In survey research, sub optimal sampling methods or formats of the questions asked can result in biased data, and so in poor results. Teaching this topic is hard because students can only "believe" the teacher and try to understand why and how biases can occur and contaminate the data. This paper introduces a new generic electronic learning environment that gives students hands-on experience with how their methodological choices affect the data. The learning environment consists of three modules. In the population module, the teacher defines a population. In the sampling module, the student can apply different sampling plans. In the survey module, the student can design a questionnaire and actually execute the survey. The resulting data file can be analyzed and compared to the population data. It is concluded that hands-on experience in a problem-based approach can support a deep understanding of several types of sampling errors and response biases.
The CAUSE Research Group is supported in part by a member initiative grant from the American Statistical Association’s Section on Statistics and Data Science Education