From mapping to better measurement: RESPOND explores smarter ways to design corruption surveys

How can researchers design better corruption surveys without starting from scratch?

That question was at the heart of the latest RESPOND online seminar, “From Mapping to Design: Using Survey Questions to Benchmark Corruption Measurement”, hosted by the Institute of Social Sciences of the University of Lisbon (ICS-ULisboa).

Inês Santos Almeida, Research Fellow at ICS-ULisboa, and Gustavo Gouvêa Maciel, Research Fellow at the Institute for Sociology of the Slovak Academy of Sciences, presented ongoing work to expand DATACORR, a database bringing together thousands of survey questions used to measure corruption. Luís de Sousa of ICS-ULisboa hosted the seminar.

The work reflects a shift from asking simply how corruption has been measured in previous research to considering how that accumulated knowledge can help researchers design better surveys in the future.

Moving beyond “corruption is difficult to measure”

Corruption is notoriously difficult to measure. Surveys have therefore become an important source of evidence, capturing perceptions, attitudes and reported experiences that cannot always be observed through administrative or criminal justice data.

The researchers argued, however, that subjective and objective approaches to measuring corruption should not be treated as competitors. Survey evidence can complement other forms of measurement, while improvements in survey design can help researchers understand dimensions of corruption that purely legal or administrative measures may miss.

DATACORR was developed to bring some order to the large body of survey research already available. Its original version compiled more than 3,000 corruption-related questions from 321 survey rounds conducted between 1976 and 2019. The updated database incorporates 23 additional survey rounds from 2020–2025 and more than 400 additional questions.

The questions fall along two key dimensions. They can be generic, asking about corruption broadly, or specific, addressing practices such as bribery, favouritism or embezzlement. They can also be sociotropic, focusing on corruption in society or institutions, or egocentric, focusing on an individual’s own experiences or circumstances.

The updated evidence confirms that generic and sociotropic questions continue to dominate corruption surveys. More specific, experience-based questions have grown, but the change has been incremental rather than transformative.

This continuity has advantages. Established questions facilitate comparisons between countries and across time. But relying too heavily on them can also limit researchers’ ability to investigate particular behaviours, experiences or institutional vulnerabilities.

Turning a database into a design tool

The next step is therefore to use DATACORR not merely as an archive of previous research, but as a practical tool for designing new questionnaires.

To demonstrate this approach, the researchers conducted a benchmarking exercise focused on the relationship between political corruption and support for democracy.

Searches covering issues including political corruption, abuse of power, conflicts of interest, political finance, media capture, integrity and democratic norms initially produced 448 potential matches. After removing duplicates, 309 unique questions remained. The research team independently assessed and discussed these, ultimately producing a benchmark set of 40 questions.

The researchers stressed that these should not be understood as the objectively “best” 40 questions. The benchmark depends on the research problem. A study examining corruption and gender, corruption-control measures or experiences of bribery would produce a different selection.

Instead, DATACORR can help researchers identify existing questions, compare their wording and conceptual approach, and decide whether to reuse or adapt them to a new context.

From quantity to quality

Discussion following the presentation pointed towards the next challenge: assessing not merely which questions exist, but how well they are constructed.

Participants explored whether DATACORR could support survey experiments, including tests of how different formulations affect responses to sensitive questions. Clearer and more specific wording may reduce ambiguity and make respondents more comfortable reporting experiences or attitudes, but experimental research is needed to test these effects.

The discussion also highlighted the importance of context. Similar concepts can be expressed differently across countries and languages, while some regions and survey programmes place greater emphasis on personal experiences of corruption than others.

Future work on DATACORR will therefore examine the conceptual and methodological quality of survey questions, including whether concepts are clearly defined, questions are understandable and formulations introduce potential ambiguity.

The wider objective is straightforward: researchers already possess decades of accumulated experience in asking people about corruption. Making that knowledge systematically searchable, comparable, and reusable can help corruption research move beyond repeatedly observing that corruption is difficult to measure, and toward designing better ways to measure it.