Article Preview
TopIntroduction
Coronavirus crisis drove school education, university classrooms, job interviews, medical surveys and many other essential practices online. Thus, it becomes more and more important to assess the authentically of online responses. For example, how do we keep students from cheating in online exams? To what extent people tend to cheat while going online? May signing the honor code be helpful, and to what extent? This paper targets to investigate these questions.
Until recently, the traditional standard approach for administering questionnaires has been to use hard copy questionnaires, i.e. paper-and-pencil questionnaires (Oppenheim 1992). However, technological changes have enabled people to complete questionnaires, surveys and online exams quickly and with a high-quality level (Strabac and Aalberg, 2011). The literature discusses the advantages and disadvantages of online questionnaires (Solomon, 2001; Denscombe, 2003; Kaplowitz et al., 2004). However, only a few researchers have focused on cheating in online questionnaires testing knowledge because it is challenging to measure cheating, even though transmitting questionnaires to online settings may significantly skew the results.
There are two main problems in measuring the knowledge level of respondents. The first problem is a known problem in knowledge surveys, which is the tendency of respondents to guess the correct answer, depending on their personality (whether they tend to guess or not) (Mondak and Davis, 2001; Mondak and Anderson, 2003; Sturgis et al., 2008). Because this problem exists for all types of questionnaires (online, hard copy etc.), it is not important when choosing how to deliver a questionnaire (i.e. an online questionnaire or a hard copy questionnaire). The second problem, which is what this article focuses on, is related to online surveys designed to measure knowledge: it is the possibility that respondents will use their Web browser to search for answers, and thus will 'cheat' in questions relating to their knowledge (Strabac and Aalberg, 2011; Jensen et al., 2014). This problem affects online questionnaires relating to knowledge. Given that the use of such questionnaires is relatively novel, the literature on this subject remains limited.
Previous studies have found that the average level of knowledge observed in Web surveys tends to be higher than the average level observed in telephone or face to face surveys. Most of these studies measure political knowledge (Ansolabehere and Schaffner, 2011; Berrens et al., 2003; Elo, 2009; Strabac and Aalberg, 2011; Jensen et al., 2014) and a few measure scholarly knowledge (Fricker, 2005). However, when they measure knowledge, some researchers do not see the action of using search engines as a source of measurement error (Fricker, 2005; Ansolabehere and Schaffner, 2011). Instead, they suggest that the high evaluation of political knowledge observed in Web surveys is due to an ongoing sampling of people with more knowledge (Ansolabehere and Schaffner, 2011). It is believed that online questionnaires have a sampling problem, and thus they do not constitute a representative cross-section of the population (Denscombe, 2003). Another belief is that the main reason that, on average, online respondents (compared to phone respondents) answer a higher percentage of questions correctly is because questions about knowledge are easier to answer when they are presented visually and when the respondents can answer them at their own pace – and indeed, online respondents take longer to answer information items (Fricker, 2005).