Data desirability, experimenter expectancy, and the results of psychological research.

19 male Es employing a Taffel-type task conducted a verbal conditioning experiment with 60 female Ss. I the Es were led to expect their Ss to show verbal conditioning, and 4 were led to expect no verbal conditioning. 4 the Es in each of these groups were led to feel that it would be desirable if their Ss showed conditioning, and J were led to feel that it would be undesirable. Those Es who (a) both wanted and expected, and (b) neither wanted nor expected their Ss to show increased use of i and WE pronouns obtained significant conditioning (p = .001). Those Es who (a) wanted but did not expect, and (b) expected but did not want increased use of I and WE pronouns obtained no significant conditioning (p = 1.00). Ss high in need for social approval arrived earlier at the site of the experiment, were less "aware" of the contingency but were no more likely to show conditioning. Ss' ratings of Es' behavior during the experiment showed significant differences between Es in different experimental conditions, between Es who were 1st vs. later born, and between Es who were high vs. low in need for social approval.