	Mentoring theory suggests that successful dyads will have mutual liking, identification, and attraction, which may explain the discrepant findings for informal versus formal mentoring, as informal mentorships are unlikely to emerge without liking and desire to be engaged in the relationship. How, then, should formal mentors be matched with protégés? Recent meta-analytic evidence has shown that deep-level similarity is the strongest predictor of perceived instrumental support, perceived psychosocial support, and relationship quality, while surface level similarity was only weakly related to these outcomes. This suggests that individuals should be matching based on values and preferences rather than superficial demographic characteristics. An interesting consideration, however, which has received less attention in the literature, is the extent to which mentors and protégés are drawn towards each other based on perceived match on visible characteristics, and how their subsequent judgments of deep-level similarity are influenced by perceptions of surface attributes. 
	Ragins offers a power perspective on diversified mentoring relationships, which are defined as mentoring dyads composed of a mentor and protégé that differ in power due to their group membership. Diversified mentoring theory draws on social identity theory and the similarity-attraction paradigm in order to explain why homogenous pairs may be more likely to form and experience greater identification and role modeling. Individuals are attracted to similar others and will initially form perceptions of similarity in part based on superficial characteristics. This may not be a problem, as homogenous pairs have been found to have greater relationship quality. Despite potential benefits, however, homogenous dominant pairs will perpetuate existing power structures, and minority protégés of minority mentors may be at a disadvantage in terms of career advancement, though Ragins proposes that homogenous minority pairs will experience high generativity and fulfillment. 
As white men have typically been greater in number and held higher-status positions, minority individuals will often be paired with non-superficially similar mentors (in power-asymmetrical dyads), while protégés from the dominant group are more likely to find themselves in homogenous pairs (power-symmetrical dyads), though the theory also encapsulates majority mentor/minority protégé pairs. These power differences have important implications for the quality and outcomes of the mentoring relationship. Given that white men have historically had the most power within (Western) organizations, they may be better able to provide instrumental career support, especially for minority group members who may have more barriers to overcome in order to succeed. This idea of the “white male mentor premium” has received support; protégés of white male mentors have been found to have higher compensation, though this may also be due to the greater likelihood of white men being the protégés of other white men. However, power-asymmetrical relationships have been found to be lower in psychosocial support and role modeling. The greatest difficulty arises in asymmetrical pairs in which the protégé is from the dominant group, and these pairs have been shown to have comparitely poor outcomes. In both cases, however, being in a diversified dyad is thought to promote empathy and interpersonal sensitivity. 
