A social relations analysis of leadership Stefano Livi a, ⁎ , David A. Kenny b , Linda Albright c , Antonio Pierro a a Department of Social and Developmental Psychology, University of Rome “La Sapienza”. Via dei Marsi, 78, I-00185, Rome, Italy b Department of Psychology, University of Connecticut, Storrs, CT, USA c Department of Psychology, Westfield State College, Westfield, MA, USA Abstract Different theoretical approaches assume, either implicitly or explicitly, that leadership operates at one of three levels of analysis: group, dyad, or individual (perceiver and target). The social relations model [Kenny, D. A. (1994). Interpersonal perception: A social relations analysis New York: Guilford] allows for all of these levels to operate simultaneously and can be used to empirically determine the amount of variance at each level. Using this methodology, this article addresses several important theoretical issues in the area of leadership emergence, beginning with the fundamental question of whether group members agree as to who leads and who does not. Based on data from several previously published and several new studies, it reaches the following conclusions: (1) There is substantial agreement about who in the group is a leader; (2) member–member agreement increases as group size increases; (3) the average level of leadership does not vary much from group to group; and (4) perceivers are prone to a self- enhancement bias. Finally, the judgment of leadership operates to a greater extent at the dyad level for socio-emotional, as opposed to task-oriented, leadership. Published by Elsevier Inc. Keywords: Social relations model; Leadership emergence; Leader–member agreement 1. Introduction Leadership, perhaps as much as any construct in the social and behavioral sciences, has been theorized to operate at multiple levels of analysis. Different theories of leadership have emphasized different levels of analysis: the individual (both the perceiver and the target), the dyad, and the group. However, Yammarino, Dionne, Chun & Dansereau (2005) have emphasized “the importance of clearly specifying the level(s) of analysis at which phenomena are expected to exist theoretically” and have noted that “it is critical to ensure the measurement of constructs and data analytic techniques correspond to asserted level(s) of analysis, so that inference drawing is not misleading or artifactual” (p. 879). Moreover, not having an explicit specification of the appropriate level of analysis can have dramatic consequences, including incorrect and misleading conclusions (Kenny, Mannetti, Pierro, Livi & Kashy, 2002; Schriesheim, Castro & Yammarino, 2000). This article provides a detailed examination of the sources of variance for leadership across several studies. Both the development of theory and the conceptualization of empirical research should be guided at the outset by an understanding Available online at www.sciencedirect.com The Leadership Quarterly 19 (2008) 235 – 248 www.elsevier.com/locate/leaqua ⁎ Corresponding author. E-mail addresses: stefano.livi@uniroma1.it (S. Livi), david.kenny@uconn.edu (D.A. Kenny), lalbright@wsc.ma.edu (L. Albright), antonio.pierro@uniroma1.it (A. Pierro). 1048-9843/$ - see front matter. Published by Elsevier Inc. doi:10.1016/j.leaqua.2008.01.003