Organizational Variables in an Air Force Program/Project Environment

Introduction

The technological and organizational complexity involved in the design, development, and introduction of new products has increased significantly in recent years. One of the major consequences of this accelerating sophistication has been its impact upon the management systems required to convert new and often uncertain technology into operating systems. Extensive innovation is required to provide a management system capable of consolidating and controlling the wide range of specialized skills needed to develop modern Air Force weapon systems. The principal result has been the development of the project or program management concept [2].

The program management concept, applying project management theory, has been adopted by the Department of Defense and the Air Force to cope with the variety of managerial challenges inherent in acquiring weapon systems. A System Program Office (SPO) has been established to bring a particular weapon system or major end item into operational use whenever anticipated research and development costs exceed $75 million or the expected cost of production exceeds $300 million [27],

In recent years, considerable attention has been directed toward improving the productivity of such SPO organizations, including considerations of both effectiveness (achievement of desired result) and efficiency (minimum resource consumption). Management as well as technical factors are involved. As General Bernard A. Shriever has stated,

Studies to improve the productivity of such organizations must involve not only a variety of both structural and behavioral variables, but must also consider in detail their interrelationships. While many studies have investigated the effects of individual organizational variables on the productivity of Air Force SPOs [7] [8] [9] [12] [14] [19] [23] [25], little effort has been made to synthesize these findings and define the pattern of relationships among these variables. This study was designed to (1) synthesize prior findings relative to nine major organizational variables into a more comprehensive perspective of program organizations as they progress through the project life cycle, and (2) through the synthesis process, create new information about the casual relationships involved to supplement the prior findings.

Preview

The variables program/project phase, organizational size, level of bureaucracy, organizational climate, role conflict, role ambiguity, role stress, and conflict intensity were identified from the prior studies as having important influences on the project organization, and they formed the basis for this study. Using the definitions presented in Table 1 as a guide, a comprehensive review was conducted to identify casual relationships that would be supported in the literature. While space prohibits a complete report of the literature review, Figure 1 presents a model of the cause and effect relationships found and includes the sources which define or imply the relationships. For example, Holtz [10] supported the view that the nature of tasks associated with each phase of a project life cycle is the primary determinant of the project’s organizational structure (level of bureaucracy), while Pugh et al. found a strong correlation between size and the structuring of activities to include standardization of functions, formalization of procedures, and specialization of roles.

TABLE 1 – DEFINITION OF ORGANIZATIONAL VARIABLES

Proposed Causal Relationships Among Organizational Variables

FIGURE 1

The pattern of relationships exhibited in the model shown in Figure 1 illustrates the close association among the structural variables (program phase, organizational size, tenure, and level of bureaucracy) on the one hand, and among the behavioral variables (role conflict, role ambiguity, role stress, and conflict intensity) on the other, as expressed in the literature. Organizational climate provided the connecting link between the structural and behavioral variables in the model. Although not directly investigated, productivity was included in the model to illustrate its position relative to the organizational variables under investigation in this research effort. This model provided the basis for the research methodology and subsequent data analysis involved in this study.

Methodology

A series of previous Air Force studies investigated several of these variables, and each variable was examined by at least one previous study. However, these studies involved different data-producing samples collected independently and at different times over a two-year period. As such, these findings were not directly relatable and attempts to synthesize the results could not be statistically conclusive However, a number of apparent, interesting and useful relationships were noted in the data collected and have been previously reported [1].

A sample of 131 individuals was selected from a population of military and civilian program managers assigned to 12 individual SPOs within the Aeronautical Systems Division of Air Force Systems Command. The sample was designed to replicate the data sources of the previous studies and to provide directly comparable measures of the variables. The new sample was stratified into four life-cycle categories which corresponded to the generally accepted life-cycle stages of a civilian project (see Table 2). Each of these life-cycle categories had approximately equal sub-sample representation. The new data was gathered using a composite questionnaire developed from appropriate portions of instruments used and validated in four of the previous studies. The questionnaire provided a measure for the program managers’ perceptions of each variable of interest to this study. The new data was then compared to the data generated in previous studies using graphical and statistical techniques. This initial analysis indicated reasonable consistency in the variable measures between the new data and that collected by previous research teams, providing a partial validation of the instrument and supporting the previous study’s findings. It also indicated that the replication was valid and provided a basis for further analysis to synthesize the previous results.

TABLE 2 CATEGORIZATION OF PROJECT LIFE-CYCLE PHASES

The vehicle used to synthesize the prior findings was the proposed causal model shown in Figure 1. The new data were evaluated using the statistical technique of path analysis to determine the validity and statistical significance of the causal relationships proposed in the model. Path analysis was originally introduced by Wright and has been popularized by Blalock [4] [5], and by Duncan [6] in the social sciences. Wright [29] [30] used path coefficients as early as 1918 and expounded upon the path analysis techniques in a series of articles dating from the early 1920s. A detailed discussion of path analysis is well beyond the scope of this paper. It is assumed, however, that the reader is conversant with regression analysis, which is the basis for the path analysis technique. The interested reader will find a good introductory summary of path analysis concepts in Nie, et al. [17], and a more detailed presentation of the topic in the several works by Blaylock [4] [5] and by Duncan [6] referenced elsewhere in this article.

Path analysis is a method of decomposing and aiding the interpretation of linear relationships among a set of variables by assuming that (1) a (weak) causal order among these variables is known or can reasonably be assumed, and (2) the relationships among these variables are causally closed [17]. Basically the assumption of weak causal ordering postulates that, given a pair of variables Xⁱ and X^j, a weak order such that Xⁱ is a cause of X^j is established if it is assumed or known that Xⁱ may affect X^j, but X^j cannot affect Xⁱ. This directional assumption does not require Xⁱ to be a cause of X^j. Causal closure assumes that, given a bivariate covariation between, say, X₁ and X₂, and a known weak causal ordering, say X₁ is a cause of X₂, the observed covariation between X₁ and X₂ may be due (1) solely to the causal dependence of X₂ on X₁ (2) to their mutual dependence on some outside variable(s), or (3) to the combination of the preceding two [17, pp. 384-385].

Of course, the basic assumptions of linear regressions concerning the error components are also operative; that is, that the error terms are independent, identically and normally distributed, they have an expected value equal to zero, and a constant variance (homoscedasticity). Path analysis, however, is primarily a technique for working out the logical consequences of the first two cited assumptions.

Path analysis uses both path (or causal) diagrams and linear regression equations to represent the proposed system of relationships among the variables being studied. If the model contains n explicit variables, then (n-l) regression equations must be solved to analyze the model, one equation for each of the proposed relationships between two variables. Ordinary F tests for individual regression coefficients are commonly used to examine the statistical significance of each proposed causal path. In the path diagram (or model), assumptions about the direction of the causal relationships are explicitly indicated by arrows leading from each determining variable to each variable dependent on it. For example, in Figure 1 the relationship between level of bureaucracy and organizational climate is a function of level of bureaucracy and all other (extraneous) factors which are not directly investigated. Path analysis procedures determine the amount of variation in the “driven” variable that is attributable to (1) the “causing” variable, here level of bureaucracy, and (2) all other causes.

This research was specifically designed to meet the assumptions of path analysis:

1. Necessary weak causal relationships among the variables were developed through an extensive review of the literature.

2. The causal relationships were grouped into a closed causal model as presented in Figure 1.

3. The basic assumptions of regression analysis held, by instrument and sample design.

The path analysis was used to test the model, determine the strength of the relationships derived from the literature and, if warranted by the data, modify the model to more accurately reflect the relationships reported by the respondents in the data. The reader should be cautioned that the identification of a causal structure does not prove causal relationships, but it does allow researchers to draw inferences which serve as a basis for recommending both managerial actions and further research.

Analysis and Discussion

The path analysis procedure produced the effect coefficients indicated in Figure 2 for the hypothesized model. Note that project phase is not a continuous variable, and was therefore analyzed using the regression technique of dummy variables. Thus, while individual path coefficients were calculated for each phase of the project life cycle, the important result for the structural variables is the overall effect of project phase on level of bureaucracy across the life cycle phases. Note that this effect can occur either directly or indirectly through organization size and/or tenure. Similarly, the important result for the behavioral variables is the overall effect of organizational climate on role stress. In this case, however, that overall effect can occur only indirectly through either role conflict or role ambiguity. The relationship between role stress and conflict intensity was also quite pronounced. All of these relationships were tested at α = .01 and are considered highly significant. The relationship between level of bureaucracy and organizational climate, however, could not be supported.

FIGURE 2

Proposed Causal Model with Path Analysis Variables Identified

*Note: Project Phase is a noncontinuous variable using dummy variables in the regression analysis. Thus C₆₉, C₇₉, and C₈₉ are different for each phase of the project life cycle. The important finding is the overall effect across phases indicated by the total effect of X₉ on X₆, C₆₉ = .5713.

The lack of statistical significance for the causal path between level of bureaucracy and organizational climate required the model be reconsidered. In evaluating the data collected for the variable level of bureaucracy, it was found that (1) they were not consistent with that obtained in previous studies across life cycle phases, and (2) this portion of the data collection instrument demonstrated only face validity and had not been widely used. It was therefore decided that level of bureaurcracy should be dropped from the model. Proceeding with the theory that organizational climate could serve as a conceptual bridge between the structural variables and the human or behavioral aspects of a program organization, the researchers revised the causal model (See Figure 3). The revised model directly relates project phase, tenure, and organizational size to organizational climate. The path analysis of the revised regression equations demonstrated that the X₅X₈ and the X₅X₇ causal paths were not statistically significant. Even after eliminating these indirect paths, however, the overall effect of project phase on organizational climate was a very high .85, significant at well above α = .01. This revised model thus provides a much better explanation of the data collected than the original hypothesized model.

FIGURE 3

Revised Proposed Causal Model

*Note: Project Phase is a noncontinuous variable using dummy variables in the regression analysis. Thus, C₅₉, C₇₉ and C₈₉ are different for each phase of the project life cycle. The important finding is the overall effect across phases indicated by the total effect of X₉ on X₅, C₅₉ = .85.

Of particular interest are the strong inverse relationships between organizational climate and both role ambiguity and role conflict. Since role ambiguity and role conflict sum to role stress by definition, it can be inferred that role stress also has a strong (inverse) relationship with organizational climate. Thus, perceived levels of role ambiguity, role conflict and role stress tend to decrease as improvement in organizational climate is perceived. Although these relationships appear to be logical, the literature did not directly address organizational climate in relation to role ambiguity and role stress. Therefore, such strong correlations (strong for sociological data) among these variables were not anticipated by the researchers.

Conclusions

Path analysis is a detailed and complex statistical procedure. It has not been possible in the confines of this paper to document all of the rigorous quantitative and analytical procedures that go into the path analysis structure. Such documentation is available in the complete study [18]. The results presented here, however, can have significant implications for the future directions of project management operations and project management research.

Based upon the present findings it is evident that structural variables have a significant effect on behavioral outcomes. Although behavioral variables are difficult if not impossible to manage directly, project managers and their superiors do control many of the structural factors which may influence project success. The model developed in this study clearly indicates that if certain structural conditions are known, then behavioral consequences can be predicted with some degree of certainty. Therefore, by controlling these structural variables, program managers may be able to indirectly influence many of the behavioral variables common to their project organizations. A brief reexamination of the causal model will help illustrate this point.

Tenure and assignment policies have long been a subject for debate in project management. Civilian industry generally recognizes that people can contribute more to an organization after they have been in a position long enough to understand its purpose and the details of its function. However, it has also been argued that increased tenure leads to increased functionalization of tasks and less dependence on other organizational personnel for task accomplishment, just the opposite of what good project management was designed to achieve.

The results of this study support the latter argument. The inverse relationships exhibited between program phase and organizational climate indicate that over time a project manager tends to rely less upon a participative or consultative management style and more upon an authoritative and independent approach to task accomplishment. Whether this tendency reflects a greater awareness of job requirements and less dependency on the expertise of others, or a tendency for project managers to withdraw from the goals and activities of the project over time, cannot be determined from the available data. In any event, the inverse relationships between climate and role conflict and between climate and role ambiguity suggest that a more authoritative, independent approach to task accomplishment leads to greater perceived levels of role conflict and role ambiguity, and consequently role stress. Further, the direct relationship between role stress and conflict intensity suggests that as stress increases, a project manager perceives a greater incidence of conflict situations emanating from program activities and organizational participants.

This sequence of relationships has a certain logical appeal when one considers the dynamic nature of the project environment. Although task functionalization and independent action among the components of an organization may work well in a stable environment, such an approach is not congruent with the complex and ever-changing requirements characteristic of modern projects. A project manager who tries to routinize his activities may run the risk of losing touch with current project objectives as these objectives evolve. Further, functionalization in one area of a project office may hamper complete integration of total organizational activities, thereby increasing interdepartmental conflict and reducing the project’s overall productivity.

Managers at all levels should be concerned with the demonstrated tendency to functionalize activities over time on both an individual and organizational basis. Project organizations should establish a structure which fosters open communication and participative action in all directions. This structure should be designed such that a degree of dependency and interaction among all components of the project organization is an absolute requirement for project success throughout the acquisition process. Group decision making should be encouraged. These and similar actions should help insure that project managers not only understand the function and purpose of their own activities but also the relationship of those activities to other components and to project objectives over time.

Attention should also be focused on preventing functionalization of the project offices themselves. Just as individuals tend to functionalize tasks over time, organizations tend to routinize activities through the use of rules, regulations, and operating instructions. As discussed previously, care should be taken to avoid limiting the responsiveness and adaptability of project organizations through over-regulation of these functions. By avoiding the restrictions and constraints of bureaucratic organizations, project organizations can concentrate their full energies on achieving the desired end product at less cost and within time limitations.

Much of this supports the classical theory of project management. In one key area, at least, new insights are provided. It has long been held that a project manager should remain with a project throughout its life cycle, retaining total responsibility for its output. This study, however, provides some clues for when his removal should be considered. Specifically, when a project manager overfunctionalizes his project, he may well be reducing the potential to achieve the desired results. In that case, it may be appropriate to consider his replacement.

Finally, the existence of an analyzed and supportable model of the relationships among major organizational variables in a project environment, as presented in this article, can provide the basis for a structured research effort into those factors that may improve the chances of project success. Such a structured research program has a much greater chance of producing valid and useful results than the small, fragmented research efforts which currently characterize this field of study.