|TITLE||Leading by Listening: An Empirical Test of Carl Rogers's Theory of Human Relationships Using Interpersonal Assessments of Leaders by Followers|
School of Business and Public Management
The George Washington University
Doctoral Dissertation: February 1997
To examine the extent to which "leaderly learning" impacts on organizational outcomes.
A sample of work teams (in facilities management, finance, information systems, marketing and patient services) was randomly selected from different departments in a large hospital system. For each manager's assessment (N=26), confidential data was provided by an average of four team members (N=105, response rate = 87.5%), who completed Barett-Lennard's (1959) Relationship Inventory, assessing Rogers's three facilitative conditions, Bass's Multifactor Leadership Questionnaire (MLQ Form 5X-Short), assessing three attributes of transactional leadership and three leadership outcomes as assessed by followers, and the LPI-Observer, assessing five practices of leaders at their best. Internal reliabilities for the LPI were .87 or above in this study.
The research created canonical correlations for the three variables of "listening" (congruence, positive regard, and empathic understanding) and for the five variables of "leadership" (challenging, inspiring, enabling, modeling and encouraging). The canonical correlation between these two "factors" was .72 (p < .0001), with .88 of the total proportion of variance extracted. In validating these results, the estimates of the canonical coefficients in one subsample were used to predict the canonical variates in the other subsample and vice versa. Randomly splitting the 105 respondents into two subsamples, the resulting canonical correlations were .60 for one group and .70 for the other group, with both correlations significant at the .0001 level. Computing a redundancy index revealed that: "...on average, therefore, 40% of the variation in the best practices of leaders is explained by Rogers's facilitative conditions for listening" (p. 93).
The canonical variate of leaderly listening was shown to significantly (p < .0001) predict leadership effectiveness using the MLQ. Regressing leaderly listening on leader's effectiveness the explained variance (or r-squared) was .74, for satisfaction with leader r-squared was .76, and for willingness to perform beyond expectations r-squared was .66.
Further analyzes revealed that "Kouzes and Posner's best practices of leaders, in conjunction with Rogers's facilitative conditions, explain more about followers' assessments of a leader's effectiveness than routine management" (p. 109), and "about followers's assessment of satisfaction with a leader" (p. 110). This finding was not confirmed for willingness to perform beyond expectations, for which "contingent reward" was found to explain more variance. This latter result lends support to contentions that rewards come very close to transformational leadership as they are connected with followers' growth needs and taken into consideration by leaders.