Abstract
Different elements of local police agencies’ terrorism preparedness may be associated with different organizational/environmental variables. We use 2003-2007 data (showing considerable adoption and desistance of practices) on medium-to-large-sized local agencies to examine relationships between contingency (vulnerability, organizational characteristics) and contagion (network/isomorphic influence) measures and preparedness elements, including terrorism special units, dedicated assignment of personnel, terrorism-related community outreach, computerized intelligence files, and interagency-shared radio frequencies. Modeling 2007 preparedness revealed consistencies and some differences in the associations between these measures and the different preparedness elements. The finding of no association between objective vulnerability score and any terrorism preparedness action particularly warrants further research attention. It will also be important to extend preparedness research into the recent period of economic recession.
Introduction
The terrorist attacks of 9/11 changed local police departments in important ways. Local agencies were expected to perform new roles as potential first responders to terrorist incidents and to assist federal authorities in developing intelligence on possible terrorist activity. For many agencies, there is a sense of a fundamental shift in focus from an earlier emphasis on community-oriented policing to a new charge of ensuring homeland security (Lee, 2010; W. M. Oliver, 2004, 2006; Ortiz, Hendricks, & Sugie, 2007). But this embrace of new roles has been uneven, with some departments taking more, or more elaborate, steps to develop new capabilities, and with little guidance or supervision by the federal homeland security establishment (Davis et al., 2004; Pelfrey, 2007, 2009; Randol, in press; White, 2004). Given the potential importance of these new roles, it is critical to understand why different agencies have or have not taken different terrorism preparedness actions. Scholars of criminal justice have engaged this emerging research question (Burruss, Giblin, & Schafer, 2010; Davis et al., 2004; Gerber, Cohen, Cannon, Patterson, & Stewart, 2005; Giblin, Schafer, & Burruss, 2009; Schafer, Burruss, & Giblin, 2009). Although there has been substantial progress in this research, there is still much to be learned. It is particularly important to study agency preparedness in a period that is long enough after 9/11 for new initiatives to be put in place, and to consider change from a measured baseline of terrorism preparedness. Application of multivariate models in data of national scope, and with separate investigation of distinct elements of terrorism preparedness, will also enrich the understanding of local agencies’ preparedness.
In the current study of 374 relatively large local police agencies (serving populations of 100,000 or more), we examine relationships between contingency (terrorism vulnerability, organizational properties, and activities) and contagion (potential network influence) factors suggested by organizational theory and five different elements of terrorism preparedness. We primarily use Law Enforcement Management and Administrative Statistics (LEMAS) data from 2007, with 2003 LEMAS data providing baseline measures of preparedness. Our results paint an interesting picture of factors associated with different preparedness actions, and help in understanding how observed preparedness reflects agencies’ organizational characteristics and processes of interagency contagion as well as the nature of terrorist threats that local agencies face. Given the importance of improving agencies’ ability to address terrorism, research findings on the factors that foster terrorism preparedness can help state and federal authorities most efficiently target resources and assistance to maximize preparedness throughout the criminal justice system.
Contingency Factors and Terrorism Preparedness
Although terrorism preparedness presents new challenges, police agencies have long needed to adapt to changing environments (Marks & Sun, 2007; McGarrell, Freilich, & Chermak, 2007). Indeed, McGarrell et al. (2007) suggest that in recent decades, the police have changed more than any other type of government agency. Contingency theory (Lawrence & Lorsch, 1967) is one way to make sense of these changes. As applied to law enforcement, the contingency perspective “view[s] the effectiveness of police organization as a product of a proper fit of organizational design and environmental exigencies” (Langworthy, 1986, p. 31). This suggests that changing agency practices are a response to environmental factors, such as the type and extent of crime in the jurisdiction, or organization, such as size, resources, and formal structure, that may affect successful policing (Jiao & Rhea, 2007; Zhao, Ren, & Lovrich, 2010). Contingency theory therefore has been applied in studies of police agencies’ adoption of various innovative practices and structures (e.g., Crank & Wells, 1991; Katz, 2001; Katz, Maguire, & Roncek, 2002; Langworthy, 1986; Maguire, 2003; Morabito, 2010; Willis, Mastrofski, & Weisburd, 2007). The contingency perspective has also been suggested as a general framework for understanding agencies’ terrorism preparedness (Burruss et al., 2010). We now discuss potential influences on terrorism preparedness that are consistent with the contingency view.
Terrorism Vulnerability
The contingency perspective’s emphasis on environmental conditions suggests that the nature of the terrorist threat faced by a local police agency will be one of the most important influences on the agency’s terrorism preparedness. Due to the difficulty of measuring actual terrorist risk, most previous studies on terrorism preparedness among local police agencies used perceived threat of terrorism by police survey respondents. Respondents were asked to rate the likelihood of terrorist attacks in general or of particular kinds (e.g., biological, nuclear, or conventional) in their jurisdiction in the near future (Burruss et al., 2010; Davis et al., 2004; Davis, Mariano, Pace, Cotton, & Steinberg, 2006; Gerber et al., 2005; Giblin et al., 2009; Schafer et al., 2009). These studies found a positive association between perceived risk and overall terrorism preparedness (Burruss et al., 2010; Gerber et al., 2005; Giblin et al., 2009; Schafer et al., 2009), indicating that high risk jurisdictions are better prepared. However, when preparedness was separated into detailed elements, perceived risk was not related to all areas of preparedness (Davis et al., 2004; Davis et al., 2006; Gerber et al., 2005), suggesting caution in interpreting the relationship between risk and preparedness.
Furthermore, use of perceived risk might be problematic when exploring the relationship between environmental conditions and counterterrorism preparedness, as perceived risk may be a function of preparedness (Giblin et al., 2009) rather than reflecting actual vulnerability to terrorism. That is, well-prepared agencies might perceive less risk whatever their actual vulnerability. Or it is possible that highly prepared agencies are more aware of and concerned with possible threats. These possibilities call for the use of a more objectively measured environmental vulnerability when exploring the relationship between risk and terrorism preparedness (Schafer et al., 2009). Davis et al. (2004) asked police agencies about the number of potential targets such as nuclear power plants, airports, dams, and bridges in their jurisdiction or nearby. However, they created a risk measure that combines number of targets with perceived threat, rather than examining objective vulnerability and preparedness directly (Davis et al., 2004). Borden, Schmidtlein, Emrich, Piegorsch, and Cutter (2007) also developed an objective “built-environment” vulnerability index for 132 urban centers; Piegorsch, Cutter, and Hardisty (2007) interpreted this measure as one important component of terrorism vulnerability. To date, this index has not been used to explore different agencies’ preparedness.
Horizontal Differentiations and Formalization
The general organizational and policing literatures have long recognized the importance of organizational attributes as factors in the adoption of innovations (e.g., Rogers, 1995; Weisburd & Braga, 2006). Police researchers have especially focused on “structural complexity” and “structural control” (Langworthy, 1986). Complexity refers to different possible elements of differentiation in the structure of an organization, including spatial, hierarchical, occupational, and functional differentiation (Langworthy, 1986). In Langworthy’s (1986) characterization of the complexity of police agencies, spatial differentiation refers to an agency’s geographic footprint, with indicators such as the number of separate field offices and patrol beats. The number of distinct command levels indicates the extent of the agency’s hierarchical differentiation, whereas occupational differentiation is indicated by the employment of technically specialized personnel, and functional differentiation by the presence of special units. Management features, including centralization of decision making, formalization of policies and procedures, and density of administrative relative to field personnel, indicate the nature of structural control (Langworthy, 1986; Maguire, 2003).
Among four components of organizational structural complexity, functional and occupational differentiation—representing the extent of horizontal complexity—are likely more relevant for adoption of innovations than spatial and hierarchical differentiation (Damanpour, 1996). Functional and occupational differentiations are thought to positively influence adoption of innovation (Damanpour, 1996). In this view, organizations with a higher level of functional differentiation will possess specialized expertise in more different functional areas, whereas higher levels of occupational differentiation will lead to more collaborative work between different kinds of specialists. Such organizations will be aware of a wider scope of potential innovations, and have more ability to produce new ideas or creatively apply existing ideas in new domains (Damanpour, 1991, 1996). In the present research setting, greater functional and occupational differentiation in a police agency should facilitate that agency’s adoption of new terrorism preparedness strategies (Burruss et al., 2010).
According to Damanpour (1991), formalization—one of the structural control elements—is also related to innovation adoption. However, the direction of this relationship is debated. Formalization might be positively associated with innovation adoption, if organizations that emphasize formal rules, policies and procedures are better suited to adopt new innovations due to less possibility of resistance, especially by subordinate members, to implementation of new practices (Burruss et al., 2010; Morabito, 2010). Adding terrorism preparedness measures such as counterterrorism policies and procedures, mutual agreements with other agencies, and terrorism special units may be easier when line officers are more constrained by formal rules (Burruss et al., 2010). However, some researchers argue that more formalized organizations will be less creative and flexible in adopting new innovations, in part because a greater reliance on formal procedures means that there will be fewer sources of potential input on new ideas (Damanpour, 1991; Randol, in press).
Among existing terrorism preparedness research, only Randol (in press) has directly measured horizontal differentiation and formalization and tested their relationships with preparedness. Functional differentiation, measured as the number of special units with full-time staff, was positively associated, as expected, with overall terrorism preparedness. Surprisingly, occupational differentiation, indicated by the ratio of administrative to sworn personnel, did not have a statistically significant relationship with the terrorism preparedness index. Randol (in press) measured formalization by each agency’s number of written policies and procedures, and found that formalization was negatively related to terrorism preparedness. This is consistent with the view that more formalized organizations are less flexible in adopting new innovations (Damanpour, 1991; Randol, in press).
Agency Size
The size of an organization is often used as a crude overall measure of contingency factors because of its relationships with environmental conditions, complexity, and structural control (e.g., Burruss et al., 2010). For police agencies, larger size could be advantageous and disadvantageous for adoption of counterterrorism measures. On one hand, larger organizations are generally more formalized and horizontally differentiated (Donaldson, 2001; Langworthy, 1986; Maguire, 2003), which may provide greater capability for managing and coordinating counterterrorism (Burruss et al., 2010). And even controlling for formalization and horizontal differentiation, large police agencies could be more likely to implement new innovations because their greater financial slack, number of specialists (e.g., grant specialists), research and technological capabilities, and other resources encourage adoption (Damanpour, 1996; Pelfrey, 2007). On the other hand, larger agencies’ complexity and potential inflexibility may inhibit innovation adoption. Along with possible negative effects of greater formalization (Randol, in press), if innovation adoption requires that different parts of an organization work together, such coordination may be easier in smaller organizations (Damanpour, 1996), leaving smaller agencies better able to introduce innovative practices. Structural inertia (Hannan & Freeman, 1989) is likely more difficult to overcome in larger agencies.
Note that another aspect of size may have particular relevance in the context of terrorism preparedness. In addition to whatever effects of size would be expected from the general organizational literature, when considering terrorism preparedness, agency size is likely one indicator of the number and attractiveness of potential terrorist targets. Larger jurisdictions certainly have more, and likely more prominent, potential targets, and therefore should be expected to put greater priority on preparedness (Burruss et al., 2010; Davis et al., 2004; Gerber et al., 2005; Jiao & Rhea, 2007; Lee, 2010; Pelfrey, 2007). In this context, size is an indicator of environmental conditions as well as being an organizational attribute. Previous studies have consistently found that the agency or jurisdiction size is positively associated with terrorism preparedness (Burruss et al., 2010; Davis et al., 2004; Davis et al., 2006; Gerber et al., 2005; Giblin et al., 2009; Marks & Sun, 2007; Ortiz et al., 2007; Randol, in press).
Community Policing
The decades before 9/11 were marked by growing acceptance of the principles of community and problem-oriented policing, so that by 2001, most agencies had adopted at least some features of this approach (Lee, 2010; Maguire, 2002; W. M. Oliver, 2009). The subsequent salience of terrorism preparedness raised the question of fit between principles of community policing and terrorism preparedness, and this question is still controversial among researchers. Some scholars suggest that community policing is inherently in conflict with terrorism preparedness. Community policing stresses openness, decentralization, and citizen–police partnership in addressing crime and its causes, whereas homeland security measures require covert operations and intelligence gathering that are inimical to warm community relations (W. M. Oliver, 2004, 2006). In addition, funding for homeland security may be at the expense of resources that previously would have been devoted to community policing (Chappell & Gibson, 2009; Lee, 2010; W. M. Oliver, 2006). Given this, agencies that are strongly committed to community policing may be hesitant to adopt elements of homeland security preparedness, especially intelligence gathering, that may conflict with community policing goals.
However, others suggest that community policing actually complements terrorism preparedness (e.g., Henry, 2002; Lyons, 2002; Murray, 2005; Thacher, 2005). In this view, strong community partnerships will encourage citizens to share information that may be relevant to homeland security, and community and problem-oriented policing initiatives may help in the prevention or investigation of terrorist activity (Chappell & Gibson, 2009; Henry, 2002; Innes, 2006; Murray, 2005; Pelfrey, 2005). Arab-American communities are an important potential source of information on homegrown or imported radical Islamic threats, so the stronger police–community partnerships and communication encouraged by community policing may be especially valuable (Lyons, 2002; Thacher, 2005). In addition, community policing and terrorism preparedness aim to reduce ordinary citizens’ fear (Chappell & Gibson, 2009; Innes, 2006).
Previous studies on the relationship between community policing and terrorism preparedness have been generally, but not perfectly, consistent. In Chappell and Gibson’s (2009) survey of Virginia police chiefs, a majority viewed community-oriented and homeland security policing as complementary, and chiefs from departments that were highly committed to community policing were more likely to take this view. In addition, Ortiz et al.’s (2007) interviews with police in jurisdictions with large Arab-American populations showed that police agencies that actively practice community policing were likely to engage in more terrorism-related community outreach. Randol’s (in press) multivariate analysis of data from a nationally representative sample of agencies found a statistically significant and positive relationship between community policing and terrorism preparedness. Similarly, Lee (2010) found that an index of community policing programs was positively associated with prioritization of homeland security planning. However, Lee also found that such prioritization was negatively related to the percentage of officers devoted solely to community policing. These results may indicate that although the principles of community policing and homeland security are not contradictory, in practice, the two approaches are in competition for resources, so allocation of funding or officers to one will be at the expense of the other. Note too that a study of South Carolina agencies found no relationship between community policing emphasis and terrorism preparedness (Pelfrey, 2005).
Contagion and Terrorism Preparedness
Scholars of innovation diffusion have taken special interest in “contagion” effects, in which knowledge and subsequent implementation of new practices or structures come about through observation of other organizations, or more directly through interorganizational network ties (e.g., A. L. Oliver & Ebers, 1998; Strang & Soule, 1998; Wejnert, 2002). The idea of institutional isomorphism, referring to tendencies toward similarity of organizational form and practices within organizational populations, has been influential in the general organizational literature (DiMaggio & Powell, 1983) and has likewise been applied to police agencies (Crank & Langworthy, 1992; Giblin, 2006). Considerations of isomorphism are typically motivated by institutional theory (Meyer & Rowan, 1977) and its emphasis on organizations’ need to appear legitimate to crucial external actors. However, Roberts and Roberts (2009) noted that the presence of network influences on police agency practices can be understood through a variety of theoretical perspectives, including contingency theory.
Research on terrorism preparedness has suggested that such contagion effects (potentially related to isomorphism and institutional forces as well as network influence) may help account for variability in agencies’ preparedness (Burruss et al., 2010; Pelfrey, 2007; Schafer et al., 2009). Burruss et al.’s (2010) “institutional pressures” factor was significantly related to overall preparedness, and in turn, the institutional pressures factor was significantly associated with agency self-reports of the perceived salience of other agencies’ preparedness in formulating their own practices. Pelfrey (2007) focused on the importance of formal accreditation in encouraging innovative practices, and accreditation likely leads to homogenization of practices across agencies. Schafer et al.’s (2009) study of small Illinois agencies found that location in the Chicago metropolitan area, with its greater access to larger agencies, was associated with more terrorism preparedness than location elsewhere.
Current Study
The current study examines relationships between contingency (terrorism vulnerability, organizational properties, and activities) and contagion (potential network influence) factors suggested by organizational theory and five different elements of terrorism preparedness in 374 relatively large (serving populations of 100,000 or more) local police agencies. These elements include presence of terrorism special units, assignment of dedicated terrorism personnel, terrorism-related community outreach, use of computerized intelligence files, and interagency-shared radio frequencies. Although terrorists could live and operate in smaller communities (Pelfrey, 2007), we focus on medium- and large-sized agencies because smaller agencies are likely to eschew formal planning and special units in favor of ad hoc reliance on community resources in times of crisis (Sharp, 2001). The measures of formalized terrorism preparedness that interest us are therefore of less relevance to small agencies (Falcone, Wells, & Weisheit, 2002), whereas the different nature of police–community relationships in larger communities requires that preparedness be formalized to be effective (Sharp, 2001).
The current study extends existing research on terrorism preparedness in four ways. Although certain studies have shared some of these features, none have had all of them. These include examination of a period that is long enough after 9/11 for influences on the new wave of terrorism preparedness to be apparent, multivariate analysis in data of national scope, analysis of separate elements of preparedness rather than only an overall measure, and use of an objective measure of vulnerability to terrorism as a predictor of preparedness. Taking these in turn, existing studies necessarily used preparedness data collected no more than a few years after 9/11 (Pelfrey, 2007; Randol, in press), but that may not have been enough time for policy changes to be implemented (Randol, in press). Our study period of 2003-2007 gives baseline measures from shortly after 9/11, but also allows sufficient time to evaluate factors related to subsequent preparedness steps. The use of a baseline also permits consideration of actual, not perceived (Jiao & Rhea, 2007; Pelfrey, 2009), change. Second, most existing work has been descriptive or bivariate in character (e.g., Davis et al., 2004; Davis et al., 2006; Jiao & Rhea, 2007; Ortiz et al., 2007; Pelfrey, 2009), and when multivariate analysis has been done, it has been with data in limited geographical areas (e.g., Burruss et al., 2010; Giblin et al., 2009; Schafer et al., 2009). Third, most previous multivariate work has created an overall terrorism preparedness scale from measures of separate aspects of preparedness, and then used that overall measure as the dependent variable (Burruss et al., 2010; Giblin et al., 2009; Lee, 2010; Randol, in press; Schafer et al., 2009). Separating different elements of preparedness actions allows a richer and clearer picture of how contingency and contagion factors relate to preparedness. For example, community-oriented policing may be compatible with terrorism-related community outreach but not with development of computerized intelligence files. Finally, perceived terrorist risk has been used in other studies to predict preparedness (e.g., Burruss et al., 2010; Davis et al., 2004; Gerber et al., 2005; Giblin et al., 2009; Schafer et al., 2009). However, perceived risk may in fact be a function of preparedness, with well-prepared agencies feeling less at risk, suggesting use of objectively measured environmental vulnerability instead (Schafer et al., 2009).
Data and Method
Dependent Variables
As discussed previously, the current study explores different elements of terrorism preparedness. Thus, the dependent variables reported the state of agency preparedness in five different areas in 2007: terrorism special unit, dedicated assignment of personnel, terrorism-related community outreach, computerized intelligence files, and interagency-shared radio frequencies. Terrorism special unit, computerized intelligence files related to terrorism activities, and interagency-shared radio frequencies were indicated by dichotomous (presence/absence) measures. Terrorism-related community outreach was measured as the number of four types of outreach (partnership with culturally diverse communities, public antifear campaigns, dissemination of information to increase citizen preparedness, and community meetings on homeland security/ preparedness) conducted by an agency. The variable representing dedicated assignment of personnel was similar, but for two types of assignments (presence of personnel assigned full-time to a multiagency terrorism task force and presence of personnel assigned full-time to terrorism intelligence gathering). Data for the various dependent variables were obtained from 2007 LEMAS.
Independent Variables
Our independent variables included vulnerability score, functional and occupational differentiation, formalization, agency size, resources, community policing, and contagion factor. Due to the problems associated with perceived risk discussed in the earlier section, we measured a city’s vulnerability more objectively by using Borden et al.’s (2007) built-environment vulnerability index. Borden et al. created this index from 40 variables measured in 132 “urban centers.” 1 The 40 variables included indicators of the age and density of residential buildings, measures of business and industrial activity, indicators of numerous aspects of infrastructure (e.g., facilities for water, nuclear and other energy, medical care, and air and ground transportation), and the presence of “monuments and icons,” including prominent buildings. Borden et al. combined the variables into a single index via principal components analysis; their work provides full details. 2
Functional differentiation is usually measured as the number of specialized units in an organization (e.g., Damanpour, 1991; Randol, in press). Following this, we measured functional differentiation through LEMAS information on the special unit types for which an agency had dedicated full-time personnel. LEMAS provided data on 18 types of special units, 3 so this measure could potentially range from 0 to 18, with larger values indicating greater functional differentiation. Damanpour (1991) envisioned occupational differentiation in terms of an organization’s range of different specialists, so that its natural measure focuses on the organization’s different occupations or job titles. We measured occupational differentiation by calculating an index of job heterogeneity that is equivalent to familiar measures of racial diversity (Blau, Blum, & Schwartz, 1982). LEMAS provided data on the proportions of agency personnel in six different job types. 4 Writing p ij for these proportions in agency i, agency i’s job heterogeneity index was calculated as follows:
Values of this index may be interpreted as the probability that two randomly selected employees have different job types. Larger values indicate greater heterogeneity. The maximum possible value of this index, 0.833, would be attained if each p ij = 1/6, and the theoretical minimum is 0, attained if p ij = 1 for one job type and 0 for all of the others.
Formalization was operationalized as each agency’s total number of written policy directives concerning on- and off-duty officer conduct, off-duty employment of officers, maximum work hours, interaction with special populations or situations, citizen complaints, and employee counseling assistance. We considered 14 LEMAS questions on written policies and procedures, 5 so this variable could potentially range from 0 to 14, with larger values indicating a higher level of formalization. Agency size was indicated by the number of full-time sworn officers plus half of the number of part-time sworn officers. We also included a measure of financial resources, as agencies with greater resources may be more able to initiate new programs (Lee, 2010; Marks & Sun, 2007). Although resources may be related to size (Damanpour, 1991), our measurement of resources as operational budget (in thousands of dollars) per 10,000 serving population correlated only weakly (r = .18) with our agency size measure. This suggests that resources represent a different characteristic than agency size, so it is valuable to include both in multivariate models of preparedness actions. Size and resources were logged for all models. The community policing measure summed 15 yes/no indicators of community-oriented policing, including community policing special units, community policing personnel, training, planning, policy, technology, “Scanning, Analysis, Response, Assessment (SARA)–type problem solving,” “fixed geographical assignments,” and citizen involvement in LEMAS. 6 The measure thus could vary from 0 to 15, with larger values indicating greater adherence to community policing principles.
For each agency, we calculated a measure of potential exposure to contagion. No available data report on network ties among these agencies; absent direct network data, we considered all other agencies as potential sources of influence on a focal agency, weighted by their size and their (inverse) distance from the focal agency. Larger organizations are likely to be seen as exemplars for imitation (Greve, 2005; Haunschild & Miner, 1997; Semadini & Anderson, 2010; Williamson & Cable, 2003), and police agencies’ information-seeking contacts tend to be directed toward larger alters (Chamard, 2003; Roberts & Roberts, 2007). For a given terrorism preparedness measure t, we calculated c i , agency i’s potential contagion exposure, based on the preparedness t j of all of the 373 other agencies. The calculation used d ij , the distance (miles) between agencies i and j, and the size s j of agency j (measured as the number of sworn officers) as follows:
The potential contagion measure is therefore a weighted average of the preparedness of all other agencies in the data, with greater weight for larger and closer agencies. We calculated such a contagion measure for each of our preparedness measures, with t j taking values 1 or 0 for yes/no preparedness measures (terrorism special unit, computerized intelligence files, and interagency-shared radio frequencies). For terrorism-related community outreach and dedicated assignment of personnel, t j was the number of the possible actions taken. In all cases, a larger value of c i corresponds to greater potential contagion exposure for that agency. 7 The appendix gives descriptive statistics for the independent variables.
Models
We examined logistic regression models for each dependent variable. Three of the preparedness measures (terrorism special unit, computerized intelligence files, and interagency-shared radio frequencies) were of the familiar binary type that is often analyzed via logistic regression. The other two (terrorism-related community outreach and dedicated assignment of personnel) were somewhat different, as they reported the number of attributes present out of the total possible (four for terrorism-related community outreach and two for dedicated assignment of personnel). Such measures can also be analyzed via logistic regression, with the assumption that the same probability of being “present” applies to each of the various attributes that contribute to the measure. This is conceptually similar to treating these two measures as continuous or count variables, but somewhat preferable in that the measures’ upper limits (the number of attributes being examined) are explicitly recognized. Along with the substantive independent variables discussed above, each model also included an independent variable indicating the lagged (measured in 2003) value of that model’s dependent variable (measured in 2007). This specification is consistent with various approaches to longitudinal data and permits a “dynamic” interpretation of model parameters as indicating associations between variables and preparedness in 2007 while accounting for the baseline preparedness in 2003.
Because an urban center could, and often did, include more than one of the agencies in the data set, it is reasonable to view agencies as being clustered by urban center. (Note that as discussed above, the same vulnerability score was applied to each agency in an urban center.) Because ignoring this clustering would threaten the validity of our significance tests, we used SAS PROC GENMOD to fit models under the generalized estimating equations (GEE) approach to clustered or correlated data (Liang & Zeger, 1986; Zeger, Liang, & Albert, 1988). Results reported in the following section assume an exchangeable correlation structure within urban centers, 8 and use “empirical” standard errors for all statistical inference.
Results
Before discussing the multivariate results, it is useful to descriptively explore change and stability in the terrorism preparedness measures between 2003 and 2007. Table 1 summarizes patterns in the terrorism preparedness measures for the 374 agencies in the analyses. The fifth and sixth columns of figures describe the extent and nature of change. Column 5 shows the percentage of agencies whose preparedness changed—in either direction—between 2003 and 2007. Column 6 gives the ratio of agencies changing from absent to present on a preparedness measure to those changing from present to absent. According to Table 1, there was substantial change in agency terrorism preparedness measures across the two administrations of LEMAS. For each measure, at least 23.8% of agencies changed between 2003 and 2007, with a maximum of 43.0% (in dissemination of information to increase citizen preparedness).
Agency Preparedness 2003-2007 (N = 374)
For most terrorism preparedness measures, the ratio of the proportion of agencies changing from absent to present to the proportion doing the reverse exceeded 1, indicating more adoption than desistance of the preparedness action. The ratio was particularly high for terrorism special unit (2.6), intelligence gathering personnel (3.0), partnership with culturally diverse communities (2.1), use of computerized intelligence files (2.0), and interagency-shared radio frequencies (2.4). These figures reflect continuing adoption of those particular homeland security measures more than 5 years after 9/11. Public antifear campaigns was the only terrorism preparedness measure with a ratio less than 1, as more agencies changed from having public antifear campaigns to not having them than vice versa. It may be that police in general felt less need to address public fear as more time passed since 9/11, as public opinion polls suggested less fear of terrorism in 2007 than in 2003 (Saad, 2011). Note that the extent and nature of change varied considerably across the different preparedness measures. This reinforces the importance of considering these measures separately, rather than combining them into one overall measure.
Next, we turn to the multivariate results. Table 2 summarizes findings on the relationship between contingency (terrorism vulnerability, organizational properties, and activities) and contagion (potential network influence) factors and five different measures of terrorism preparedness (terrorism special unit, allocation of terrorism-related personnel, community outreach, computerized intelligence files, and interagency-shared radio frequencies). 9
Logistic Regression Results for 2007 Preparedness Measures (N = 374)
Note: Standard errors are reported in parentheses.
p < .10. *p < .05. **p < .01 (two-tailed).
There were certain elements of consistency across results for the different preparedness measures. Four independent variables—those measuring vulnerability, formalization, community policing, and (log) resources—were not significantly associated with any of the preparedness measures. Moreover, in each case, the baseline (2003) preparedness measure was significantly associated with 2007 preparedness (p = .0504 for use of computerized intelligence files). As expected, given the descriptive information in Table 1, these associations were positive, so that 2007 preparedness was estimated to be more likely for agencies which were similarly prepared in 2003 than for those which were not.
Some other results were substantially consistent across the different preparedness measures. Agency size, measured as the logged number of sworn officers, was significantly and positively associated with four preparedness measures (terrorism special unit, assignment of terrorism personnel, community outreach, and use of computerized intelligence files). Its association with the presence of interagency-shared radio frequencies was also positive, but not statistically significant. (In fact, aside from the 2003 baseline, none of the independent variables were significantly associated with this preparedness measure.) The relationship between size and the presence of a terrorism special unit was noteworthy, as a 10% increase in the number of sworn officers was, net of other variables, estimated to be associated with 15% greater odds of having a terrorism special unit. This seems consistent with the general expectation that larger agencies show greater functional differentiation, but as for the functional differentiation measure (the agency’s number of special units) itself, there was a significant association with only terrorism-related community outreach; each additional special unit was estimated to be associated with 4% greater odds of an outreach type being present.
There was less consistency in results for occupational differentiation and contagion. Occupational differentiation, measured by job heterogeneity, was significantly associated with assignment of terrorism personnel and community outreach (marginally so for assignment of personnel), but none of the other measures. Furthermore, its association with assignment of personnel was positive and its association with community outreach was negative. The contagion measure was significantly associated with only the presence of a terrorism special unit (marginally so) and assignment of terrorism personnel. An increase of .10 in the contagion measure was estimated to be associated with 15% greater odds of having a terrorism special unit. However, estimates of similar magnitude for contagion and use of computerized intelligence files and shared radio frequencies were not statistically significant. 10
Note that we have discussed statistical significance in a model-by-model (preparedness measure-by-preparedness measure) perspective. We could also view the collection of results as addressing, in total, the more general question of whether a particular independent variable is associated with preparedness in any form. In that case, it would be natural to use a Bonferroni-type adjustment to significance levels; with five measures, the conventional .05 level would be adjusted to .01. Of the independent variables discussed above as showing at least some significant associations with preparedness measures, all but the functional differentiation measure (number of special units) meet this adjusted standard in at least one model.
Conclusion and Discussion
The current study used organizational theory to examine relationships between contingency factors (terrorism vulnerability, organizational properties, and activities) and contagion (potential network influence) and various different elements of terrorism preparedness in a sample of relatively large local police agencies. It focused on the post-9/11 (2003-2007) period, examining separate elements of preparedness, rather than a single overall preparedness measure, in multivariate analysis of data from throughout the United States. It also used an objective measure of infrastructure vulnerability instead of perceived terrorism risk. Perhaps the most noteworthy result is the finding of no association between objective vulnerability score and any terrorism preparedness actions. An obvious concern is that this may indicate an inefficient allocation of national preparedness resources. Furthermore, our findings differed from those of most previous studies that showed a positive association between perceived risk and overall preparedness (Burruss et al., 2010; Davis et al., 2004; Gerber et al., 2005; Giblin et al., 2009; Schafer et al., 2009). This suggests a mismatch between perceived risk and objective vulnerability, so that perceived risk may in fact be a by-product of preparedness (Giblin et al., 2009). The previous findings of a positive relationship between perceived risk and preparedness may suggest that agencies with higher levels of preparedness are more concerned with and cognizant of possible threats, regardless of actual vulnerability.
The nonsignificant association between formalization and preparedness may result from formalization in some ways fostering and discouraging adoption of new innovations (Burruss et al., 2010; Randol, in press). On one hand, more formalized agencies are better able to adopt new innovations because they can more easily overcome any resistance among line officers (Burruss et al., 2010; Morabito, 2010). On the other hand, agencies that emphasize formal rules, policies, and procedures will be less open to new ideas (Damanpour, 1991; Randol, in press).
The current research findings suggest that community policing does not encourage counterterrorism preparedness actions, but neither does it hinder them. Recall that Lee (2010) found adherence to community policing principles was positively associated with prioritization of homeland security, but that this prioritization decreased as more funds were allocated to community policing (as indicated by the percentage of officers solely engaged in community policing tasks). This suggests that although community policing is not in principle contradictory to terrorism preparedness, in practice, they compete for resources. For the current research, LEMAS does not provide the proportion of an agency’s budget that is allocated to community policing, and LEMAS data on the proportion of officers engaging in community policing do not indicate whether community policing is these officers’ sole task. Therefore, we could not measure community policing funding separately from other aspects of community policing, and such separation is necessary to untangle the potentially complicated relationship between community policing and homeland security policing. Similar considerations may apply to our financial resources variable. Because LEMAS currently does not provide details of operational budget allocation, or information on receipt of homeland security grants, we used overall operational budget per serving population as our financial resources variable. Although we found no relationship between overall resources and preparedness actions, Davis et al. (2004) and Davis et al. (2006) found a positive relationship between resources allocated specifically for terrorism and preparedness immediately after 9/11.
Results for agency size supported previous terrorism preparedness studies (Burruss et al., 2010; Davis et al., 2004; Gerber et al., 2005; Giblin et al., 2009; Marks & Sun, 2007; Ortiz et al., 2007; Randol, in press). Even with the restricted (to medium and large agencies) distribution of agency size in our sample, size was the most consistent predictor of preparedness. However, only certain preparedness actions were significantly associated with horizontal differentiation and contagion. Functional differentiation was only related to terrorism-related community outreach (and even in that case, its significance did not reach the adjusted Bonferroni standard). Occupational differentiation was related to terrorism-related personnel and community outreach, but the directions of those relationships were different (positive for personnel and negative for community outreach), and our contagion measure was significantly associated with only terrorism special unit and assignment of terrorism-related personnel. This may indicate that other agencies’ preparedness is more visible and influential for some elements than for others. Direct data on network ties—unavailable here—also may be useful.
Very few independent variables were statistically significant in models for use of computerized intelligence files and shared radio frequencies. If interagency cooperation is critical to effective prevention and response to terrorist attacks (McGarrell et al., 2007; Sloan, 2002), these elements of preparedness are especially important. Sharing intelligence with other local, state, and federal agencies is a key preventive step against terrorist attacks (Chappell & Gibson, 2009; Henry, 2002; Sloan, 2002), and the use of computerized terrorism-related intelligence files makes such information sharing easier. “Interoperability” among public safety agencies in the same or nearby jurisdictions would allow a more coordinated response to an actual terrorism event (Chappell & Gibson, 2009; Gerber et al., 2005; Henry, 2002). Thus, it is important for future studies to identify the variables associated with these elements of preparedness. One possibility is that these elements depend heavily on an agency’s overall technical expertise. Interviews or focus groups with personnel engaged in these elements of preparedness at different agencies may help identify promising factors for further investigation of the correlates of such interagency cooperation.
Although the current study explored several different elements of counterterrorism preparedness, there are others it could not examine, including training, equipment acquisition, and entering into mutual aid agreements with nonpolice agencies. 2003 LEMAS included questions on counterterrorism equipment acquisition and mutual aid agreements with city, county, transit, public works, and other agencies, but those questions did not appear in 2007, preventing the sort of analysis shown here. The RAND surveys (Davis et al., 2004; Davis et al., 2006) are a potential source of longitudinal preparedness data and would be valuable to exploit, particularly if further waves are conducted.
The importance of further longitudinal analysis is accentuated by the onset of economic recession shortly after our study period. Our results showed no association with resources (funding) net of agency size and other organizational properties. However, since 2007, many local governments have faced budget crises, often of considerable severity, as tax revenues have not kept pace with the need for social and other government services. This could change the relationship between resources and terrorism preparedness, and perhaps lead to a different relationship between objective vulnerability and preparedness as well. If the economic environment more often forces local police departments to choose between terrorism preparedness measures and everyday policing, it may be that preparedness, or at least the most costly elements of preparedness, will be increasingly limited to those agencies that confront the greatest vulnerability. It will therefore be important to extend the current study’s focus on the immediate post-9/11 era with new research on preparedness in an era of economic recession.
Footnotes
Appendix
Descriptive Statistics for Independent Variables (N = 374)
| M | SD | Minimum | Maximum | Definition | |
|---|---|---|---|---|---|
| Vulnerability score | 2.10 | 3.40 | −3.61 | 8.58 | Built-environment vulnerability index developed by Borden, Schmidtlein, Emrich, Piegorsch, and Cutter (2007) |
| Functional differentiation | 8.05 | 4.18 | 0.00 | 18.00 | Number of specialized units |
| Occupational differentiation | 0.54 | 0.09 | 0.17 | 0.67 | Job heterogeneity index |
| Formalization | 12.61 | 1.39 | 6.00 | 14.00 | Number of written policy directives |
| Agency size | 610.17 | 2,125.72 | 97.00 | 35,973.00 | Number of full-time sworn officers plus one half of the number of part-time sworn officers |
| Financial resources (thousands) | 2,181.73 | 886.43 | 143.65 | 6,760.94 | Operational budget per 10,000 serving population |
| Community policing | 9.41 | 2.92 | 0.00 | 15.00 | Number of community policing practices engaged in by agency |
| Contagion factor | Weighted average of other agencies’ specific preparedness measure, weighted by size and (inverse) distance from focal agency | ||||
| For terrorism special unit | 0.33 | 0.09 | 0.13 | 0.64 | |
| For assignment of terrorism-related personnel | 0.86 | 0.19 | 0.43 | 1.51 | |
| For terrorism-related outreach | 1.48 | 0.23 | 0.85 | 2.60 | |
| For computerized intelligence file | 0.33 | 0.10 | 0.13 | 0.66 | |
| For interagency-shared radio frequency | 0.71 | 0.10 | 0.32 | 0.93 |
Acknowledgements
We thank Susan L. Cutter and Walter W. Piegorsch for identifying which counties were included in each urban center in the calculation of the built-environment vulnerability index.
Declaration of Conflicting Interests
The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.
Funding
The author(s) received no financial support for the research, authorship, and/or publication of this article.
