Chapter 9. The Less Than Awesome Power of Expectations to Bias Perception, Memory and Judgment

Abstract

The chapter critically evaluates the conclusions often reached on the basis of the early research testifying to the power of expectations to bias perception, memory, and judgment.  It leads off by walking through several real world examples involving sports, academics, and the stock market, all of which strongly suggest that such biases are generally quite limited in daily life.  It then revisits some of the early classic studies widely interpreted as demonstrating large and dramatic expectancy-confirming biases generally: 1) provided more evidence of accuracy or unbiased responding, than of bias; or 2) have never been replicated.  After debunking the justifiability of testaments to the power of bias on the basis of these early classics, the chapter turns to meta-analyses of the wider literature – and shows that all meta-analyses assessing the effects of stereotypes or other expectancies on perception, judgment, and memory produce, on average, very small bias effects (averaging about r=.10).  The inexorable conclusion is that the biasing effects of expectations on judgments is, in general, very small.

EXCERPT:
    So, what have the meta-analyses [of the biasing effects of expectations] found?  The second half of Table 6.1 has presented the results, but I review them here …  The simple average of the effects obtained in these seven meta-analyses is .09.  The weighted average (weighting the effect size by the number of studies included in each meta-analysis) is .07.
Translated into lay English, this means that expectations bias judgments, on average, about 5-10% of the time (see footnote 4).  Or, put another way, on average, expectancies fail to bias judgments about 90-95% of the time.  I conclude, therefore, that neither the handful of high-impact studies often cited as demonstrating powerful biases, nor the broader, more general literature, demonstrate that expectancies typically have very powerful effects on perception and judgment.  Such biases are undoubtedly real — they occurred in many studies reviewed here, and the meta-analyses consistently found evidence of bias.  That being said, however, the only viable conclusion from this literature is that such biases are, in general, quite small.