Using pirated software, gambling, and visiting pornographic websites are examples of behaviors that are hard to measure because people fear the consequences of being truthful. This reluctance can hinder efforts to understand and address these problems. However, one study suggests that it may be possible to elicit more honest responses without compromising confidentiality.

Samuel S.K. Kwan, Mike K.P So and Kar Yan Tam of HKUST applied a tested research method not commonly used in information systems research, randomized response technique (RRT), to study software piracy. They were able to yield strong evidence that it makes people more forthcoming than when they are asked direct questions on sensitive topics.

RRT dilutes the stigma or embarrassment caused by sensitive questions, by mixing them up with innocuous questions and letting the respondent choose which to reply to. For instance, they can flip a coin and reply to question 1 for heads and question 2 for tails, but only they will know which is which. Researchers will determine what question they answered based on probabilities, rather than being told directly by the respondent.

The success of the authors' approach is important in an industry like software piracy, which has been estimated to cost the industry US$48 billion a year. Researchers frequently cite response distortion as a limitation in their findings, but until now little effort has been made to resolve this problem.

"Response distortion generally happens when the answer provided by the subject doesn't accurately reflect his genuine opinion, belief, feeling, intention or behavior. It may be that the respondent finds the questions or answers sensitive or worries that their response might incur legal liability. This poses serious difficulties for behavioral research and, ironically, may result in a lack of research in areas where objective investigations are badly needed," they say.

Their research addresses this concern. It involves two online studies. In each study, one group was asked directly about their attitudes, intentions and behaviors regarding software piracy, while two other groups were tested using RRT. The RRT groups chose a number between one and four and kept it to themselves. One group answered a question if their number matched a randomly generated number that flashed on their computer screen. The other group answered if it did not match their number.

The wording of the questions in RRT was carefully executed to pair an innocuous question with a sensitive one so as to gain respondents' trust that their response would not directly link them to illicit behavior (for example, in one pairing they responded to either "I have a lot of software that was copied without authorization" or "I like holidays").

Given the methods and probabilities, the authors were able to confirm that RRT elicited more frank responses than direct questioning.

"The distortions appear to be more serious in self-reports of intentions and behavior compared with attitudes toward pirated software. In general RRT respondents showed greater willingness to report politically incorrect or illegal attitudes, intentions and behaviors," they say.

The authors also expanded the use of RRT by considering multiple variables, in this case considering how punishment certainty and software cost influence attitudes and how this links to piracy behavior. Previously RRT has only been used in single variable questions. They found the influences of attitudes and punishment certainty may have been over-reported in other studies, while the effect of cost may be under-reported.

Overall, the research confirmed the validity of using RRT to investigate sensitive topics. "Our work represents a pioneering effort to provide empirical evidence to substantiate the worries of many previous researchers about response distortion in software piracy research. With better understanding of the extent of response distortion and its effects on causal relationships, efforts to combat piracy could be more precisely directed," they say.

"This problem may also exist in the context of other sensitive topics such as data privacy, cyberslacking, hacking, online gambling, and pornography viewing. RRT is recommended as a method to tackle response distortion in research related to sensitive topics."