# 5.2 How Can Complexity Be Measured?

A reasonably straightforward approach is for the prosecutor’s office to identify key factors influencing prosecution decisions, and then assess the extent to which individual cases contains these factors. A key issue for this procedure is how to rate the complexity of each case in a way that is reasonably reliable.

The simplest approach is to ask prosecutors to rate the complexity of each case using a pre-determined list of factors believed to make a case complex. Each prosecutor would assess the case based on the presence or absence of each factor. The complexity rating for each factor could be only “yes” or “no”. After considering the number of “yes” factors, the prosecutor would place the case into one of three or four levels of complexity, such as: (1) very complex; (2) somewhat complex; 3) somewhat straightforward; or (4) straightforward. This type of rating is subjective. Not all prosecutors would judge the same case the same way, and the ratings of a prosecutor might be affected by how they felt on a given day.

The problem of subjectivity can be somewhat alleviated by having two or even three prosecutors separately rate each case. This approach also would be strengthened by providing detailed definitions of the complexity levels, including the presence or absence of selected complexity factors. A candidate list of such factors is contained in Exhibit 5-1 below.

A second, more involved approach increases the precision of the complexity ratings. This involves developing a computerized calculation of complexity for each case as in the first approach, but each factor would be assigned a numerical value within a rating scale. For each case, a prosecutor would rate each complexity factor on its level of influence on the case.

An example of such a complexity rating system is presented in Exhibit 5-2 below. The rating system uses three levels of influence on the case valued at 0, 1, and 2. The system allows for additional nuance of each factor to be accounted for in the scoring process.

Once each factor has a rating, a user can then add the ratings across each factor to come up with an overall complexity score, which then can be grouped into a small number of overall complexity levels, such as: (1) high complexity; (2) medium complexity; and (3) low complexity. The prosecutor’s office must establish the range of values for each of these levels. For example, “low complexity” cases might range from 0-5; “medium complexity” might range from 6-10;  and “high complexity” might be 11 and over.

Simply adding the ratings across factors implies that each one has equal importance and influence. This is unlikely to be true. For example, victim intoxication may make a case exponentially more complicated than other complexity factors. Thus, prosecutors’ offices may want to consider weighting complexity factors based on their perception of added complexity in a case. Rather than using 0, 1, and 2 in the rating system, prosecutors could assign a number weighted based on relevance to complexity. So, a victim’s lack of injury may be rated a 2, but victim intoxication rated an 8. This process allows certain factors to weigh more heavily on the overall complexity score.

The process for adding the ratings of each factor and then assigning a level of case complexity could be digitized based on whatever software program an office chooses to use – meaning prosecutors would not need to make manual calculations. From there, only simple analysis is needed to determine the number and percentage of cases that fall into the three or four complexity categories during a particular reporting period. Next, this data can be linked to each case’s level of success to provide the percentage of cases at each complexity level with successful outcomes.

The computer-calculated rating system is preferable because of its reduced subjectivity. Some offices, however, might want to start out with the simpler procedure and then switch to the computer-calculated procedure at a later date.

More details on complexity rating procedures are provided in Exhibit 5-3 below.

##### Initial List of Candidate Factors for Complexity Ratings88

1. Lack of statement from the defendant.
2. Lack of DNA or other forensic evidence, especially if the offender was unknown to the victim.
3. Lack of physical evidence that the victim was assaulted (g., damaged clothing).
4. Lack of third-party witnesses to events or statements before, during, or after the assault.
5. Lack of sexual assault kit evidence.
6. Lack of documented physical injury.
7. Lack of supportive testimony from medical forensic examiner.
8. Lack of other corroborating evidence.
9. Lack of use of a weapon by the offender.
10. Prior sexual relationship between the victim and offender.
11. Offender claim of consent to act.
12. Victim is unable to participate in prosecution.
13. Witness is unable to participate in prosecution.
14. Victim has a disability.
15. Significant difficulties understanding the victim’s or offender’s speech.
16. Victim has a history of mental health issues, alcohol/drug use/abuse, and/or activity   associated with commercial sexual exploitation.
17. Offender has a history of mental health issues.
18. Lack of offender history of sexual abuse or other crimes (felonies or misdemeanors).
19. Victim’s use of alcohol or drugs at time of assault.
20. Offender’s use of alcohol or drugs at time of assault.
21. Lack of physical resistance by the victim.
22. Delayed report.
23. Significant forensic and/or digital content in case (requiring major effort and cost).
24. Prominent status of offender or offender’s connections/family.
25. Significant political/media attention pressuring prosecution.
26. Unusual elements with which prosecutors have little or no experience.
27. Involvement of multiple offenders.
28. Consensual interactions between offender and victim at, or near, the time of the assault.
29. Offender has a disability.
##### Example of a Complexity Rating Scale

Adapted from Complexity Rating Scale Developed by Assistant District Attorney, Rick Spoentgen, Sauk County District Attorney’s Office

1. Statement from the defendant

0 – Confession / very inculpatory admission
1 – Partial admission with slight inculpatory statements
2 – No statement / strong consent defense (i.e., only part of defendant’s account that conflicts with other available evidence account is issue of consent)

0 – Autosomal / strong Y-STR return in non-consent case
1 – Autosomal / strong Y-STR return in consent case – weak Y-STR return in non-consent case
2 – No match

1. Lack of sexual assault kit evidence

0 – Kit evidence with highly probative/inculpatory findings
1 – Kit complete but with nothing probative
2 – No kit

1. Lack of other corroborating evidence

0 – Probative corroborative evidence exists
1 – Corroborative evidence exists, but not probative
2 – No corroborating evidence exists

1. Consent defense

0 – Inapplicable or not claimed
1 – Imperfect consent defense (i.e., some parts of defendant’s account match other available evidence, but not all)
2 – Strong consent defense (i.e., only part of defendant’s account that conflicts with other available evidence account is issue of consent)

1. Lack of victim participation

0 – Victim is active participant
1 – Victim is partial participant (labor intensive/reluctant)
2 – Victim is non-participant / unwilling participant (openly hostile /unable to be located)

1. Lack of witness participation

0 – Witnesses participating
1 – Witnesses reluctant / statement changing
2 – Witnesses nonparticipating

1. Presence of credibility concerns with victim’s account

0 – No relevant issues
1 – Has issues raising of credibility concerns
2 – Has issues which significantly impact case or credibility

1. Lack of defendant history of sexual assault

0 – Significant probative history found
1 – Slight history / debatable probative value
2 – No history found

1. Victim use of alcohol or drugs at time of incident

0 – No use OR use to the extent it clearly shows victim is incapacitated
1 – Slight-moderate use with limited impact on perception, memory, or other issues related to credibility
2 – Moderate-heavy use with severe impact on perception or memory, but not incapacitated

1. Consensual sexual interactions at or near time of assault89

0 – No physical intimate contact within a week of incident
1 – Intimate contact without sex within week of incident
2 – Full oral/anal/vaginal/digital sex within week of incident

1. Complicating legal factors/motions:

0 – None / boilerplate
1 – Novel and not time/labor intensive OR standard motion and highly fact-specific
2 – Novel and time/labor intensive

##### Step 1: Identify and select the complexity factors to be assessed in each case.90

Exhibit 5-1 above provides a non-comprehensive list of factors. Prosecutor’s offices should solicit input from as many sexual violence prosecutors as is feasible, not only to increase the accuracy of the factors identified but also to increase prosecutors’ interest in, and understanding of, the information that comes from this process.

##### Step 2: Selecting a rating system.

The individuals deciding the ratings for each case’s complexity should use this scale.

Option A: The simplest option is to count the number of complexity factors, developed in Step 1, that are present in a particular case and sum the total. For instance, if a prosecutor is using the above suggested list of complexity factors, and has determined that 5 are present in a particular case, each factor would add a value of 1 to the rating, and thus the complexity rating for that case would be 5.

Option B: A more challenging option is to rate each factor according to the extent it is present in the case using a pre-determined rating scale (like that discussed earlier), rather than only assessing the presence or absence of a factor as in Option A. For instance, the complexity factor might have three levels (0-2), where “0” means the particular factor did not increase case difficulty at all and “2” means the factor greatly increased case complexity. Prosecutors’ offices can enhance the reliability of these complexity scores by developing definitions for each grade of each complexity factor (as exhibited in Exhibit 5-2 above).

For example, when rating the factor, “supportive testimony from medical forensic examiner,” if the testimony indicated that the defendant was guilty, the complexity factor would likely be rated 0 or 1. However, if the testimony was weak, the rating would likely be 2 or 3. If no such evidence was available, the rating would depend on the circumstances of the case. If the lack of such evidence was a minor problem, a 0 or 1 might apply, but if it was a major problem, a rating of 2 might be appropriate.

Another example: When rating “statement from the defendant,” if the statement was particularly powerful, the factor would likely be rated 2. If the defendant did not testify on his behalf, the rating would depend on how the prosecutor judged the effect of that choice; if the statement inculpated the defendant, the factor might be 0.

Option C: Yet another option is to consistently weight the complexity factors across all cases, while varying the weight of each factor to abide by the site’s predetermined value. For instance, “involvement of multiple offenders” (#24 in Exhibit 5-1 above) could be assigned a weighted numerical complexity rating of 2, while “lack of cooperation of witness” (#13 in Exhibit 5-1 above) could be assigned a weighted numerical complexity rating of 5. Thus, in all cases with a disabled offender, the factor would be assigned a rating of 2, and in all cases that lack witness cooperation, the factor would be assigned a rating of 5.

The upside of this approach is that it reduces some of the subjectivity present in Option B. However, it also reduces the precision of the case’s overall complexity rating. For instance, perhaps the evidence available in your case was so strong that the lack of a cooperating witness did not add much complexity to the case; you would still need to add a weighted complexity factor of “5” to the case’s overall complexity score.

Exhibit 5-4 below provides an example of Options, A, B, and C for a subset of complexity factors. The first column names a subset of five factors, and the second column illustrates a system that rates the presence/absence of each factor (Option A). Each factor, if present, would receive a score of 1 with a total possible case complexity score of 5. The third column illustrates a system where an attorney would rate the extent to which each factor matters for particular cases on a scale from 0-2 (Option B). The total possible case complexity score would be 10. The last column illustrates a system where factors would be consistently weighted across all cases, with the value predetermined by sites (and represented in each row) (Option C). The total possible complexity score using this system would be 12.5.

##### Step 3: Select a set of three, four, or five non-numerical categories that represent overall levels of case complexity.

For example, a three-level complexity scale might select the categories: “Low Complexity,” “Medium Complexity,” and “High Complexity.” Sites with small numbers of sexual violence cases might prefer fewer categories than sites with large numbers. The total score for each case (from Step 2 above) is used to place each sexual assault case into one of the complexity levels, using a procedure like that discussed below in Step 4. The calculations preferably would be done automatically using computer software, not manually.

##### Step 4: Select the numerical complexity ranges for each level identified in Step 3.

This will depend in part on the number of factors selected in Step 1 above and non-numerical categories selected in Step 3 above. (The more the complexity factors, or the fewer the non-numerical categories, the larger the ranges will be). The simplest approach is to use equal intervals calculated by dividing the maximum case score by the number of levels chosen (see Step 3 above).

For example, if the prosecutor’s office selects five complexity factors (Step 1) and uses a rating scale of 0-2 (Step 2), the range of possible values for a case would be 0-15. If the office selected the three levels of complexity noted in Step 3 and choses to use equal intervals, the ranges would then be:

Low Complexity = 0 – 5
Medium Complexity =  6 – 10
High Complexity = 11 – 15

Instead of using equal intervals, the office may choose to “validate” the complexity scale using a small number of cases and cutting the intervals at the point it makes sense in reality. This would involve looking at how cases fall within categories and determining what the thresholds are for each complexity levels — i.e., is a case one of medium complexity with a score of 4; and one of high complexity with a score of 13? If so, the ratings might be adjusted to reflect 0-4 = Low Complexity; 5-13 = Medium Complexity; and 13-15 = High Complexity.

A problem can arise when the individual examining a case believes it to be of high complexity even when its score does not reflect this. In this case, an option is to add a procedure to override the automatic calculation. In that case, a written explanation for the override should be provided. The site is advised to test its procedure with a sample of cases (perhaps 10% of a caseload within a given timeframe) to decide whether the option is needed.

##### Step 5: Develop a process to systemically record information regarding case complexity for every matter that comes to the prosecutor’s office.

The process should specify who is responsible for calculating the ratings of each factor and assessing the category of complexity in which it falls. Options include: the prosecutor assigned to the case, a prosecutor who has not worked on the case, and the unit chief. The process should also specify at which point in a case the complexity should be rated. Case complexity could be evaluated: early in the case (e.g., at the time of charging — which would make these ratings useful in case and office workload planning); and/or after the case is resolved (where complexity can be compared with level of case success). Offices may have varying opinions about the utility of case complexity ratings at each stage of the case, and these opinions should be considered when determining the timing. For example, early ratings can help prosecutors estimate the amount of effort and resources the case may require, and encourage them and other team members to identify ways to meet challenges posed by the identified complexity factors. The risk is that such a determination early on may bias the process in favor of a less than optimal disposition because of the perceived difficulty of securing a conviction at trial. Offices might consider not disclosing early complexity ratings until after the case is resolved.

Rating complexity after the case has been resolved may provide a more accurate picture of the complexity. The post-resolution case review could encourage prosecutors to identify challenges in future cases as a way to improve their practices.

Ratings at both points — early in the case and after disposition — may be best, where feasible. Conducting both assessments may yield important information, such as the usefulness of an expert witness in securing a favorable outcome, and may challenge the prosecutor’s own pre-conceived ideas about the case. However, complexity factors rating will take additional prosecutor time.

Offices might wish to seek outside assistance to help set up the procedures. Faculty or students from nearby universities, colleges, etc., might be available to help. Once the process has been established, implementing it can become a routine task.

##### Step 6: Once the complexity rating process is established, an analytic software program or case management system can calculate and provide the number and percentage of cases for each complexity level and any reporting period, quickly and accurately.

The prosecutors’ role, once the process is implemented, is to make and record their ratings of each case, as called for by Step 2 above.

The analytic software program can calculate the “percentage of guilty resolutions (and/or not guilty resolutions) of cases resolved during the reporting period that fell into each level of complexity. Resolutions should be considered alongside determinations of case success, which is discussed in detail in Chapter 7. This could provide outcome measures in the following form: Percentage of moderately complex cases within a particular reporting period resolved with satisfactory outcomes and a higher number of promising practices implemented. The analytic process also could provide outcome values categorized by each victim characteristic or circumstance group, such as “percentage of moderately complex cases involving a teenaged victim with successful outcomes.” A detailed discussion of victim demographic subgroups can be found in Chapter 4.

##### Methodology Comparison: Case Complexity Factors for a Prosecutor’s Office
 Complexity Factor Option A: Rating Factors Based on Presence in Case Option B: Rating Factors on a Scale of 0-2 in Each Case Option C: Weighting Factors Consistently Across all Cases Lack of SAK evidence 1 0-2 1 Prior sexual relationship between victim and offender 1 0-2 3 Victim’s use of alcohol or drugs 1 0-2 4 Lack of participation by victim 1 0-2 3 Delayed report 1 0-2 1.5 Total possible case complexity score 5 10 12.5

88 This initial list of candidate factors for complexity ratings is based on findings from research on sexual violence case attrition, i.e., reasons victims do not report sexual violence cases, reasons police close cases without referring them to prosecution, and reasons prosecutors decline to charge cases. See Chapter 3 for more on case attrition.

89 Although consensual sexual conduct following sexual violence is a common victim behavior, this factor is listed here because such a circumstance would need to be properly explained to the factfinder. It is also listed here because it may raise questions as to the offender’s identity, if there is an identity defendant, since multiple individuals’ DNA may be extracted from a sexual assault kit.

90 Uniform definitions of complexity factors and complexity levels across prosecution offices would be ideal, facilitating comparisons among prosecutor offices to identify successful (“best”) practices. However, this is unlikely to be immediately feasible. It may be better, at least initially, for each community to select its own definitions and parameters for rating case complexity.