Bias (+ Confounding?)
Fragment of a discussion from Course talk:STAT 550
1) Attrition Bias: Caused by loss of participants. People who drop out of a trial might just be ignored by investigation, but the drop out could be experiment related. For example in cancer investigation someone might drop out to get treatment abroad due to drugs not working.
2) Publication Bias: (Bit different) Reports are far more likely to be published if they contain significant results rather than null hypothesis result. Often leads to unreported results which could effect other experiments.
3) Performance Bias: Systematic differences in the care provided to the participants in the comparison groups other than the intervention under investigation.