Meet the Experts

Registration (via Zoom)   |  

Slides (1,92 MB) |   Presentation on YouTube   |   MTE Playlist

The Lecture will be held in English.

Survey programs often choose different wordings or response scales to capture the same concepts. Furthermore, wordings or response scales can also change during a long running time series within a survey program. This can pose a challenge if we want to compare or combine survey data across different survey programs or across time. A central challenge is that such question differences lead to different measurement units, meaning that the numerical values in the different datasets imply different levels of the same concept. The QuestionLink R-Package was designed to help researchers with this challenge. QuestionLink uses a psychometric approach called observed score equating in a random groups design (OSE-RG) to increase the comparability across different question wordings and response scales. The main advantage of QuestionLink is that it finds opportunities for harmonization in large, combined datasets from different survey programs automatically. However, in its current form, QuestionLink harmonizes data within countries, but not across countries.

In this talk, I will (1) briefly explain the harmonization challenge, (2) explain how QuestionLink resolves this challenge, and (3) give some practical guidance on how to tell if QuestionLink is the right fit for your project.

Presenter:

Dr. Ranjit K. Singh

Registration (via Zoom)   |  

Slides (5,81 MB) |   Presentation on YouTube   |   MTE Playlist

The Lecture will be held in English.

Designing questionnaires is considered an art that requires knowledge and experience. To improve the scientific nature of this process, researchers can use the freely accessible web-based tool, the Survey Quality Predictor (SQP) 3.0. Similar to weather forecasting, SQP estimates the measurement quality of survey questions based on their formal-linguistic characteristics. When comparing different question versions, their quality estimates provide valuable guidance in selecting the better option. Additionally, they can be used to correct for measurment error in substantial analyses.

As a supplement to conventional methods such as cognitive pretesting and expert reviews, SQP can also be used to refine questions and support international survey projects by identifying discrepancies between an original suvery question and its translations.

In this talk, I will (1) briefly provide an overview of key decisions involved in survey question design, (2) explain how the quality of a survey question can be estimated, and (3) offer some practical guidance on using SQP to predict the quality of a survey question. This tool is particularly valuable for anyone designing their own survey questions, working on international survey projects, and seeking to correct for measurement error in their analyses.

Presenter:

Dr. Lydia Repke

Registration (via Zoom)   |  

Slides |   Presentation on YouTube   |   MTE Playlist

The Lecture will be held in English.

ZIS is an open access repository that provides validated measurement instruments. As such it fosters transparent and reproducible research. In this talk, we will explore how ZIS empowers researchers to discover, create, and share essential instruments for data collection. We’ll navigate the platform’s search and filtering features, e.g. for finding suitable scales, then outline the process for publishing new or adapted instruments. By showcasing best practices, we will demonstrate how using ZIS can streamline research workflows, promote quality assurance, and advance open science principles.

Presenter:

Dr. Piotr Koc

Registration (via Zoom)   |

Slides |   Presentation on YouTube   |   MTE Playlist

The Lecture will be held in German.

Kognitive Pretests prüfen, ob Fragen bzw. Items verständlich sind und wie intendiert von den Befragten interpretiert werden. Die GESIS Pretest-Datenbank ist ein Open Access Repositorium von Survey-Fragen, die mittels kognitiver Pretests bei GESIS evaluiert wurden. Sie richtet sich an Personen, die kognitiv getestete Fragen recherchieren, entweder im Rahmen der Fragebogenkonstruktion, um diese für eigene Erhebungen zu nutzen, oder um Informationen darüber zu erhalten, wie Fragen, die in Umfragen eingesetzt wurden, von den Befragten verstanden und beantwortet werden. Die Datenbank enthält aktuell mehr als 3.000 kognitiv getestete Fragen bzw. Items aus vielfältigen Themenbereichen (z.B. Arbeit & Beruf, Gesellschaft & Soziales, Gesundheit, Politik). Für die Fragen sind die detaillierten Ergebnisse der Pretests dokumentiert, sowie etwaige Vorschläge zur Optimierung der Fragen. Der Vortrag beginnt mit einer kurzen Einführung zum Hintergrund kognitiver Pretests und stellt im Anschluss die Inhalte der Pretest-Datenbank, ihre Funktionen und ihre Recherchemöglichkeiten anhand verschiedener Beispiele vor.

Vortragender:

Dr. Timo Lenzner