7
measure of trust in government (Keele 2003) where higher values on this variable indicate greater levels
of trust. All of these variables are time series data available at the yearly level. The mean for policy
mood during the 1983 to 1999 time period is 61.77, mean macropartisanship is 54.23, and mean trust in
government is 41.84. Their standard deviations are 3.02, 2.62, and 4.52, respectively.
Other Political Institutions
The second set of predictor variables focuses on the temporal relationship between the public
approval of two prominent political institutions – Congress and the presidency – and public approval of
the bureaucracy. Scholars have long recognized a close relationship between Congress, the President, and
the bureaucracy (Rourke 1984; Wilson 1989; West 1995), and there is a good deal of evidence suggesting
the formal powers that Congress and the president exercise over the bureaucracy can lead, at least to some
degree, administrative agencies to be responsive to their political principals (Wood and Waterman 1991,
1993). Congress, for example, sets agency budgets (Bendor and Moe 1985; McCubbins 1985; Calvert,
McCubbins, and Weingast 1989), holds oversight hearings (McCubbins and Schwartz 1984; Aberbach
1990, 2002), and imposes administrative procedures on the implementation of legislation (McCubbins,
Noll, and Weingast 1987, 1989; Bawn 1995; Potoski 1999, 2002). The President’s powers over the
bureaucracy include some administrative appointments and the agenda control arising from recent
centralization of decision making via relying more heavily on White House staff and the Office of
Management and Budget (Moe 1985; West 1995). If presidents and/or Congress are fully responsible for
bureaucratic performance, then we might expect that citizens’ assessment of bureaucratic performance are
really just evaluations of their principals, one step removed.
There are, however, good reasons to doubt this hypothesis. First, many scholars doubt whether
these principals really exercise such clear cut authority over bureaucracy performance. Krause (1996), for
example, suggests that bureaucrats are quite capable of influencing the behavior of the political principals
who monitored them. Moe (2003) similarly found that bureaucrats could influence the election of the
political principals that oversee their behavior. Second, and adding further complexity to the interplay
Convention