6.5 Ideal experiments: Examples
- Bauer (2015): “Ideally we would conduct a randomized field experiment to maximize both internal as well as external validity. We would draw a large random sample from our target population (persons living in Switzerland) and measure the level of trust of all sample members by directly accessing their thoughts before and after the treatment. We would recruit homogeneous offenders that randomly treat half of the sample with exactly the same negative factual experience (treatment group) and leave the other half in peace (control group). Random assignment of the treatment would allow for estimating an unbiased (internally valid) causal effect, as it assures that the treatment Di is unrelated to the potential out- comes (Angrist and Pischke, 2008: p. 15). This ideal experiment would have strong external validity, as the sample is representative of a larger Swiss population and the treatment is a real-life experience. Besides, we would control the timing of both, outcome measurement and treatment assignment. Clearly, this ideal experiment can not be realized for ethical and practical reasons. Thus, we have to resort to ‘natural’ variation of our treatment, i.e. victimization across individuals. The described ideal experiment, however, serves as the benchmark to which we can compare our research design to reveal potential validity threats” (Bauer 2015, 399–400).
References
Bauer, Paul C. 2015. “Negative Experiences and Trust: A Causal Analysis of the Effects of Victimization on Generalized Trust.” Eur. Sociol. Rev. 31 (4): 397–417.