How long does it take? An exploration of the time between case assignment and a detective’s first investigative activity

Abstract

Citizens and victims perceive the criminal justice system to be protracted. Blame for lengthy delays has disproportionately fallen on detective work. To that end, these analyses explore how long it took 184 detectives to work index crimes. Data suggest that more serious crimes, cases with a known suspect, and offenses that did not result in an arrest are most likely to be investigated and worked more promptly by detectives. Detectives, however, continue to exercise a great deal of discretion in if and when cases are worked once they are assigned. These findings are discussed as they relate to police performance indicators, the allocation of departmental resources, issues of crime control, and citizen satisfaction with the police.

Keywords:

• Case processing
• delayed justice
• police
• detective
• discretion

Notes

1. For our purposes, the term detective refers to all sworn officers that specialize in retrospective criminal investigations. In some departments, the terms inspector or investigator will be synonymous. Detective, however, does not include administrative personnel.

2. Retrospective criminal investigations are also known as secondary or follow-up investigations.

3. Index crimes include aggravated assaults, arsons, burglaries, forcible rapes, homicides, larcenies, motor vehicle thefts, and robberies.

4. Suspect status may change during the course of a criminal investigation. Throughout this text, suspect status is based on the time the case was assigned unless otherwise noted.

5. Case assignments are often influenced by organizational patterns and distinguished by offense types, geographic areas, detective rotations, and existing detective workloads (Greenberg & Wasserman, 1979; Greenwood et al., 1977).

6. It should be mentioned that these calculations do not account for crimes that are not reported to police. Accordingly, the figures presented in this section should be considered liberal evaluations of investigative success.

7. The data used to support these analyses, were collected in cooperation with the HPD and with an endorsement from the Houston Police Officers Union (Hoover, personal communication, September 19, 2014).

8. This duel criteria was selected because only assigned cases are available in these data. Information based on when the offense became known to police are not available.

9. Consider an offense that occurred on the last day of the period of observation (i.e., June 19, 2013). This offense would likely have few, if any, observations of a detective’s initial effort.

10. This measure, however, does not mean that the identified suspect was the person who was arrested. This information was not available in these data. Where multiple arrests existed in a single case (n = 7) the earliest date was used in these analyses.

11. These figures mirror expected counts during the observation period when detective sampling is taken into consideration (not depicted).

12. Cases with an arrest but no initial investigative effort were assigned with the suspect already in custody.

13. Accordingly, subsequent analyses will avoid generalizations to small subsamples.

14. Aggravated assaults with a known suspect did not follow this trend: when there was no known suspect in these cases aggravated assaults were 10.7% less likely to be worked.

15. Homicides with an arrest were contrary to this pattern: when there was no arrest in these cases, homicides were 14.3% less likely to be worked.

16. Aggravated assaults with a known suspect did not follow this trend: when there was no known suspect in aggravated assaults, cases were worked nearly two days slower by detectives.

17. In comparison to other offenses, fewer sexual assaults were worked and among those that were worked, they typically took longer than a week to be worked by detectives (n = 10, 55.6%, $\overline{x}$ = 9.6).

18. It is important to note that the standard deviation among the majority of these estimates exceeds the observed means. This indicates that a few extreme cases may be inflating these results. The prevalence of low mode estimates (≤3) and high maximum values support this supposition.