Flawed Social Experiments

摘要: This paper describes, somewhat overlapping, types of flaws that have occurred in social experiments. Each flaw is illustrated with examples from previous Some these problems result minor hurdles, while others cause experiments to fail — is, the experiment unable provide a valid test hypothesis interest. An accompanying summary table lists flaws, indicates circumstances under which they occur, their potential seriousness, and approaches for minimizing them. The most important are response bias resulting attrition; failure adequately implement treatment as designed; too small sample detect impacts. third can insufficient marketing, an initial target group, disinterest on part group participating (if voluntary), or attrition. To considerable degree discussed this article be minimized. For instance, implementation failures usually avoided sufficient effort planning often mitigated example, through increased follow-up efforts conducting surveys.