A methodological experiment aiming at investigation of the frequency of false an­swers and causes of such behaviour is presented in this poster. The experiment was conducted du ring a Web survey of a sample of mailbox users in one of the most popular Polish Web portals (N=12000). The subject of the questionnaire was not directly related to the internet.

Both the response format (drop-down menu vs. radio buttons) and the number of possible answers were manipulated in the experiment. Moreover, repeated measurement analysis is conducted: "hard" demographic data collected in the survey were compared to the answers given while registering mailbox (full anony­mity of respondents was ensured). Responses to questions regarding the location of respondents while answering were compared to location data extracted from their IP numbers. Concluding, we are going to discuss some possible methods to control credibility and reliability of data gathered in online research.