A stratified (based upon state of residence, gender, area of practice, and academic degree), systematic randomly selected sample of 2,500 social workers in the U.S. were surveyed using a mailed questionnaire, with 33 percent responding (N = 833). The typical respondent was a Caucasian (90%) female (78%) with a MSW degree (92%). Respondents completed the Leadership Practices Inventory (Observer), the Job Satisfaction Scale (Koeske, et al. 1994), and responded to a question about their satisfaction with the leadership behavior of their supervisor. An overall measure of leadership was computed with the LPI, rather than using the five separate leadership practice scales, and the JSS was also revised. Cronbach's alpha for the revised LPI was .98 and the revised JSS was above .86. Respondents were asked to complete the LPI twice: Once to indicate how they expected their supervisor to perform and again for how they perceived their supervisor was actually performing.
Just over 90 percent of the respondents indicated that the LPI items reflected the leadership behavior of their supervisor. There was a significant difference between the expected and actual LPI scores, with the former being uniformly higher than the latter. Thirteen of the 17 items from the revised Job Satisfaction Scale were significantly associated with the mean differences in leadership expectations and behaviors. "These results suggest that the greater the difference between expected and actual leadership practices, the more dissatisfied the social respondents were with their jobs." The three factors that accounted for the greatest variance in the differences between scores were the leadership behavior of your supervisor, recognition given your work by your supervisor and the quality of supervision you receive.