Inside the broader question of how many police officers a jurisdiction should employ, the number of officers assigned to follow-up investigations is a policy decision. Analyzing crime rates, case flow, and investigative practices in similar-sized cities gives little guidance to a police administrator when deciding how to allocate scarce human resources. Similarly, within the group of investigators, the relative deployment to subspecialties also involves policy calls. While empirical analysis can assist in making such decisions, no universal standard exists for either the percentage of officers who should be assigned as investigators or the relative deployment within crimes against property, capers, vice, et cetera.
Unlike patrol allocation formulas, which have received at least some attention over the last two decades, little empirical research has been conducted on investigative staffing formulas. The research that has been done has been more theoretical than applied, i.e., abstract formulas have been developed without the existence of reliable data to utilize the formulas. Indeed, a major shortcoming in the development of investigative staffing models is the near total lack of reliable data regarding the actual tasks performed by investigators and, importantly, the time ranges necessary to perform such tasks. Also missing is information about the likelihood of case clearance relative to time spent and techniques utilized. For example, no researcher has captured how long the average interview of a forgery victim takes, or how much time, on average, is spent by an investigator at the scene of a homicide. Indeed, the few studies that have been conducted suggest that investigators actually spend most of their days either in the office doing paperwork or on the telephone, or traveling by automobile from one location to another. The time spent on this latter factor naturally has the potential for wide variability, depending on the size and nature of the jurisdiction as well as whether investigators are centralized or de-centralized. (Centralized investigators in a major city would presumably have to drive further per trip than decentralized investigators.) Similarly scant data exists on the relationship of investigative time spent and subsequent identification of criminal offenders.
What is known is that the single greatest contributor to a successful criminal investigation is the quality of information gathered by the responding patrol officer. Indeed a review of case clearance rates by investigators nationwide shows a very low number, less than ten percent, of serious offenses are cleared by follow-up investigative work. Unfortunately, the low clearance rates in larceny/theft type offenses, which make up the bulk of serious crime, are aggregated in this number and obscure whatever higher rate of success might have been attained by follow-up investigators in murder, rape, and robbery cases.
This is not to suggest that investigators do not serve an important police function. Indeed, arresting offenders is only one goal of investigators. For instance, the societal outrage at murders mandates that each receive follow-up investigative attention, even though homicides constitute a very small percentage of overall crime. Indeed, public expectations are that homicides will be investigated until an offender is identified and brought to justice. Unsolved cases often are carried in an open status and continue to utilize some level of investigative resources months after their occurrence. Similar public policy concerns require extensive resources to be devoted to sexual assault investigations and robbery cases. Only in property crimes does a police department have some degree of latitude in deciding which cases to pursue further and the level of investigative effort to be expended.
There are three fundamental approaches in examining investigative staffing:
Benchmark comparison approach – This strategy examines investigative staffing levels nationwide as well as within a proximate area to a jurisdiction. Cities of comparable population to that of a given jurisdiction are always used as benchmarks. It should be noted that this strategy, while containing a fair level of face validity, is subject to criticism and results must be viewed cautiously. The underlying problem is understanding why other cities have decided on their particular levels of staffing. Perhaps they are understaffed, or even overstaffed, with investigators. Accordingly, a comparison to these jurisdictions may simply establish that a given jurisdiction is no worse off than other police departments, without actually determining whether any of the agencies is adequately staffed. This “keeping up with the neighbors” approach says little about whether staffing levels are adequate within the individual agency. Additionally, two key variables can greatly affect the conclusions drawn from such comparisons: the level of responsibility of the uniformed patrol officer in the investigative process, and case screening/assignment practices. An agency that empowers patrol officers to conduct complete preliminary investigations will likely need fewer follow-up investigators while a department where patrol officers simply hold the scene and notify detectives upon discovery of a felony will need proportionately more investigators. Likewise, agencies that assign every case for investigative follow-up will either need more investigators or allot less time to each case than a department where cases are screened prior to assignment. Such screening for “solvability” or for other purposes will greatly influence caseload levels and the number of investigators needed. These caveats should be kept in mind when reviewing comparative data.
Longitudinal time comparison approach – A longitudinal time study of a given jurisdiction’s investigative staffing levels is desirable. Essentially this strategy involves comparing current staffing levels with the staffing levels from some years earlier. Factored into this equation would be the presumed increase in criminal events over the intervening time frame and a determination as to how well staffing levels have kept up with changing workload demands. While this view does provide some guidance in establishing contemporary staffing needs, any judgments made should be made cautiously because of the underlying assumption that the earlier staffing levels were appropriate at the time. Here too the base staffing levels against which current levels are compared may have been deficient or overly generous at the time. Further, internal reorganization within a particular investigative unit may mean that it gains additional responsibilities, thus necessitating additional resources. This dynamic alteration of work makes year to year staffing comparisons problematic. For example, if the Burglary and Theft Division creates an internet crime unit, a common practice in recent years, staffing alterations would be expected, irrespective of the change in the overall number of burglaries and thefts.
Workload projection approach – Investigative units which maintain not only investigator performance data (e.g., number of cases assigned, arrests made, case clearances) but also the amount of regular work hours and overtime hours logged are taking the best approach. This data at least provides general observations about the average time necessary to process a case. Accordingly, in deciding upon future investigative staffing levels, one may assume that each new investigator added will take, on average, about the same amount of time to process a case as the role incumbents. Thus, a reasonable estimate can be made regarding the effect that each newly assigned investigator will likely have on case clearances, arrests, and other investigative outcomes.
A thorough review of investigative staffing needs might employ all three approaches, noting the limitations of any one. The focus of Justex’s Allocation Model for Investigations, however, is upon workload projection.
Accounting for Administrative Time
As in the AMP program, AMI provides for structured, non-discretionary time that is part of any work schedule, termed fixed practice. For detectives we have provided three categories of fixed time:
Administrative:
Meal breaks, routine breaks, routine paperwork
Court:
Detectives spend a significant amount of time in court, testifying occasionally but mostly waiting to testify and participate in disposition negotiations. AMI treats court time as distinct from investigative time, and counts it as hours per week rather than on a case by case basis, i.e., rather than factoring it in to case investigative time. The assumption is that a more accurate figure will ensue from estimates of overall and on average how much time our investigators spend in court.
Leave:
Like the AMP model, staffing of investigative units must consider vacation, sick, and other forms of scheduled leave time.
The resultant database is used to establish an estimated number of detectives needed to staff at a given service level, with underlying assumptions. As noted earlier, various departmental policy decisions can cause the estimated number of detectives to increase or decrease.
Approaches to Data Entry
Developing an allocation model for investigations requires that the time spent on each of the 78 case types be averaged. There are two ways an agency can determine the average amount of time per case, one detailed and, relatively speaking, tedious, the other involving “reasoned estimating,” but far easier. The first method entails having detectives log the actual amount of time they spend on a sample of cases. In determining the presumptive numbers in AMI we sampled January and July—the extreme of seasons. One could obviously sample a different time frame for varying durations. At one extreme, an agency could require detectives to log the time on every investigation for a full year, but that is not very practical. One could develop a more accurate log if the log was done prospectively, that is detectives for an upcoming month log the actual time they spend on a case. Alternatively, time could be estimated retrospectively, providing detectives with a list of cases they handled in January and asking them to estimate the time they spent on each case with the total having to add to the number of hours they worked in the particular time frame. Forms are provided with the allocation packet to gather such data, and a spreadsheet is also provided for data logging and calculation.
Alternatively, an agency can review the presumptive numbers provided in the sample agency, and estimate any variation from those presumptive numbers. The primary worksheet for AMI contains two columns, Current Status and Projections. The Current Status column is provided to log the data in terms of existing staffing levels. The data must be modified such that the estimated number of detectives required equals the actual number of detectives currently working. This is the way AMI is validated for a particular agency’s practices. If numbers are entered for each of the 78 cells, and the estimated number of detectives equals 100 whenever there are only 50 detectives actually working, there is considerable exaggeration in the amount of time spent. Adjustments must accordingly be made until the estimated number of detectives is equivalent to the actual number of detectives.
Projecting Needed Investigative Resources
The value of AMI is the ability to project investigative resources with specific enumeration of the value received, or lost, from changes in resources (the number of detectives). AMI provides several routes to estimating investigative resources with projected levels of effort. Estimates can be done in any or all of three sections of the model: performance objectives, crime trend statistics, and investigations activity statistics. Each will be discussed in turn.
Performance Objectives provide a means for “macro” estimates of additional investigators needed to increase any combination of five levels of effort: crimes against persons, crimes against property, crimes against public order, non-criminal cases, or proactive anti-crime activities. If for example an additional 10 percent of effort in aggregate is desired to investigate crimes against persons, the 10 percent figure is entered and AMI calculates how many additional detectives will be required to generate that level of additional effort. AMI automatically takes into account the specified time spent on indirect activity. In Performance Objectives the user may specify either an increase in hours per month of investigative effort for a given category or a percent of current amount of effort. AMI is designed to base calculations off of either.
Crime Trend Statistics estimate changes in investigative resources. Like the previous section, the projected changes can be made either in terms of a raw number or percent of current amount. However, while the performance objectives projections are premised upon specified raw numbers in hours per month, the crime trend statistics section uses number of offenses per month. If a crime trend in a jurisdiction indicates that crimes against persons is increasing at a rate of 3% per year then the increase can be expressed either in terms of the raw number that the increase represents or, on the second line of each category, the percent of current amount. Again, the user should not enter numbers in both rows but rather specify either an increase in the number of cases per month or a percent increase in the current amount of cases.
Investigations Activity Statistics are the most sophisticated component of the AMI simulation model is the section titled Investigations Activity Statistics. This allows a police manager to closely analyze investigative effort by offense type with four levels of suspect status. The specificity will allow the projection of resources recovered by, for example, early case closer by patrol officers of larceny/theft cases. Minimum agency specific data required is the annual number of cases in each of the 27 categories. Based on previous survey work by Justex Systems, presumptive percents of each case category involving suspect unknown, possible suspect ID, known suspect at large, and suspect in custody is provided. These numbers should be carefully examined by you to confirm that they indeed represent the experience of your agency. For example, the example data for murder/criminal homicide stipulates that the number of such cases as a percent of the total number of cases is 0.25% (1/4 of 1%), and of this number 5% involved the suspect unknown, 10% possible suspect identification, 50% known suspect at large, and 35% suspect in custody. These numbers may vary depending on jurisdiction characteristics. They are changeable by the user agency.
There are three sets of columns in this section. The first, a single column, specifies the total cases. A second set of two columns specifies the number of cases as a percent of the total. The third set of columns represents the hours spent by case category. Again an agency should review the presumptive numbers to ascertain whether they reflect the reality of that jurisdiction’s environment. The presumptive numbers, for example, indicate that while murder / criminal homicide, constituted only .25% of the total number of cases, the investigative effort constituted 2.5% of the total hours available. Similarly, while the presumptive numbers indicate that robbery constitutes 2.0% of the total number of cases, it represents 5% of the investigative hours dedicated annually. Some categories go in the opposite direction. The presumptive numbers indicate that larceny/theft represents 26% of the total number of cases, but only 20% of total investigative effort.
It should be emphasized that any or all of the presumptive numbers can be modified by the user agency. They are provided as a starting point. For most agencies they will be reasonably accurate, given the pretesting of the model. But variation will exist depending upon agency practices and priorities. For example, if an agency routinely provides that simple larceny/theft cases with no suspect identification are closed by patrol officers with no follow-up investigative effort, then the percent of time dedicated to larceny/theft by investigators will likely be lower than the 20% stipulated in the model. If, on the other hand, an agency routinely assigns all larceny/theft to an investigator, and expects a thorough check of pawn shop records, and expects multiple victim callbacks, and expects an effort to link a given theft to theft patterns, then a higher proportion of investigative effort may in fact exist. As noted above, one of the attributes of AMI is that it provides a way to reasonably estimate the investigative staff savings, or costs, of a policy change in degree of follow-up effort stipulated. Indeed one of the uses of AMI is to provide a means to systematically review investigative follow-up policy and its impact upon necessary staffing levels.
Non-Criminal Investigative Effort
There are ten categories of investigative effort for which suspect status information is not relevant. These are handled at the end of the investigations activity statistics section by simply logging the number of such cases and the estimated percent of investigative time consumed. If the investigations division never handles a particular non-criminal activity type, then a zero should be entered in that category. Additionally, this section is not designed to calculate resources for proactive investigations. This is relevant to the last four categories listed—code enforcement, drug/narcotics investigations, prostitution/vice, and gambling. In this section the only resources that should be entered are investigations in a reactive mode by regular detectives. The number of detectives currently assigned to proactive investigative activities is entered in the first section of the model, Current Statistics. Such assignments are ultimately an agency policy determination and not subject to simulation modeling. The macro staffing for such functions is included in AMI such that the model takes account of all investigative staffing. An agency could omit those numbers, simply recognizing that the model then provides only the number of detectives for reactive investigations.
Calculated Statistics
Drawing from the statistics previously provided, AMI provides six categories of summary data.
Investigation Statistics Based on Established Norms calculatesthe average annual caseload per detective is provided in terms of total cases by suspect category, hours spent by category and average hours per case by category.
Percent Calculations for Agency Norms draws upon data provided in the standards and agency norms section expressed as a percent of all investigative effort.
Average Percent Time Spent Traveling uses the average number of miles traveled per shift per investigator, with an agency stipulated average speed, to calculate percent of time spent.
Desired Increases for Case Investigations provides a summary of the hours increased by offense type given projections in previous sections.
Projected Annual Increases/Decreases in Assigned Cases provides a summary of projected increases in the number of cases and the projected hourly investigative effort requirements that are derived.
Resulting Calculation provides a proposed annual increase in investigative time, the resulting total annual hours needed for all CID activities, current plus projected, the total annual paid hours per investigator, and the total investigators recommended given the desired increases or decreases in investigative effort.
Performance Objectives provide a means for “macro” estimates of additional investigators needed to increase any combination of five levels of effort: crimes against persons, crimes against property, crimes against public order, non-criminal cases, or proactive anti-crime activities. If for example an additional 10 percent of effort in aggregate is desired to investigate crimes against persons, the 10 percent figure is entered and AMI calculates how many additional detectives will be required to generate that level of additional effort. AMI automatically takes into account the specified time spent on indirect activity. In Performance Objectives the user may specify either an increase in hours per month of investigative effort for a given category or a percent of current amount of effort. AMI is designed to base calculations off of either. 
There are ten categories of investigative effort for which suspect status information is not relevant. These are handled at the end of the investigations activity statistics section by simply logging the number of such cases and the estimated percent of investigative time consumed. If the investigations division never handles a particular non-criminal activity type, then a zero should be entered in that category. Additionally, this section is not designed to calculate resources for proactive investigations. This is relevant to the last four categories listed—code enforcement, drug/narcotics investigations, prostitution/vice, and gambling. In this section the only resources that should be entered are investigations in a reactive mode by regular detectives. The number of detectives currently assigned to proactive investigative activities is entered in the first section of the model, Current Statistics. Such assignments are ultimately an agency policy determination and not subject to simulation modeling. The macro staffing for such functions is included in AMI such that the model takes account of all investigative staffing. An agency could omit those numbers, simply recognizing that the model then provides only the number of detectives for reactive investigations.