On 19 September 2012, Dr. Karin Lasthuizen, Associate Professor in Leadership and Governance Studies, VU University Amsterdam (The Netherlands) gave a talk at the Lee Kuan Yew School on research techniques to measure 'sensitive' issues such as corruption.
Dr. Lasthuizen began by identifying the problems associated with measuring corruption, including its causes, visibility, awareness of corrupt practices, and problems regarding the reporting, investigating and sanctioning of it. She then spoke on the types of data sources, and research strategies available. Some data is visible and investigated, while others remain hidden because of confidential internal investigations, or the reporting behaviour of victims and perpetrators.
One way to measure corruption is through reputational data. Surveys, such as those collected by Transparency International, an non-governmental organisation that publishes the widely cited Corruption Perception Index, ask respondents to estimate the amount of corruption in their society. However, reputational data tends to be concerned with opinion, rather than evidence. While it can be reliable and useful, especially in comparative studies, generalised perceptions and preconceived opinions mean it suffers from bias. This may explain why some countries remain at the same level for years. The terms of definitions in the perceptions index also tend to have a Western bias.
Another method of collecting data is through criminal investigations and internal investigations data available from reports and prosecution files. Most internal investigations following workplace and citizen complaints do not lead to convictions. This data is thus limiting as they do not provide substantial evidence and only circumstantial. This kind of data also suffers from “dark problems” that relate to insufficient information to make robust assumptions, hidden data, and interpretation problems. This includes disparities between organisations which are more willing to investigate and report versus those that are not. Thus, the question remains, does a high number of investigations indicate high levels of corruption or rather, high levels of willingness to investigate?
A third method of collecting data about corruption is to survey employees, by asking questions regarding the extent of perceived unethical behaviour and integrity problems. Again, this method measures perceptions rather than actual occurrences. Questions are raised over how much employees can observe and relate. There might be problems with regard to translation and identity and/or equivalence problems mainly to do with cultural bias and definitions. Questions arise over what exactly respondents tend to report, due to loyalty to, or on the other hand, frustration with the organisation or on the other hand, frustration with the organisation.
Another method Dr. Lasthuizen discussed is victimisation data. Self-reporting of corrupt behaviours are complicated by cross-cultural studies. Different cultures may accord different meanings to words such as “bribe” or “corruption”, while the social atmosphere and level of cynicism within their society may have an affect on the willingness to report as well. While this method is ideal for research into sensitive behavours, the question remains over how to ask about self incriminating behaviour?
The last method discussed by Dr. Lasthuizen was the experimental data method. Where the answer is randomised, Dr. Lasthuizen's experiment called for rolling a six-sided dice, which only the respondent can see. Regardless of if the deed is done or not, the respondent is advised to say no, if he or she rolls a “1” or “2”, and to say yes, if a “5” or “6” is rolled. And if the respondent rolls a “3” or “4”, he or she is advised to answer truthfully. Needless to say, this method provides cover for self incriminating answers as well as does not suffer from insufficient data problems. However, there are technical problems, problems with lying and cheaters to contend with. Also, several respondents have a problem with saying yes, when they had not done anything wrong.
In conclusion, Dr. Lasthuizen summarised that there are many methods of measuring corruption and sensitive behaviour, which fundamentally depend on the availability of data. The experimental data method and with other new methods are promising in addressing the sensitive nature of corruption Researchers can also use multiple methods to ensure the same answer through triangulation. However, the biggest obstacle to measuring corruption remains cultural bias, as corruption still tends to be defined in Western perspectives.
is a second-year Master of Public Policy student at the LKY School of Public Policy.