Chapter 13. Teacher Expectations:
Accuracy and the Quest for the Powerful Self-Fulfilling Prophecy

Abstract

     This chapter is framed around reviewing the scientific and empirical literature that has addressed three questions regarding teacher expectations: 1. How powerful are expectancy effects in the classroom? 2. How accurate is the typical teacher expectation? 3. Have any conditions been identified under which truly powerful self-fulfilling prophecies do occur?  The results regarding the first two questions are vividly clear because they have been so well-established by so many studies: Expectancy effects in the classroom exist, but are generally weak, fragile, and fleeting; and teacher expectations predict student achievement primarily because those expectations are accurate.  A far more limited body of scientific research has addressed the third question, so that conclusions regarding conditions under which powerful self-fulfilling prophecies must be held more tentatively.  Nonetheless, unusually large self-fulfilling prophecies have been found among students suffering from some sort of stigma (race, class, low achievement).  Whether self-fulfilling prophecies primarily help or harm students, however, is currently unclear from the existing literature, with my own research finding helpful self-fulfilling prophecy effects among the largest yet found, but several other studies finding more evidence of harmful than of helpful self-fulfilling prophecies.

EXCERPT:
Table 13-1 summarizes the results of every quantitative (i.e., they reported results in terms of numbers, not researchers’ impressions) naturalistic study of self-fulfilling prophecies in the classroom of which I am aware.  Two features of those results are particularly worth noting.  Self-fulfilling prophecy effect sizes range from 0 to .4, with most falling between .10 and .20. Depending on how it is calculated, the overall mean effect size is between .07 and .17 (see Table 13-1 for more details [note: .17 is the simple average; .07 is the average weighted by sample size).  It remains unclear to me how researchers can justify testaments to the power of self-fulfilling prophecies (see quotes in Chapters 4 and 5) in the face of these data.
Second, Table 13-1 clearly shows that the larger the sample size, the smaller the self-fulfilling prophecy effect size (on average). Indeed, the correlation between sample size and effect size is -.71.
Why would larger samples produce such consistently smaller effects?  This is the wrong question, because I am sure the large samples do not actually cause the smaller effects.  Instead, as the statistically inclined can readily verify, effect sizes are more variable in smaller samples.  Effect sizes that are near zero rarely get published.  As the statistically inclined know, small sample studies require larger effect sizes than do large sample studies to obtain the holy grail of statistical significance.  Without a statistically significant result, a self-fulfilling prophecy study is unlikely to get published.  This primarily leaves the small scale self-fulfilling prophecy studies that produce larger effects in the literature.  The larger effect sizes obtained in the published small studies than in published large studies probably reflects the inherently greater random noise in such studies rather than any substantively generalizable evidence of larger self-fulfilling prophecies.