| The poetry of reality|
|We must know. |
We will know.
|A view from the|
shoulders of giants.
Experimenter bias is the introduction of any bias into the conclusions of an experiment due to the preconceptions or goals of the experimenter being made clear to participants in the study. The bias can be introduced at any stage of the experiment, from the construction of the experiment, the selection of subjects, the administration of the experiment, the analysis of the data, or the conclusions drawn.
Experimenter bias stems from the deep-seated beliefs of an experimenter and is therefore difficult to prevent. Modern psychology generally supports the belief that is not possible to prevent experimenters from being biased, so it is necessary to design experiments in ways that bias cannot influence the results.
A great number of experimental methodologies are attempts to limit the unintentional introduction of experimenter bias. These include randomized controlled trials and double blind. Double-blind experiments are advisable whenever possible. In cases where measurements are recorded by human observers rather than by mechanical instruments, it is advisable to train people as observers who have little or no stake in the outcome of the experiment until their recorded observations correlate well on trials.
Minimizing interaction between the theorist(s) and experimental subjects to the barest minimum will limit the effects of bias, although this is not possible for all types of investigation.
Involvement in woo
Most experiments that purport to show evidence for psychic powers, alternative and quack medicine, or other woo fail to use proper methods and controls to limit experimenter bias. (For example, the doctor telling patients that homeopathy will work while failing to do so for the placebo). When replication attempts are made that do control for experimenter bias, the results are shown to be no better than chance.