Conservatism Bias

Conservatism bias, in cognitive psychology and decision science, is the tendency to revise a belief too little in light of new evidence, updating more slowly than the mathematically correct amount would call for. The psychologist Ward Edwards introduced the concept in 1968 through experiments in which participants estimated the probability that a given bag, among two available, was the source of a sample of colored chips; when the sample strongly favored one bag, the mathematically correct posterior probability was around 97 percent, but participants' estimates typically clustered around only 70 percent, revealing that people update their beliefs in the right direction but by a smaller amount than the evidence actually supports. The bias reflects a tendency to over-weight a prior assumption and under-weight new evidence, rather than a complete failure to revise a belief at all, and it has been applied outside the laboratory to explain findings in finance, where investors have been observed to under-react to new corporate information such as earnings reports and dividend announcements.