| Title: | Presidents, Chiefs of Staff, and White House Organizational Behavior: Survey Evidence from the Reagan and Bush Administrations. |
| Subject(s): | |
| Source: | |
| Author(s): | |
| Abstract: | Presents information on a study which investigated how presidential and chief of staff management styles shape the United States administration's organization structure. How the effect of presidential management style on White House's organization structure vary across the Reagan and Bush presidencies; Chiefs of staff from 1969-2000; Roles assumed by the chiefs of staff; Conclusions. |
| AN: | 3497282 |
| ISSN: | 0360-4918 |
| Database: | Academic Search Elite |
Section: ARTICLES
The authors set forth a behavioral model of the White House 5 organizational structure by taking into account the management styles employed by both the president and chiefs of staff, as well as how well they work together. Using survey data drawn from both Reagan and Bush administration elites, the statistical results show that these factors are important in explaining the White House's organizational structure. The authors also obtain evidence that presidential management style's effect on White House organizational structure does vary across the Reagan and Bush presidencies, but not within each administration across different chiefs of staff Although chiefs of staff have their own unique way in shaping the organizational structure of the White House, it fails to translate into altering the effect of presidential management style on the White House's organizational structure.
There has been an extensive body of research that investigates the organizational aspects of the American presidency (Buchanan 1990; Burke 1992, 25; George 1980; Hart 1995; Henderson 1988; Hess 1988; Johnson 1974; Porter [1980] 1988; Thompson 1992; Walcott and Hult 1995; Warshaw 1997). This line of scholarship has examined how the White House operates as an organization. This body of research has yielded fruitful insights yet does not provide a behavioral framework to understand the workings of the White House as an organization. Providing a better understanding of this phenomenon is important for students of executive politics, given that the effectiveness of presidential administrations in the policy and administrative arenas is thought to be related to the White House's organizational structure (e.g., Burke 1992; George 1980; Hart 1995; Kessel 1983, 1984; Krause 2001; Pfiffner 1999). How the White House is organized in practice and how the president actually behaves determine the functional organizational structure.(n1) This, in mm, has implications for determining whether or not a given administration is successful or not along the dimensions of policy formulation, policy implementation, and policy outcomes.
Our purpose is to augment this burgeoning line of research by setting forth a simple behavioral model of how presidential and chief of staff management styles shape the White House's organizational structure. We use original survey data gathered from former Executive Office of the President (EOP) staffers and cabinet members in both the Reagan and Bush administrations to assess our model empirically. We wish to determine whether the individuals operating in these institutional positions really matter in determining the working or behavioral structure of the American presidency or whether it is based on systemic relationship patterns. We now turn our attention to a critical position that has developed in the modern White House-that of the chief of staff.
Chiefs of staff occupy a crucial and indispensable position in the modern White House. They are primus inter pares, or first among equals, on the White House staff. No one person, except the president himself, holds more potential power in an administration. To a great extent, the success or failure of a president lays in the hands of the chief of staff, as it is his or her task to oversee the functioning of the White House process, to advise the president on important matters of politics and policy, and to protect the president from forces that could destroy his administration (Kernell and Popkin 1986). The best chiefs' footprints on the White House landscape are unmistakable. Too often, they receive little acclaim for a job well-done and must be content to let the president receive the accolades better meant for the chief of staff. On the other hand, chiefs of staff who fail in their duties are often removed from their job and are conspicuous in the organizational and public relations mess they leave behind.
The foundations of the modern chief of staff can be seen as far back as 1789 when George Washington hired Tobias Lear to be his private secretary, a position that evolved over time into the complex and important position of chief of staff today (Cohen 2000, chap. 2). Presidents in the premodern age, because of the relatively smaller size of the White House staff and executive branch as a whole, were free to organize and manage their White House with little assistance. However, in today's modern political system, presidents are unable to do so and must delegate responsibility to a chief of staff for handling the management function of the White House (Hult 1993; Kernell and Popkin 1986; Pfiffner 1996). The chief of staff has become a permanent fixture of the presidential landscape (Pfiffner 1996; Neustadt 1999). Every president since Richard Nixon has chosen to have a chief of staff at one time or another (see Table 1 for a listing of chiefs of staff in the modern era). Even Presidents Gerald Ford and Jimmy Carter, who began their administrations without a chief of staff, revamped their White House structures by incorporating a chief of staff system. All presidents, beginning with Ronald Reagan, have used a chief of staff system throughout their entire tenures.
The chiefs of staff covering the Nixon through Clinton administrations are listed in Table 1. One striking pattern is revealed from these data-that presidents go through more than one chief of staff during their tenure in the White House. In fact, the average professional life expectancy of a chief of staff during this period is less than two years. This suggests that this position is a difficult one that provides transitory stability at best.
Chiefs of staff assume three major roles in the modern White House: administrator of the White House organizational system, adviser to the president on political and policy matters, and guardian of the president's interests (Cohen 1997, 2000). For our purposes, we are interested in the first role-administrator of the White House organizational system. The chief of staff's special institutional position means that he or she helps shape the organizational structure of the White House. Thus, we posit that a chief of staff's style of management and general working relationship with the president serve as two important systemic variables that help explain the organizational structure of the White House. The next section presents a behavioral model of the White House's organizational structure.
The focus of our study is on the functioning of the White House's organizational structure. Thus we are not interested in explaining the formal organizational chart of the White House, but rather we are concerned with how the organizational structure works in practice, as has been the focus of past studies of the American presidency investigating the internal operation of the White House (Burke 1992; Hess 1988; Kernell and Popkin 1986; Patterson 1988; Walcott and Hult 1995). Simply, our approach for understanding the organizational structure of the White House is rooted in the behavior of both presidents and their chiefs of staff.
What explains the functional nature of organizational structure in the White House? Research on this subject maintains that both presidents and chiefs of staff are responsible for shaping the organizational structure of the White House (Cohen 2000; Kernell and Popkin 1986). Both individuals sit at the top of the White House's organizational hierarchy and both possess the power and authority to organize and change that structure. Because of their formal authority as chief executive, presidents have the power to organize the White House in ways they deem appropriate. At the same time, however, presidents also delegate authority to chiefs of staff. That is, a chief of staffs power and authority is informally derived from the president himself. Thus, both presidents and chiefs of staff, irrespective of which individuals are occupying these positions at any given point in time, invariably place limits to one another's behavior while acting for the presidency as an institution. Therefore, each separate actor, as well as the general working relationship between both actors, should be largely responsible for affecting the White House's organizational structure.
Figure 1 provides a graphic portrait of the theoretical mechanism by which we assert the functioning of the White House's organizational structure is determined. This diagram indicates that there are three systemic behavioral factors hypothesized to account for variations in the organizational structure of the White House in our theoretical model: presidential management style, chief of staff management style, and the general working relationship between these two institutional actors.
Specifically, we posit that the more a president concentrates on managing his own White House, the less hierarchical it will function as an organizational entity. A president who wants to be "in the loop" would not want his chief of staff acting as a filter; rather, such a president would gravitate toward a spokes-of-the-wheel system in which there were multiple access points to himself for diffusing power among individuals (e.g., Jimmy Carter, Bill Clinton). On the other hand, a president not concerned with the intricacies of White House management would have a tendency to delegate management authority (e.g., Ronald Reagan). We expect to find that in such a situation, the White House would operate in a much more hierarchical fashion with a chief of staff tightly controlling the reigns. The following testable hypothesis regarding the impact of presidents' management style on the White House's organizational structure is as follows:
Hypothesis 1: As a president's management style becomes increasingly activist, the organizational structure of the White House will function in a less hierarchical manner.
In principal-agent parlance, when the president (principal) is more willing to delegate authority to his chief of staff(agent), the more hierarchical the White House's organizational structure becomes in practice, holding all else constant. Thus, as the "supply" of authority presidents are willing to delegate to their chiefs of staff rises, the president will become an increasingly vertical actor in the way the White House functions as an organization.
In our second hypothesis, we assert that the chief of staff also has a vital role to perform. We surmise that the more a chief of staff has a controlling, or active, style of management, the more hierarchical that particular White House's organizational apparatus will function in practice. In general, one can assert that most chiefs of staff prefer a hands-on management style-if they didn't, they probably would be the wrong people for the job. Instead, it is more a matter of degree of control that they wish to exert through their post. In other words, the more dominant the chief of staff's management style, the more vertical the organizational structure of the White House will become in terms of its operation, ceteris paribus. The following testable hypothesis regarding the impact of chiefs of staff's management style on the organizational structure of the White House is as follows:
Hypothesis 2: As the chief of staff's management style becomes increasingly activist, the White House's organizational structure will function in a more hierarchical manner.
In addition, the working relationship between a president and his chief of staff is another key theoretical variable that should be crucial in determining the White House's organizational structure behavior. We expect that the dynamic of the general working relationship between the president and chief of staff will affect the resulting operation of the White House as an organization. Simply, the better the president and chief of staff work with one another, the more apt the president will be to delegate authority to the chief of staff and thus create a more hierarchical organizational structure in practice. Chiefs of staff have a natural inclination toward a hierarchical White House organization (Kernell and Popkin 1986). Not only does this increase their personal prestige; it also helps to augment their influence because most aspects of the political and policy process must pass through them. Thus, the better the chief of staff and president work together, and consequently the more that the president trusts his chief of staff, we expect that the more the White House will operate in a hierarchical fashion.
The following hypothesis pertains to the linkage between presidential-chief of staff general working relationship and functioning of the White House's organizational structure:
Hypothesis 3: The better the general working relationship between a president and chief of staff is, the White House's organizational structure will function in more a hierarchical manner.
Interpreted within a principal-agent framework, this hypothesis implies that greater congruence between the president (principal) and chief of staff(agent) will result in the operation of a more vertical organizational structure, all else being equal.
We also explore the possibility that individual differences will occur between administrations. For example, conventional wisdom holds that the White House's organizational structure functioned in a less hierarchical manner under Bush compared with Reagan (Burke 1992). We also examine whether the management styles of the president and chiefs of staff, as well as their working relationship, have a differential impact on the nature of the White House's organizational structure across presidential administrations and chiefs of staff, respectively. Evidence of such a differential effect will show that the organizational structure of the White House reflects the importance associated with the individual operating within the broader context of the Oval Office.
To test our hypotheses and competing theoretical perspectives, we employ individual-level survey data drawn from the Chief of Staff Project (COSP) Questionnaire, which focuses on the intraorganizational aspects of the presidency. This is a survey conducted by the first author of former White House staffers during the Reagan and Bush administrations at the special assistant level and above, as well as cabinet and deputy cabinet members (see Cohen 2000). The survey was administered via postal mail from October 1998 through January 1999.(n2) All survey instruments are measured on a conventional 7-point scale (see Appendix A for precise variable definitions) and provide a representative sample of individuals in various positions within the White House Office. The data cover the period with the following chiefs of staff during the Reagan administration: James Baker (1981-85), Donald Regan (1985-87), Howard Baker (1987-88), and Kenneth Duberstein (1988-89); and during the Bush administration: John Sununu (1989-91) and Samuel Skinner (1991-92).(n3)
Although survey data are a fruitful way of addressing our research question as to what determines the White House's organizational structure, some caveats must be noted before we proceed with our analysis. First, since we only have data from two presidential administrations, our ability to generalize to other presidents is limited. With this said, however, our subsequent empirical analysis will provide a conservative test of differences in the nature of organizational structure of the White House across presidents, since the gulf between Democratic and Republican administrations is viewed as greater than intrapartisan administration differences (Pfiffner 1996). Thus, we surmise that the differences in organizational structure between the Reagan and Bush administrations should be marginal compared to a pairwise analysis of other administrations.
In addition, given that our data are perceptual based from a survey sample, we must be cautious about representativeness of the sample of respondents and about whether a selection bias exists based on the different perceptions generated from those employed within the EOP versus those working in cabinet-level positions. Regarding the former issue, there is nothing in our sample to suggest that it is biased in a way that is not representative of the target population.(n4) Concerning the latter problem, response bias may exist because some individuals' positions are found within the White House, while others are not. Our preliminary analysis, however, reveals that this is not problematic for the purposes of this study based on our supplementary statistical findings displayed in Appendix B of this article.
There are a couple of key similarities when one compares the organizational aspects of the Reagan and Bush White House operations. Both presidents have been generally viewed as having a passive White House management style, especially in the area of domestic policy, consistent with Barber's (1985) typology. However, the sample mean values (arithmetic mean of sample) of the survey respondents concerning presidential management style reveal that Bush was typically more involved in White House management compared with Reagan, and that this difference was statistically significant at conventional levels (Arithmetic mean of sample[sub Presidential Management Style-Bush] = 3.44 and Arithmetic mean of sample[sub Presidential Management Style-Reagan] = 2.11; t-statistic = 6.44, p<.001). Not surprisingly, there also exists a nontrivial difference in the mean value of the respondents' view of the organizational structure of the White House under each administration, where the Reagan White House exhibits much greater hierarchical organizational behavior than the Bush White House (arithmetic mean of sample[sub Organizational Structure-Reagan] = 6.05 and arithmetic mean of sample[sub Organizational Structure-Bush] = 5.12; t-statistic = 4.36, p<.001). Thus, the structure of the White House as an organization under both presidents was rather similar in that both functioned in a pyramidal fashion, although the degree of hierarchy was somewhat less pronounced in the Bush administration.
At the same time, however, at an absolute level, respondents viewed these Republican presidents in tandem as leaning more toward the practice of a vertical organizational structure. This is consistent with Pfiffner's (1996) claim that Republican administrations prefer a hierarchical White House organizational structure with an assertive chief of staff. In addition, presidential management style appears to be much less "hands-on" compared with chiefs of staff's management style covered in this sample period (Arithmetic mean of sample[sub Presidential Management Style] = 2.48 and Arithmetic mean of sample[sub Chief of Staff Management Style = 5.39; t-statistic = -19.49, p<.001). This result implies that presidents delegated considerable authority to chiefs of staff in overseeing the execution of White House Office tasks during the Reagan and Bush presidencies (Burke 1992; Hart 1995; Pfiffner 1996).
Although informative, these tests are of limited utility because they do not control for other factors. Thus, we perform multivariate regression analysis as a more rigorous approach to test our hypotheses. There are two classes of regression models analyzed in this study. The first involves the use of nonstratified data that do not differentiate survey respondents by the chiefs of staff with whom they worked. The purpose here is to understand whether the operation of the White House's organizational structure under the Reagan and Bush presidencies is driven by (1) systemic patterns of presidential and chief of staff behavior irrespective of the individual president in office, (2) individual differences across presidents, (3) both (1) and (2), or (4) the interaction between the individual president residing in office and the institutional context within which he and his chief of staff operate within the broader organizational structure.
The second regression analysis is more revealing since it entails stratifying the survey respondents according to the individual chief of staff with whom they concurrently served so that there is a regression equation representing each chief of staff in our sample. This approach requires using a system estimation approach involving seemingly unrelated regressions (SURs) developed by Zellner (1962) to enhance the efficiency of the parameter estimates that ordinary least squares (OLS) fail to exploit.(n5) A major advantage of this approach is that it takes into account useful information in the residual covariance (correlation) across each chief of staff equation, which reflects the possibility that the behavioral functioning of the White House as an organization for a given chief of staff will be indirectly linked to this same phenomenon under a different chief of staff. Moreover, a systems estimation approach allows us to easily test coefficient restrictions across each of these equations to determine whether these exogenous variables possess the same substantive impact on the structure of White House organization across these various chiefs of staff. Thus, we can determine whether individual administration or chief of staff differences occur when assessing the impact of these systemic behavioral variables, irrespective of the individuals occupying in these positions.
The first set of regression analyses of the White House's organizational structure is displayed in Table 2. The results in the Behavioral Model show that as the president's management style becomes less active, the more hierarchical the organizational structure will become in practice. Conversely, as a chief of staff's management style becomes more active, the structure of the White House as an organization behaves in a more hierarchical fashion and less like a "spokes-of-the-wheel" structure. These findings in tandem make intuitive sense given the roles played by presidents and chiefs of staff. Presidents who are less willing to delegate authority will be more inclined to employ a "spokes-of-the wheel" (or less hierarchical) organizational structure, in which they are more involved in running their own White House. In principal-agent terms, this suggests that the principal (president) who is less inclined to delegate authority will behave in a more horizontal fashion within the White House's organizational structure. Chiefs of staff who have an active management style will typically be associated with an increasingly hierarchical organizational structure because the White House's internal processes must go through this position before reaching the president. In other words, the chief of staff functions as a "bottleneck" through which individuals and information must go through to get to the president. Furthermore, the better the general working relationship between the president and the chief of staff, the more hierarchical the organizational structure of the White House will operate in practice as presidents become more willing to function in a "vertically" structured White House.
Comparing the Behavioral and Individual Models reveals that although Bush preferred a less hierarchical approach than Reagan, the systemic importance associated with presidential and chief of staff management style and their general working relationship is noticeably more important than cross-administration differences in explaining the White House's organizational behavior. This finding is not terribly surprising. Specifically, although these presidents had different preferences for the organizational structure of the White House, they nonetheless did not differ as sharply as one may presume if the comparison were to be made between the highly dissimilar organizational characteristics of the Carter and Reagan administrations (Burke 1992; Campbell 1986; Hess 1988).(n6)
However, a combined model indicates that systemic variables and differences between the Reagan and Bush administrations are both important in explaining how the survey respondents viewed the functioning of the White House's organizational structure during their tenures. The model specification containing interactive terms between the systemic forces and individual administration differences (interactive model) reveals that both chief of staff style and the general working relationship between the chief of staff and the president exert a systematic effect on the organizational structure of the White House that is not predicated on the individual serving as chief executive. In other words, there is no conditional effect of chief of staff management style or presidential-chief of staff general working relationship on the White House's organizational structure predicated on differences between these administrations. Moreover, significant individual presidential administration effects disappear altogether. In only a single instance does the uniqueness of an individual presidential administration affect the behavior of the White House's organizational structure. Specifically, the survey data reveal that presidential style in managing the White House has a differential effect on its organizational structure between the Reagan and Bush administrations. This finding implies that Bush's relatively more "hands-on" management style resulted in the functioning of a more horizontal organizational structure of the White House than Reagan's. At a broader level, these empirical results indicate that presidents, not chiefs of staff, are able to account for variation in how these systemic factors affect the organizational structure of the White House. Although the systemic influences embedded into our behavioral model do matter in determining White House organizational behavior, the statistical results also show that differences across presidents are important.
These empirical results, however, may also be responsible for attenuating heterogeneity that reflects the varying impact of each systemic variable in our behavioral model across individual chiefs of staff in determining the operation of the White House's organizational structure. For instance, do these systemic effects vary across chiefs of staff? If so, then what implications does this have for managing the White House from an organizational perspective? We address this subject by testing our behavioral model by stratifying respondents according to the chief of staff that they worked with while serving in a given administration. This issue is a critical one to address given the "hands-on" management style associated with chiefs of staff relative to presidents and chiefs of staff's subsequent importance in explaining the behavior of the White House's organizational structure.
Statistical Results across Individual Chiefs of Staff
The results based on the analysis of individual chiefs of staff appear in Table 3. Presidential management style is significant in explaining the White House's organizational structure under James Baker (Reagan presidency) and for both John Sununu and Samuel Skinner's (Bush presidency) tenures as chiefs of staff. For instance, when Ronald Reagan was comfortable with his chief of staff during his first term, his management style was more hands-off, thus making the organizational structure of the White House behave in an increasingly hierarchical manner. A statistically significant positive relationship between chief of staff activism and degree of hierarchical organizational structure occurs in four of the six possible instances-James Baker, Howard Baker, Kenneth Duberstein, and Samuel Skinner. Although on the surface it appears that James Baker's role was limited to administration,(n7) his political and management ability enhanced his own status over time and he soon became primus inter pares (Burke 1992, 151; Pfiffner 1996, 26; Smith 1988, 308-17). As James Baker's power increased and as his management style became more hands-on, the White House began functioning more hierarchically with Baker acting as a bottleneck. In the cases of Howard Baker, Kenneth Duberstein, and Samuel Skinner, on the other hand, each of these chiefs was employed at unique times in their respective administrations. We suspect that Howard Baker's main purpose as chief of staff was to get the Reagan White House under control following the Iran-Contra scandal. This objective, coupled with the fact that Reagan administration personnel began to leave for other job opportunities, resulted in a vacuum that may have been instrumental for strengthening the impact of chief of staff management style on White House organizational behavior during his tenure. Kenneth Duberstein, who followed Howard Baker as chief of staff, served the last several months of Reagan's presidency. Thus, it should be no surprise that his management style had a proportionally greater effect on the resulting organizational structure of the White House since staffers were leaving in droves for secure employment in the private sector. Samuel Skinner served as chief of staff in the Bush administration during a very tough reelection campaign. With most White House staffers responsible, at least in part, to working toward the president's reelection, the opportunities for a chief of staff to exert a significant impact on the functioning of the White House's organization should have increased.
The working relationship between the president and chief of staff only exerts a statistically meaningful impact on the functioning of White House's organizational structure during James Baker and Donald Regan's tenures. The stronger the working relationship was between these chiefs of staff and Ronald Reagan, the more hierarchical the organizational structure of the White House became according to the respondents. In sum, these stratified results uncover considerable heterogeneity in the behavioral relationships across individual chiefs of staff that cannot be uncovered in our earlier tests, which do not allow for such explicit differences.
We perform chi-square (Chi[sup 2]) cross-equation coefficient equality restriction tests on the SUR system of regression equations estimated in Table 3. The results of these ancillary hypothesis tests are presented in Table 4. These tests allow us to determine whether presidential and chief of staff management styles and their general working relationship exert a differential effect on the organizational structure of the White House across individual chiefs of staff. These statistical tests are extremely valuable because central importance is attached to chiefs of staff in implementing the type of organizational structure on behalf of the president. These tests are performed for different chiefs of staff within a single administration and across both the Reagan and Bush administrations. The null hypothesis maintains that there should be no significant differential impact of presidential management style, chief of staff management style, and president-chief of staff general working relationship on White House's organizational behavior.
Although these tests indicate that the impact of presidential management style on the White House's organizational structure does not vary within either Reagan and Bush presidencies -Chi[sup 2] ~ (4): 5.82 (p = .21), it does appear that they are rather likely to vary across chiefs of staff across administrations-Chi[sup 2] ~ (5): 10.28 (p = .07). These results corroborate the findings from our previous analyses by clearly showing that the effect of presidential management style on the functioning of the White House's organizational structure reflects the importance of the individual president operating within the context of the institution. In other words, substantive differences occur only across the Reagan and Bush administrations regarding the impact of presidential activism on the White House's organizational behavior.(n8)
The chi-square test results clearly reject the null hypothesis for the chief of staff variables in both instances at p<.01. This indicates that the statistical relationship between chief of staff management style and the functioning of the White House's organizational structure varies across the individuals occupying this position. This finding underscores the importance of individuals who serve as chief of staff, an influence that cannot be gleaned from analyses that fail to make such distinctions and that lead us to falsely conclude that the impact of chief of staff management style is independent of the individual in that position.
Finally, the general working relationship between the president and his chief of staff may vary through time. It does not, however, result in a differential impact of this particular variable on the White House's organizational structure across individual chiefs of staff that is statistically meaningful. This result is true, irrespective of whether we examine across or within administrations in a multivariate context. Thus, while variations in the strength of the working relationship between presidents and chiefs of staff may exist, the general working relationship effects on the organizational structure of the White House are sufficiently stable so as not to induce a notable differential effect across chiefs of staff either within or between different presidential administrations.
Although numerous illuminating studies investigating the functioning of the White House exist (e.g., Burke 1992; Hart 1995; Hess 1988; Johnson 1974; Pfiffner 1996; Redford and McCulley 1986; Thompson 1992), there have been few theoretical-based systematic empirical quantitative studies that attempt to explain the organizational structure of the presidency in terms of its actual operation. We have attempted to address this subject by analyzing the behavior of the organizational presidency during the Reagan and Bush administrations using survey data from elites serving under each president. We find that the systemic factors contained in our behavioral model of White House organization are important predictors of the degree by which the White House operates in a horizontal fashion (i.e., "spokes-of-the-wheel") versus a vertical manner (i.e., hierarchical). Furthermore, our analysis of the elite survey responses leads us to infer that the behavioral component of the White House's organizational structure is an outgrowth of heterogeneous individuals operating under a broader institutional context. Specifically, we find that the individuals operating within an institutional context are important when analyzing the manner of how chiefs of staff and presidents shape the organizational structure of the White House. This conclusion is most consistent with the view that the interaction between the individual occupant and the office itself is vital for understanding the behavior of the presidency as an institution (Burke 1992; Feldman 1993; Hargrove 1993; Pika 1988). Thus, individual presidents can shape the institution as much as the institution constrains them. As Burke (1988) notes,
Presidents are not . . . simply at the mercy of the institution. . . . The presidency takes on a different coloration from administration to administration, from one set of staff advisers to another. Presidents and their staffs are by no means hostage to the institution; often they have been able to benefit from the positive resources it provides while deflecting or overcoming institutional forces that detract from their goals. (P. 373)
In other words, presidents are both shapers of, and are also shaped by, this institution. The results of our study suggest that presidents and chiefs of staff behave as individual actors operating within a larger institutional setting.
In addition, a critical subplot to this study has been to determine whether chiefs of staff determine the structure of White House organization. Research has shown that EOP units are relatively stable across presidential administrations (Kessel 1984), implying that individual-based models are not critical for understanding organizational structure. However, our empirical results suggest otherwise when one views the actual functioning of organizational structure in terms of the degree of hierarchy that is observed by officials employed in the White House. Furthermore, our statistical evidence clearly shows a differential effect for chief of staff management style on the organizational structure of the White House, whether one focuses within or across administrations. This finding suggests that chiefs of staff have varying capabilities in shaping the operational structure of the White House as an organization. Therefore, diverse management styles of individual chiefs of staff can create variations in the organizational structure within which presidents operate.
The common thread running throughout this study is that the presidency consists of both complex individuals and institutional features. Both individual and institutional considerations come into play in understanding the structure of presidential organization. This complexity manifests itself because both presidents and chiefs of staff have a variable, yet notable, influence on the White House's organizational structure. The cross-presidential and chief of staff differential effects observed in this study are intriguing when one considers (1) the similarities involving Republican administrations noted elsewhere (Pfiffner 1996) and (2) that the presidency has become fully institutionalized since the early 1970s, thus largely restricting the ability of individual action by these actors outside the realm of institutional behavior (Krause and Cohen 2000; Ragsdale and Theis 1997). Although one may not expect to observe behavior reflecting individual-level heterogeneity across presidents and their chiefs of staff for these two reasons, our data suggest otherwise. The major implication that can be drawn from our study is that research on the presidency requires an institution-based perspective; however, this need not come at the expense of investigating individual-level differences that do occur in the analysis of executive politics.
Although the Reagan and Bush administrations provide a rich empirical laboratory to investigate the nature of the White House's organizational behavior, interesting answers to this question can only be acquired by studying additional presidencies. This topic is an important one because the organizational behavior of presidential administrations will have an impact on presidential action. For example, one very promising direction for studying the presidency is to link its organizational features to its policy behavior and outcomes (Hult 1993; Light 1993). This line of inquiry, in turn, can give us answers in how variations in the structure of the White House as an organizational unit shape different presidential activities and subsequent policy outputs. This approach can yield fundamental insights into the presidency as it has for Congress, where the organization of political parties within Congress has been shown to affect policy outputs (e.g., Aldrich 1995; Cox and McCubbins 1993).
AUTHORS' NOTE: Equal authorship. The authors' names are listed in alphabetical order. We thank George Edwards and Brad Gomez for their many helpful comments on an earlier version of this article. All statistical analysis was performed using EVIEWS (version 3.1).
(n1.) By functional we refer to the actual operation or behavior of the organizational structure of the White House, as opposed to its technical placement on an organizational flow chart. Hence, our discussion of organizational structure will refer to its functional operation throughout the remainder of this article.
(n2.) The survey was sent to 404 individuals listed in the White House Office as well as select positions within the Executive Office of the President (Chair of the Council of Economic Advisers, Office of Management and Budget Director and Deputy Director, and the U.S. Representative to the United Nations) and the executive branch (cabinet and deputy cabinet members). The United States Government Manual, which is published annually and contains a listing of all upper-level employees in the White House and executive departments, was used as the primary source to determine survey recipients. This survey had a response rate of 27 percent (174) out of the 643 surveys administered. Individuals were sent a survey for each chief of staff with whom they worked.
(n3.) James Baker's tenure as chief of staff during the Bush administration is excluded from our survey since his five-month stint at the end of the presidential term almost solely pertained to reelection and the subsequent transition to the Clinton administration. Baker was transferred from his post as secretary of state for the explicit purpose of rejuvenating George Bush's floundering presidential reelection campaign (Broder 1992; Gergen 1992).
(n4.) The most represented groups in the sample were the ones that contained the largest number of staff positions in the Executive Office of the President (i.e., White House assistant, White House deputy assistant, and White House special assistant), while those represented in smaller numbers reflected positions with only a handful of personnel in each category (i.e., department secretary, deputy secretary, and White House/cabinet mixed). Moreover, the Reagan administration respondents outnumbered the Bush administration respondents by nearly a 2.5 to 1 ratio (n[sub Reagan] = 124; n[sub Bush] = 50), and their respective survey response rates were not noticeably different from one another (Reagan: 29 percent, Bush: 24 percent). For additional information on the survey sample, the interested reader can refer to Cohen (2000).
(n5.) The seemingly unrelated regressions (SUR) estimation approach allows one to estimate a system of equations via generalized least squares (GLS) that are related through its residual terms, and when none of the endogenous variables appear on the right-hand side of the other equations. SUR estimates yield greater efficiency if(1) each of the equations do not have the same, identical specification (i.e., right-hand-side variables) and (2) the cross-equation covariances do not equal zero. This gain in efficiency is especially important given the small sample sizes estimated in each of the regression equations in the individual chiefs of staff empirical model. A more detailed treatment of this technique can be found in a standard econometrics textbook such as Pindyck and Rubinfeld (1998, 358-61).
(n6.) John Kessel's (1984) study of the organizational structure of the White House shows that the relative importance of units within the Executive Office of the President (EOP) remains stable across the Carter and Reagan administrations. However, one cannot infer from Kessel's research that this necessarily translates into the fact that presidents desire the same organizational matrix nor the same relationship of subordinates to presidents in terms of organizational structure.
(n7.) In the case of James Baker, Chief of Staff Baker, Counselor to the President Edwin Meese, and Deputy Chief of Staff Michael Deaver were part of what became known as the White House Troika, a group that largely controlled the first Reagan administration. Although Baker was the only one of the three with the official designation of chief of staff, important duties that were traditionally the province of the chief of staff alone were initially divided among the troika with Baker in charge of administration, Meese responsible for domestic policy, and Deaver handling public relations, scheduling, and liaison between the West Wing staff and the First Lady (Burke 1992, 150-51; Hess 1988, 147; Pfiffner 1996, 25; see also Smith 1988, chap. 10).
(n8.) However, one must exhibit caution in interpreting this result because it pertains to only two presidential administrations.
Legend for Chart:
A - Chief of Staff
B - Tenure
C - President
D - Party
A B C D
H. R. Haldeman 1969-73 Nixon Republican
Alexander M. Haig Jr. 1973-74 Nixon Republican
Donald H. Rumsfeld 1974-75 Ford Republican
Richard M. Cheney 1975-77 Ford Republican
Hamilton Jordan 1979-80 Carter Democrat
Jack Watson 1980-81 Carter Democrat
James A. Baker III 1981-85 Reagan Republican
Donald T. Regan 1985-87 Reagan Republican
Howard H. Baker Jr. 1987-88 Reagan Republican
Kenneth Duberstein 1988-89 Reagan Republican
John H. Sununu 1989-91 Bush Republican
Samuel K. Skinner 1991-92 Bush Republican
James A. Baker III 1992-93 Bush Republican
Thomas F. McLarty III 1993-94 Clinton Democrat
Leon Paneaa 1994-96 Clinton Democrat
Erskine B. Bowles 1997-98 Clinton Democrat
John D. Podesta 1998-2000 Clinton Democrat
Source: Cohen (2000, Table 3.4).
Legend for Chart:
A - Variable
B - Behavioral
C - Individual
D - Combined
E - Interactive
A B C D
E
Constant 3.40[***] 6.05[***] 3.54[***]
(.61) (.11) (.61)
3.59[***]
(.85)
Presidential style -.32[***] -- -.26[***]
(.06) (.07)
-.13
(.08)
Chief of staff style .39[***] -- .37[***]
(.08) (.07)
.32[***]
(.09)
Working relationship .18[**] -- .17[**]
(.08) (.08)
.16[*]
(.10)
Administration dummy -- -.93[***] -.39[**]
(.21) (.19)
.71
(1.09)
Presidential Style x -- -- --
Administration Dummy -.29[**]
(.12)
Chief of Staff Style x -- -- --
Administration Dummy .12
(.18)
Working Relationship x -- -- --
Administration Dummy .09
(.14)
Adjusted R[sup 2] .32 .10 .35
.34
Standard error of estimate 1.10 1.26 1.09
(SEE)
1.08
White Test 1.96[*] 0.39 2.01[*]
[.08] [.53] [.06]
1.57[*]
[.10]
Wald F-statistic -- -- --
1.89
H[sub 0]: Beta[sub 5] =
Beta[sub 6] = Beta[sub 7] = 0 [.13]
n 171 173 171
171
Note: The numbers in parentheses in the behavioral, combined, and
interactive models are White heteroskedastic-consistent standard
errors and OLS standard errors for the individual model. The
numbers in brackets are probability levels. The Wald F-test
statistic: Beta[sub 5] = Beta[sub 6] = Beta[sub 7] = 0 simply
means that the joint effect of the interaction variables (i.e.,
Presidential Style x Administration Dummy, Chief of Staff Style x
Administration Dummy, and Working Relationship x Administration
Dummy) is equal to zero.
[a.] OLS = ordinary least squares.
[*] p < .10.
[**] p < .05.
[***] p < .01.
Legend for Chart:
A - Variable
B - Reagan Administration J. Baker
C - Reagan Administration D. Regan
D - Reagan Administration H. Baker
E - Reagan Administration K. Duberstein
F - Bush Administration J. Sununu
G - Bush Administration S. Skinner
A B C D
E F G
Constant 3.48[**] 3.72[***] 1.69
(1.59) (1.07) (2.73)
1.55[*] 4.98[***] 2.61[***]
(.88) (.92) (.68)
Presidential style -.38[**] .15 -.25
(Gamma[sub i]) (.19) (.17) (.25)
.02 -.30[***] -.26[**]
(.12) (.08) (.12)
Chief of staff style .21[*] .19 .33[*]
(Delta[sub i] (.13) (.14) (.18)
.57[***] .14 .78[***]
(.11) (.11) (.14)
Working relationship .34[*] .24[*] .48
(Theta[sub i]) (.20) (.13) (.38)
.17 .15 -.21
(.13) (.17) (.19)
Adjusted R[sup 2] .16 .05 .05
.52 .26 .67
Standard error of estimate
(SEE) 1.09 1.07 1.36
.82 .79 .89
n 38 32 33
22 31 17
Rho Epsilon[sub JB] -- -- --
-- -- --
Rho Epsilon[sub DR] .03 -- --
-- -- --
Rho Epsilon[sub HB] -.13 -.05 --
-- -- --
Rho Epsilon[sub KD] .11 .19 -.10
-- -- --
Rho Epsilon[sub JS] -.38[**] -.01 .41[**]
.10 -- --
Rho Epsilon[sub SS] .06 -.20 .02
-.36[**] .03 --
Note: The numbers in parentheses are standard errors. The
determinant residual covariance equals. 176. The bottom of this
table displays the contemporaneous residual correlation matrix
among each of these regression equations estimated via the
seemingly unrelated regression (SUR) technique. The alphabetical
subscripts refer to the initials of the chief of staff in
question.
[a.] SUR = seemingly unrelated regression; GLS = generalized
least squares.
p < .10.
[**] p < .05.
[***] p < .01.
Legend for Chart:
A - Type of Chi[sup 2] Test Variable
B - Across Administrations Null Hypothesis: No Variation in
Effect of Variable on White House Organizational Structure
across Chiefs of Staff
C - Within Administrations Null Hypothesis: No Variation in
Effect of Variable on White House Organizational Structure
across Chiefs of Staff
A B C
Presidential style
Gamma[sub JB] = Gamma[sub DR] =
Gamma[sub HB] = Gamma[sub KD] =
Gamma[sub JS] = Gamma[sub SS] 10.28[*] --
[.07]
Gamma[sub JB] = Gamma[sub DR] =
Gamma[sub HB] = Gamma[sub KD];
Gamma[sub JS] = Gamma[sub SS] -- 5.82
[.21]
Chief of staff style
Delta[sub JB] = Delta[sub DR] =
Delta[sub HB] = Delta[sub KD] =
Delta[sub JS] = Delta[sub SS] 20.27[***] --
[.00]
Gamma[sub JB] = Gamma[sub DR] =
Gamma[sub HB] = Gamma[sub KD];
Gamma[sub JS] = Gamma[sub SS] -- 20.22[***]
[.00]
Working relationship
Theta[sub JB] = Theta[sub DR] =
Theta[sub HB] = Theta[sub KD] =
Theta[sub JS] = Theta[sub SS] 5.86 --
[.32]
Theta[sub JB] = Theta[sub DR] =
Theta[sub HB] = Theta[sub KD];
Theta[sub JS] = Theta[sub SS] -- 2.63
[.62]
Note: The numbers in brackets are probability levels. The
alphabetical subscripts refer to the initials of the chief of
staff in question.
[*] p < .10.
[**] p < .01.
DIAGRAM: FIGURE 1. A Behavioral Model of the White House's Organizational Structure.
Aldrich, John H. 1995. Why parties? The origins and transformation of party politics in America. Chicago: University of Chicago Press.
Barber, James David. 1985. The presidential character: Predicting performance in the White House. 3d ed. Englewood Cliffs, NJ: Prentice Hall.
Broder, David S. 1992. Baker team settles in at White House. Washington Post, August 25, A5.
Buchanan, Bruce. 1990. Constrained diversity: The organizational demands of the presidency. Presidential Studies Quarterly 20 (fall): 791-822.
Burke, John P. 1988. The institutional presidency. In The presidency and the political system, 2d ed., edited by Michael Nelson. Washington, DC: CQ Press.
Burke, John P. 1992. The institutional presidency. Baltimore, MD: Johns Hopkins University Press.
Campbell, Colin. 1986. Managing the presidency: Carter, Reagan, and the search for executive harmony. Pittsburgh, PA: University of Pittsburgh Press.
Cohen, David B. 1997. George Bush's vicar of the West Wing: John Sununu as White House chief of staff. Congress & the Presidency 24 (spring): 37-59.
-----. 2000. Effectiveness and influence in the White House: The chief of staff during the Reagan and Bush years. Ph.D. diss., University of South Carolina.
Cox, Gary W., and Mathew D. McCubbins. 1993. Legislative Leviathan: Party government in the House. Berkeley: University of California Press.
Feldman, Martha S. 1993. Organization theory and the presidency. In Researching the presidency: Vital questions, new approaches, edited by George C. Edwards III, John H. Kessel, and Bert A. Rockman. Pittsburgh, PA: University of Pittsburgh Press.
George, Alexander L. 1980. Presidential decisionmaking in foreign policy: The effective use of information and advice. Boulder, CO: Westview.
Gergen, David. 1992. Enter the new deputy president. U.S. News & World Report, August 24, 42.
Hargrove, Erwin C. 1993. Presidential personality and leadership style. In Researching the presidency: Vital questions, new approaches, edited by George C. Edwards III, John H. Kessel, and Bert A. Rockman. Pittsburgh, PA: University of Pittsburgh Press.
Hart, John. 1995. The presidential branch: From Washington to Clinton. 2d ed. Chatham, NJ: Chatham House.
Henderson, Phillip G. 1988. Managing the presidency: The Eisenhower legacy-From Kennedy to Reagan. Boulder, CO: Westview.
Hess, Stephen. 1988. Organizing the presidency. 2d ed. Washington, DC: Brookings Institution.
Hult, Karen M. 1993. Advising the president. In Researching the presidency: Vital questions, new approaches, edited by George C. Edwards III, John H. Kessel, and Bert A. Rockman. Pittsburgh, PA: University of Pittsburgh Press.
Johnson, Richard Tanner. 1974. Managing the White House: An intimate study of the presidency. New York: Harper & Row.
Kernell, Samuel, and Samuel L. Popkin, eds. 1986. Chief of staff: Twenty-five years of managing the presidency. Los Angeles: University of California Press.
Kessel, John H. 1983. The structures of the Carter White House. American Journal of Political Science 27 (August): 431-63.
-----. 1984. The structures of the Reagan White House. American Journal of Political Science 28 (May): 231-58.
Krause, George A. 2001. The decline in presidential domestic policy making: An organizational perspective. In Reinventing the presidency, edited by George C. Edwards III and Kenneth J. Meier. New York: Chatham House. Forthcoming
Krause, George A., and Jeffrey E. Cohen. 2000. Opportunity, constraints, and the development of the institutional presidency: The issuance of executive orders, 1939-1996. Journal of Politics 62 (February): 88-114.
Light, Paul C. 1993. Presidential policy making. In Researching the presidency: Vital questions, new approaches, edited by George C. Edwards III, John H. Kessel, and Bert A. Rockman. Pittsburgh, PA: University of Pittsburgh Press.
Neustadt, Richard E. 1999. Does the White House need a strong chief of staff? In The managerial presidency, 2d ed., edited by James P. Pfiffner. College Station: Texas A&M Press.
Patterson, Bradley H. 1988. The ring of power: The White House staff and its expanding role in government. New York: Basic Books.
Pfiffner, James P. 1996. The strategic presidency: Hitting the ground running. 2d ed. Lawrence: University of Kansas Press.
-----, ed. 1999. The managerial presidency. 2d ed. College Station: Texas A&M Press.
Pika, Joseph A. 1988. Management style and the organizational matrix: Studying White House operations. Administration & Society 20 (May): 3-29.
Pindyck, Robert S., and Daniel L. Rubinfeld. 1998. Econometric models and economic forecasts. 4th ed. New York: McGraw-Hill.
Porter, Roger B. [1980] 1988. Presidential decision mating: The economic policy board. Reprint, New York: Cambridge University Press.
Ragsdale, Lyn, and John J. Theis III. 1997. The institutionalization of the American presidency, 1924-92. American Journal of Political Science 41 (October): 1280-1318.
Redford, Emmette S., and Richard T. McCulley. 1986. White House operations: The Johnson presidency. Austin: University of Texas Press.
Smith, Hedrick. (1988). The power game: How Washington really works. New York: Ballantine.
Thompson, Robert J. 1992. Contrasting models of White House staff organization: The Eisenhower, Ford, and Carter experiences. Congress & the Presidency 19 (autumn): 113-36.
Walcott, Charles E., and Karen M. Hult. 1995. Governing the White House: From Hoover through LBJ. Lawrence: University of Kansas Press.
Warshaw, Shirley Anne. 1997. The domestic presidency: Policy making in the White House. Boston: Allyn and Bacon.
Zellner, Arnold. 1962. An efficient method of estimating seemingly unrelated regressions and tests for aggregation bias. Journal of the American Statistical Association 57:348-68.
On a scale of 1 to 7, how would you characterize the organization of the White House-hierarchical, with a strong chief of staff in charge, a spokes-of-the-wheel system in which the president operated as his own chief of staff, or something in between? Spokes of the wheel Hierarchical 1 2 3 4 5 6 7
On a scale of 1 to 7, how would you characterize the president's general style of managing the White House (i.e., how much time and involvement was spent by the president on White House management issues)? Delegating Hands-On 1 2 3 4 5 6 7
Chief of staff management style
On a scale of 1 to 7, how would you characterize the chief of staffs general style of managing the White House (i.e., how much time and involvement was spent by the chief of staff on White House management issues)? Delegating Hands-On 1 2 3 4 5 6 7
On a scale of 1 to 7, how would you characterize the chief of staffs general working relationship with the president? Poor Excellent 1 2 3 4 5 6 7
In addition to the statistical analysis we present in the text of this article, we performed additional analyses to ensure that our results were not sensitive based on the composition of respondents included in our sample. Thus, response bias may exist given that some individuals surveyed have positions within the White House, while others do not. We incorporate a dummy variable in these supplementary specifications that takes on a value of one for individuals located outside the Executive Office of President (EOP)-that is, cabinet departments-and equals zero for those individuals serving within the EOP. This coefficient is negative, suggesting that those individuals outside EOP view the White House's organizational structure as operating in a slightly less hierarchical manner compared with those individuals surveyed working within EOP. The reason for this differential perception is perhaps because cabinet and deputy cabinet members do not wish to, or need to, see the president as often as EOP staff does. Thus, when a cabinet member asks to meet with the president, a relatively uncommon occurrence compared with higher-level EOP staff, their request is often granted, especially if they do not abuse this privilege, thus leaving the cabinet member with the impression that they have open and unfettered access to the president when this is not necessarily the case. However, this difference is statistically trivial at conventional levels.
More important, the results of the other relatonships are substantively identical to those presented in Table B. 1 containing the unstratified empirical tests in every way except that the president-chief of staff general working relationship variable in the interactive model barely misses statistical significance at p<.10 in the supplementary analysis found in Table B. 1. The supplementary analysis involving the statistical tests examining across chief of staff variation displayed in Table B.2 demonstrates in only a few instances where these results diverge from being substantively identical to those presented in Table 3. Specifically, the presidential management style coefficient is significant at p approximately equal to .06, and chief of staff management style just misses obtaining statistical significance at p<.10 in the James Baker equation in this table. The president-chief of staff general working relationship coefficient's probability level slightly declines in the Donald Regan equation when the institutional position of the survey respondent is explicitly considered as an independent variable. The chief of staff management style coefficient for the Howard Baker equation is much larger and significant at a higher level of confidence when the institutional position of the survey respondent is accounted for in the model specification. Finally, inspection of the Howard Baker equation reveals that the institutional position of the survey respondent matters in determining their view of the White House's organizational structure. Those outside EOP (cabinet-level officials) will possess a less hierarchical view of the White House's organizational behavior than those individuals surveyed who served within EOP. Table B.3 reveals that the Wald Chi[sup 2] cross-equation coefficient equality tests are substantively identical to those listed in Table 4, except that the p-value drops from .07 to .03 for the across administration-based test, which inspects whether presidential management style has a similar effect on the White House's organizational behavior across individual chiefs of staff when the institutional position of the survey respondent is incorporated as an independent variable.
Legend for Chart:
A - Variable
B - Behavioral
C - Individual
D - Combined
E - Interactive
A B C D
E
Constant 3.40[***] 6.06[***] 3.53[***]
(.61) (.11) (.61)
3.62[***]
(.86)
Presidential style -.31[***] -- -.38[***]
(.07) (.08)
-.12
(.08)
Chief of staff style .39[***] -- .43[***]
(.08) (.07)
.32[***]
(.09)
Working relationship .18[**] -- .17[**]
(.08) (.08)
.16
(.11)
Administration dummy -- -.92[***] -.38[**]
(.22) (.19)
.77
(1.10)
Presidential Style x -- -- --
Administration Dummy -.29[**]
(.12)
Chief of Staff Style x -- -- --
Administration Dummy
.13
(.18)
Working Relationship x -- -- --
Administration Dummy
.09
(.14)
Institutional position of -.17 -.09 -.13
survey respondent (.32) (.38) -.33
-.18
(.34)
Adjusted R[sup 2] .32 .09 .33
.34
Standard error of estimate 1.10 1.26 1.09
(SEE)
1.08
White Test 2.87[***] 4.66[**] 3.09[***]
[.00] [.01] [.00]
2.10[**]
[.02]
Wald F-statistic -- -- --
1.95
H[sub 0] = Beta[sub 5] =
Beta[sub 6] = Beta[sub 7] = 0 [.12]
n 171 173 171
171
Note: The numbers inside parentheses are White
heteroskedastic-consistent standard errors. Probability levels
are listed in brackets. The Wald F-statistic restricting the
coefficients of the institutional variables to zero in the
interactive model is statistically significant at conventional
levels (F-statistic = 8.80, p < .001). The Wald F-test statistic:
Beta[sub 5] = Beta[sub 6] = Beta[sub 7] = 0 simply means that the
joint effect of the interaction variables (i.e., Presidential
Style x Administration Dummy, Chief of Staff Style x
Administration Dummy, and Working Relationship x Administration
Dummy) is equal to zero. [a.] OLS = ordinary least squares.
[**] p < .05.
[***] p < .01.
Legend for Chart:
A - Variable
B - Reagan Administration J. Baker
C - Reagan Administration D. Regan
D - Reagan Administration H. Baker
E - Reagan Administration K. Duberstein
F - Bush Administration J. Sununu
G - Bush Administration S. Skinner
A B C D
E F G
Constant 3.45[**] 3.23[***] 2.06
(1.59) (1.10) (2.64)
1.57[*] 4.69[***] 2.65[***]
(.87) (.88) (.68)
Presidential style -.37[*] .19 -.20
(Gamma[sub i]) (.19) (.17) (.24)
.07 -.30[***] -.26[**]
(.14) (.08) (.12)
Chief of staff style .2O .21 .50[***]
(Delta[sub i]) (.13) (.13) (.19)
.57[***] .14 .77[***]
(.11) (.11) (.14)
Working relationship .35[*] .28[**] .31
(Theta[sub i]) (.20) (.13) (.37)
.15 .20 -.20
(.13) (.17) (.19)
Institutional position of
survey respondent 0.30 .60 -1.82[**]
(.44) (.53) (.75)
-.62 .12 -.19
(.81) (.28) (.57)
Adjusted R[sup 2] .15 .04 .06
.49 .25 .65
Standard error of estimate
(SEE) 1.10 1.07 1.38
.84 .80 .92
n 38 32 33
22 31 17
Rho Epsilon[sub JB] -- -- --
-- -- --
Rho Epsilon[sub DR] .05 -- --
-- -- --
Rho Epsilon[sub HB] -.14 .05 --
-- -- --
Rho Epsilon[sub KD] .13 .12 -.12
-- -- --
Rho Epsilon[sub JS] -.37[**] -.06 .53[***]
-.11 -- --
Rho Epsilon[sub SS] .04 .20 .07
-.37[**] .06 --
Note: The determinant residual covariance equals .137. The bottom
of this table displays the contemporaneous residual correlation
matrix among each of these regression equations estimated via the
seemingly unrelated regression (SUR) technique. The alphabetical
subscripts refer to the initials of the chief of staff in
question. SUR = seemingly unrelated regression; GLS = generalized
least squares.
[*] p < .10.
[**] p < .05.
[***] p < .01.
Legend for Chart:
A - Type Of Chi[sup 2] Test Variable
B - Across Administrations Null Hypothesis: No Variation in
Effect of Variable on White House Organizational Structure
acrossChiefs of Staff
C - Within Administrations Null Hypothesis: No Variation in
Effect of Variable on White House Organizational Structure
across Chiefs of Staff
A C B
Presidential style
Gamma[sub JB] = Gamma[sub DR] =
Gamma[sub HB] = Gamma[sub KD] =
Gamma[sub JS] = Gamma[sub SS] -- 12.03[**]
[.03]
Gamma[sub JB] = Gamma[sub DR] =
Gamma[sub HB] = Gamma[sub KD];
Gamma[sub JS] = Gamma[sub SS] 6.26 --
[.18]
Chief of staff style
Delta[sub JB] = Delta[sub DR] =
Delta[sub HB] = Delta[sub KD] =
Delta[sub JS] = Delta[sub SS] -- 20.30[***]
[.00]
Gamma[sub JB] = Gamma[sub DR] =
Gamma[sub HB] = Gamma[sub KD];
Gamma[sub JS] = Gamma[sub SS] 20.30[***] --
[.00]
Working relationship
Theta[sub JB] = Theta[sub DR] =
Theta[sub HB] = Theta[sub KD] =
Theta[sub JS] = Theta[sub SS] -- 5.62
[.35]
Theta[sub JB] = Theta[sub DR] =
Theta[sub HB] = Theta[sub KD];
Theta[sub JS] = Theta[sub SS] 3.17 --
[.53]
Note: The numbers in parenthesis are probability levels. The
alphabetical subscripts refer to the initials of the chief of
staff in question.
[**] p < .05.
[***] p < .01.
Overall, these results indicate that the statistical findings
presented in the text of the article are not contaminated by who
is responding to the survey questionnaire. Because our sample
does not hint at such a bias that is not representative of the
target population, this supplementary analysis corroborates the
content discussed in Footnote 4.
~~~~~~~~
By David B. Cohen, University of Akron and George A. Krause, University of South Carolina
David B. Cohen is an assistant professor in the Department of Political Science at the University of Akron. His research on executive politics has been published in American Politics Quarterly, Congress & the Presidency, Presidential Studies Quarterly, and Southeastern Political Review. His primary areas of interest are executive politics, Congress, and U.S. foreign policy making
George A. Krause is an associate professor of political science in the Department of Government and International Studies at the University of South Carolina. He is the author of A Two-Way Street: The Institutional Dynamics of the Modern Administrative State (University of Pittsburgh Press, 1999) and of articles in journals including the American Journal of Political Science, the Journal of Politics, the British Journal of Political Science, Political Analysis, Public Choice, and Legislative Studies Quarterly.