A further development in the literature is to incorporate the concept of symbolic representation, which, unlike active representation, works cognitively on the public. Thus, when bureaucrats share the identification, experience, and characteristics of a portion of the public, that audience will perceive the actions of those bureaucrats as legitimate, even if the bureaucrats are not purposefully representing that group. Nick Theobald and Donald Haider-Markel (2009), by examining citizen attitudes about actions by police officers, show that actions by bureaucrats are more likely to be perceived as legitimate if citizens and bureaucrats share demographic characteristics. If this holds across agencies, it suggests that citizen attitudes about bureaucrats and policy implementation can be changed without actions on the part of bureaucrats that are expressly designed to represent certain groups (active representation). Furthermore, they argue that methods used by those studying representative bureaucracy have relied on aggregate data, which makes it difficult to know if their findings demonstrate active or symbolic representation. Considering the implications for democratic governance, it is important to clarify this question.