speech-to-text algorithms for the transcription of speech data from surveys. The results showed that the variation in accuracy between different ASR systems varies considerably, highlighting the need to compare different ASR systems for transcribing speech data. We investigated the use of speech input in surveys and found that spoken responses tend to be longer and somewhat more informative than written responses. This suggests that the use of speech input in surveys could potentially lead to richer and more nuanced data. Finally, we examined the concept of ideal research designs (IRDs), which can help develop better research designs.

Publications

• Comparing Speech-to-Text Algorithms for Transcribing Voice Data from Surveys. Center for Open Science.
  Landesvatter, Camille; Behnert, Jan & Bauer, Paul C.
  
• From ideal experiments to ideal research designs (IDRs): What they are and why we should use them more. Center for Open Science.
  Bauer, Paul C. & Landesvatter, Camille
  
• Dissertation: Methods for the classification of data from open-ended questions in surveys. Dissertation verteidigt an der Universität Mannheim am 16. April 2024
  Landesvatter, C.
  
• How Valid Are Trust Survey Measures? New Insights From Open-Ended Probing Data and Supervised Machine Learning. Sociological Methods & Research, 54(2), 534-564.
  Landesvatter, Camille & Bauer, Paul C.
  
• Do emotions affect political trust judgments?. Center for Open Science.
  Bauer, Paul C. & Landesvatter, Camille
  
• Information content in text and voice response formats for open-ended survey questions. Center for Open Science.
  Landesvatter, Camille & Bauer, Paul C.