Dr. Matthew L. Ferrara

Matthew L. Ferrara, PhD, is a licensed psychologist and Licensed Sex Offender Treatment Provider in private practice in Austin, Texas. He is responsible for creating all the major sex offender treatment programs operated by the State of Texas, including the Sexually Violent Predator Program. Along with Professor Mark Stafford, Texas State University, he conducted research that led to the current risk assessment deregistration criteria and methodology. He has trained all the deregistration evaluation specialists in Texas. For more information about deregistration visit Dr. Ferrara’s website: www.deregistertexas.com

Expert Punishment Testimony in a Sex Trial

Sentencing Issues

The genius of Winston Churchill is as manifold as it is undisputed. With the hope of having some of Mr. Churchill’s genius rub off onto this discussion, I would like to use his quote as a jumping off point to discuss sentencing issues in a sex offender case.

I believe that sentencing the sex offender is a riddle wrapped in a mystery, inside an enigma because a satisfactory answer, which is comprehensive and fair, cannot be found. This is true whether you are an attorney for the defense or an attorney for the State.

Let’s unpackage Mr. Churchill’s proclamation as it applies to sentencing in a sex offender trial. At the heart of the sex offender sentencing issue is a riddle. Actually, there are two riddles. Let’s begin with the first riddle.

Riddle #1: What type of offender is least likely to reoffend with the type of crime that originally got them into trouble? Let me simplify the riddle a little bit to give you a chance to answer the riddle correctly. Does a sex offender have a greater chance of reoffending with a new sex offense as compared to the chance of a non-sex offender reoffending with a non-sex offense?

To answer this riddle, you have to come up with two numbers: the percentage of sex offenders released from prison who commit a new sex offense and percentage of non-sex offenders released from prison who commit a new non-sex offense. Do you have those two numbers in mind? Good. Here is what the research says.

In a longitudinal research study, the federal government studied 9,691 sex offenders and 262,420 non-sex offenders released in 1994 from state prisons in 15 states, including Texas (Langan, Schmitt, & Durose; 2003). In the 3-year follow-up period, 5.3% of the sex offenders were arrested for a new sex crime and 68% of the non-sex offenders were arrested for a new non-sex offense.

As you can see, the sex offender re-arrest rate for new sex offenses (5.3%) is considerably lower than the non-sex offense re-arrest rate (68%) for non-sex offenders. But that may not be the biggest surprise of this study, which leads us to the second riddle.

Riddle #2: Which group of offenders produces more sex offenses, sex offenders, or non-sex offenders? To answer this riddle, you need to come up with two numbers: the number of sex offenses produced by known sex offenders and the number of sex offenses produced by convicted criminals with no history of sexual offenses. Here is what the research says.

Using the same governmental study as Riddle #1, the answer to Riddle #2 is this: Sex offenders produce fewer sex crimes than criminals with no history of sexual offending. In this study, 517 sex offenders were arrested for a new sex offense and 3,328 “non-sex” offenders were arrested for a new sex offense. In other words, 92% of new sex offenders are criminals with no prior sexual offense history and 8% of new sex offenses are caused by known sex offenders.

It might be tempting to dismiss these numbers because they are so counterintuitive, but if you were to dismiss government research numbers, you would still have to explain the numbers that come from decades of research generated by non-governmental entities.

Dr. Karl Hanson conducted two sex offender recidivism meta-analysis studies (Hanson & Bussiere, 1998; Hanson & Morton-Bourgon, 2005). A meta-analytic study is a study that examines existing research articles published in peer-reviewed journals. Dr. Hanson’s 1998 study examined decades of peer-reviewed research as documented in 61 published articles about the reoffense rate of 23,393 sex offenders. In his second study, Dr. Hanson retained some of the research articles from his 1998 study, he eliminated other articles, and he added some new articles. The result was a meta-analysis published in 2005 that examined 82 peer-reviewed research articles involving 19,267 sex offenders. The sexual reoffense rates for sex offenders were 13.4% and 13.7%, respectively, for the first and second meta-analytic studies. The rate of sexual reoffending by sex offenders (avg. = 13.55%) found in Dr. Hanson’s studies is considerably lower than the rate of non-sexual offending by non-sex offenders (68%) in the government study.

You might be wondering why the study done by the federal government found sex offenders have a rate of sexual reoffense around 5%, whereas Dr. Hanson’s studies put the sex offender reoffense rate at around 13.55%. The explanation is simple: time. The recidivism study conducted by the federal government followed sex offenders for three years, whereas Dr. Hanson’s numbers rely on studies that followed sex offenders on average six years and in some cases up to 27 years. As the amount of time increases the opportunities for reoffense increases, so the reoffense rate increases.

Mystery. Once the riddles about the sex offender have been answered and you understand that there is a scientific basis for the assertion that few sex offenders reoffend, you are faced with a mystery: Why is the offender least likely to reoffend often given the harshest sentence?

This mystery is only a mystery when you look at the low rate of sexual reoffending among sex offenders. When you look at the nature of the sex offense, it comes as no surprise that the sex offender gets a long prison sentence. Sex offenders typically violate children or women. Judges and juries often use sentencing to punish the sex offender for harming vulnerable victims and as a means of protecting women and children in the general public.

Additionally, sex is not just an act of procreation. In our society and among humans in general, sex is social cement (Diamond, 1993). It is used to cement intimate relationships. The sexual act can touch those involved to the core. When a sex crime is committed, it has the potential of hurting the victim to the core. Even though a third or more of all children who are sexually abused never experience any mental health symptoms (Finkelhor, 1990), in a modicum of cases the sexual abuse victims suffer severe mental health problems. In recognition of the serious harm that sex crimes can or may cause, judges and juries hand out long prison sentences.

Enigma: Given the low threat to the public, most sex offenders should be considered for placement on probation. Given the emotional upshot of a sex offense, most sex offenders should be considered for a term of incarceration. Here is the enigma: An offender can’t receive both a mild and a harsh sentence.

The enigma of sentencing the sex offender lends itself to no easy solution. In fact, attorneys for the state and defense debate and argue in every sex offense trial trying to come up with a solution to this enigma. But the solution is not the province of the defense or the state. It is the province of the trier of fact. The rest of this article is dedicated to a discussion of the type of expert testimony that could help the trier of fact reach a satisfactory solution of the enigma of sentencing in a sex offense trial.

Expert Testimony about Risk Assessment

The best way to talk about sexual reoffense rates during the sentencing phase of a sex offense trial is to have an expert testify about the defendant’s risk for sexual reoffense. The expert’s testimony should be based upon an evaluation of the defendant—which includes the use of formal risk assessment instruments. At a minimum, the expert should provide testimony based on the use of two actuarial risk assessment instruments: Static-2002 and the Level of Service Inventory-Revised.

Before discussing each of these risk assessment instruments, it is important to have an understanding of the basics of modern risk assessment, including how actuarial risk assessment instruments are created and what a risk assessment is and is not.

If you wanted to create a risk assessment instrument for sexual reoffending, you would collect as much research as you could find regarding sexual reoffense among sex offenders. You could not use all the research. You would have to eliminate research that was based upon single-subject case studies or opinions. You would only want to consider studies that relied upon statistical analyses of groups of sex offenders. The best studies would be those that conducted statistical analyses of sex offenders and non-sex offenders over a long period of time.

Once you have statistical studies regarding sexual reoffense among sex offenders, you would go through the studies and identify the factors that have the strongest association with sexual reoffense. For example, child molesters who select unrelated, male children as victims are known to have many sexual abuse victims (Abel, Becker, Mittelman, Cunningham-Rathner, Rouleau & Murphy, 1987). On the other hand, there is research that shows many sex offenders use drugs or alcohol while committing the sexual offense that gets them put on probation or sent to prison, but the research does not identify substance as a factor associated with re-arrest of a sex offender for a new sex offense (Hanson & Bussiere, 1998).

So, if you were constructing a risk assessment instrument for sexual reoffense among sex offenders, you would want to include the factor “male victim who is unrelated to the perpetrator.” You would not want to include in your risk assessment instrument a factor related to substance abuse. As you go through the scientific literature, you would perform this task repeatedly: Add items to your risk assessment instrument with a strong statistical relationship to reoffense and eliminate from consideration any factors with a weak or no statistical relationship to reoffense. When you are done looking at all the scientific research, you will have a list of factors that are “empirically derived”—i.e., the factors are derived from the research.

Empirically derived is not as good as empirically validated. When you empirically validate a factor, you use a scientific study to prove, or validate, that your factor is associated with risk for reoffense. This is exactly how the Static-2002 and the Level of Service Inventory-Revised were created. The authors of these risk assessment instruments created a list of risk factors and then examined how reoffense was related to each of these factors. Each of these instruments was studied in multiple studies involving thousands of offenders. Only those factors that were statistically related to reoffense were included in the final versions of the risk instrument.

Now that you have an understanding of the manner in which a risk assessment instrument is created, it is important to understand what a risk assessment instrument is and is not.

  • What a risk assessment is: A good way to understand what a risk assessment is is to view the results of a risk as­sess­ment in the same way that you look at the gas gauge in your car. If your gas gage is low, you won’t go far. If your gas gauge is high, you can potentially go very far. The same is true of a risk assessment. If a defendant gets a low score on a risk assessment instrument, the defendant has been identified as a person who likely won’t go far down the path towards the target behavior. On the other hand, if the defendant gets a high score on a risk assessment instrument, the defendant has been identified as one who will likely go far down the path toward the target behavior.Just like your gas gauge measures something that exists, a risk assessment measures something that exists. The gas gauge is measuring the gas in your gas tank. The risk assessment is measuring the risk in the defendant. It might be useful to think of the gas gauge and the risk assessment as measures of potential energy. Potential energy is stored energy. Potential energy is the potential to do work, or in the case of a risk assessment, it is the potential to do wrong.
  • What a risk assessment is not: Even though risk assessment is concerned about the future it is not a prediction of future behavior. At this time, it is impossible to predict the future. Risk assessments do not do the impossible. Risk assessments do the mundane, commonplace task of measuring some psychological characteristic that a person possesses right now, in the present.

Using the foregoing discussion of modern risk assessment, it is time to consider the two risk assessment instruments that should be part of the expert’s testimony during the sentencing phase of a sex offender trial.

Sexual Reoffense Risk Assessment

Risk is the likelihood that a person will exhibit a specific target behavior. The Static-2002 can be used to determine a defendant’s risk for sexual reoffense.

The Static-2002 is an actuarial risk assessment instrument that was empirically derived and empirically validated (Phenix, Doren, Helmus, Hanson & Thornton, 2003). The Static-2002 was subjected to rigorous scientific testing, which included the study of sex offenders who were in the community for ten years or more. The Static-2002 has good reliability (i.e., consistently measures risk accurately) and validity (i.e., scores correlate very well with other risk assessment instruments and with future behavior). The Static-2002 is used extensively by Licensed Sex Offender Treatment Providers. Fourteen items comprise the Static-2002.

1.  Current age: 50 years old or older = 0 points; 35 to 49.9 years old = 1 point; 25 to 34.9 years old = 2 points; and 18 to 24.9 years old = 3points

2.  Prior sentencing occasion for sexual offense: no prior sex offenses = 0 points; one prior sex offense = 1 point; two to three prior sex offenses = 2 points; and four or more sex offenses = 3 points

3.  History of juvenile sex offending: no arrest for a sexual offense prior to age eighteen = 0 points; one or more arrests for a sex offense prior to age eighteen = 1 point

4.  Rate of sexual offending: less than one sentencing occasion for a sex offenses every 15 years = 0 points; one or more sentencing occasion for a sex offense every 15 years = 1 point

5.  Any sentencing occasion for a non-contact sex offense: no = 0 points; yes = 1 point

6.  Any male victims: no = 0 points; yes = 1 point

7.  Young, unrelated victims: does not have two or more victims under age 12, one of which is unrelated = 0 points; does have 2 or more victims under age 12, one of which is unrelated = 1 point

8.  Any unrelated victims: no = 0 points; yes = 1 point

9.  Any stranger victims: no = 0 points; yes = 1 point

10.  Any prior involvement with the criminal justice system: no = 0 points; yes = 1 point

11.  Prior sentencing occasions for anything: less than 3 prior sentencing occasions in the individual’s life = 0 points; between 3 and 13 prior sentencing occasions in the individual’s lifetime = 1point; and, 13 or more sentencing occasions = 2 points

12.  Any community supervision violation: no = 0 points; yes = 1 point

13.  Years free prior to index sex offense: more than 36 months free from confinement prior to committing the sexual offense that resulted in the index conviction and more than 48 months free prior to index conviction = 0 points; less than 36 months free from confinement prior to committing the sexual offense that resulted in the index conviction or less than 48 months free prior to the conviction for the index sex offense = 1

14.  Any prior non-sexual violence sentencing occasions: no = 0 points; yes = 1 point

The authors of the Static-2002 have classified scores on the Static-2002 into groups such as low, low-moderate, moderate, high-moderate, and high. Their classification is difficult to use because they use labels such as “low moderate”—e.g., how is a “low” risk offender different from a “low-moderate” offender? In 2007, Calvin Langton, PhD, and a group of researchers developed a research-based Static-2002 classification system (Langton, Barbaree, Hansen, Harkins, & Peacock, 2007). These researchers classified low-risk offenders as having a Static-2002 score of four or less, medium-risk offenders have scores of five or six, and high-risk offenders have scores of seven or higher. Dr. Langton and his colleagues found the sexual reoffense rates for sex offenders in the community for ten years or more for low-, medium-, and high-risk offenders were 4.2%, 15.1% and 27.6%, respectively. Dr. Langton’s classification system is probably the best one to use because it is simple, appeals to common sense, and is based on research.

There is an important limitation to the Static-2002. The scoring guide for the Static-2002 states specifically that if a defendant’s only crime is child pornography, the Static-2002 cannot be used. In those cases, there are alternative sexual reoffense risk assessment instruments that can be used with pornography-only offenders, such as the SVR-20-20 (Boer, Hart, Kropp, & Webster, 1997). Like the Static-2002, the SVR-20 is widely accepted in the field, has been subjected to peer review, and has a known error rate.

The results of the Static-2002 risk assessment can be used during the plea bargaining process as well as during the sentencing portion of a trial. Defense attorneys trying to substantiate their position that a defendant warrants probation could use their client’s low score on the Static-2002 to help the state’s attorney see that the defendant would pose little risk to the public if he is placed on supervised released. On the other hand, the state’s attorney could use a defendant’s high score on the Static-2002 to show the defense that a prison term is warranted and plea bargain discussion about putting the defendant on supervised release make no sense from a public safety standpoint.

Risk Assessment for Violation of Conditions of Probation

Recall that risk is the likelihood that a person will exhibit a specific target behavior. The Level of Service Inventory-Revised is used to determine a defendant’s risk for non-sexual law violations and for violations of his or her conditions of probation.

It is important to understand just how specialized the LSI-R is, and I will call your attention to the fact that the LSI-R can assess a defendant’s risk for violations of his or her conditions of probation. Think about that for a moment. The LSI-R does not just assess an individual’s risk for law violations. It can assess the individual’s risk for violations of such probation rules as “refrain from contact with felons,” “pay probation fines and fees,” and “perform community service.” The fact that the LSI-R can assess risk for such unique requirements as those found in the conditions of probation should not be lost on anyone.

Another matter also deserves your attention. Recall the government study in which 5% of the sex offenders had a new sex crime. This study showed that up to 43% of the sex offenders had a new non-sex crime. Based upon these data, it should be clear that the sex offender’s largest risk to the community is for non-sex offenses, and that LSI-R can provide the trier of fact with scientifically based estimate of the sex offender’s risk for non-sexual offending. Because the sex offender is more at risk for non-sexual reoffense than for sexual reoffense, the LSI-R may be more relevant to sex offender sentencing decisions than the Static-2002 because the LSI-R assesses the sex offender’s most routine form of risk.

As for the LSI-R, it has been around a long time (Andrews & Bonta, 1995). This instrument was first researched and published in the 1970s. The LSI-R went through a major revision in the 1990s, and it has remained more or less the same since then, although there are some variants of the LSI-R (Lowenkamp, Lovins, & Latessa, 2009) and an LSI-R for youth has been developed (Schmidt, Hoge & Gomes, 2005).

The authors of the LSI-R note that the LSI-R samples many of the major and minor risk factors of criminal activity to provide comprehensive risk/needs assessment. There are 54 items on the LSI-R and a defendant is awarded one point for every item that applies to him or her. Defendants with low scores are good candidates for community supervision. In fact, researchers have developed a ranking system for LSI-R total scores: minimum risk (0 to 7), medium risk (8 to 15), and maximum risk (16 and up). The following are the risk categories of the LSI-R:

1.  Criminal History: number of arrests, history violent crime and revocation of supervised release

2.  Education: highest grade completed, any suspensions or expulsions from school, and relationship with teachers and classmates

3.  Employment: willingness to work, ever fired, and relationship with bosses and coworkers

4.  Financial: ability to pay bills and use of public assistance

5.  Family/Marital: relationship with significant other, parents, and siblings

6.  Accommodations: three or more residence changes in a year, satisfactory housing, and high crime neighborhood

7.  Leisure/Recreation: use of free time and participation in group-based recreation activities

8.  Companions: do friends and acquaintances have a criminal orientation

9.  Alcohol/Drug Problems: current or past alcohol or drug problems

10.  Emotional Functioning: current or past emotional problems

11.  Attitude: attitude toward society, societal standards and supervised release

Dr. Simourd (2004) provided a good synopsis of three decades of research regarding the LSI-R. Dr. Simourd notes that the LSI-R has been subjected to scientific study and published in peer-review articles and is accepted for use with male offenders, female offender, minority offenders, Native Americans, and juveniles. Separate studies of the use of the LSI-R with sex offenders show that the LSI-R provides an effective means for assessing risk among sex offenders (Simourd & Malcolm, 1998).

On a practical level, defense attorneys will find the LSI-R useful because most sex offenders do not have a history of criminal behavior, and the LSI-R can be used to statistically demonstrate the defendant’s low risk for non-sexual crimes. Prosecuting attorneys can use the LSI-R in those instances in which the sex offender does have a history of criminal conduct. In these instances, the prosecuting attorney should use the results of the LSI-R to show the defendant’s attorney that the defendant is a risk to the public, and, consequently, plea bargain discussions need to focus on incarceration, not probation.

Treatment and Prognosis

Prognosis is the likelihood that a person will make progress in treatment. In the sentencing portion of a sex offense trial, the defendant’s ability to make progress in sex offender treatment is the concern.

The available research shows that admission of guilt and acceptance of responsibility for the crime are both significantly associated with prognosis (Barrett, Wilson & Long, 2003). Sex offender treatment also places a premium on the offender’s ability to empathize with those harmed by his or her sex crime (Pithers & Gray, 1996). So, an assessment of prognosis should address both accountability and empathy. Some of the criteria for conducting an assessment of prognosis are delineated below:

  • Accountability: The defendant who accepts responsibility should be able to articulate the ways in which his or her crime was wrong. The defendant should be able to discuss how the crime violated societal standards (legal and ethical) and personal standards (personal ethics). When explaining how the crime was wrong, the defendant should not sidestep responsibility by blaming others or blaming the circumstances. Above all, the client should not blame the victim. It is important for the expert to be realistic about the accountability assessment. Taking a step away from sentencing issues and looking at human behavior in general, it is not uncommon for us to fail to see the complete picture of how our mistakes and misdeeds are wrong and most of us also have on occasion blamed others or circumstances for our missteps. Refocusing on the issue of sentencing, to expect a defendant to give “textbook” example of accountability is unrealistic, especially when the magnitude of the misconduct and its consequences are unprecedented in the defendant’s life.
  • Empathy: Empathy is the ability to feel what another person feels. If the defendant can feel the emotional pain that he or she caused others, the defendant can use these feelings as motivation to do well in sex offender treatment. The defendant with fully developed empathy will be able to recognize three categories of victims: Primary Victim (the actual victim); Secondary Victims (the family and friends of the victim and the defendant); and Tertiary Victims (coworkers, general public, those in the criminal justice system, taxpayers, etc.)

The assessment of prognosis is based almost entirely upon the expert’s interview with the defendant, although there is a psychological test that might be useful. The Personality Assessment Inventory (Morey, 1990) is an objective personality test with a statistically validated scale that identifies a person’s readiness for treatment. Treatment readiness is divided into three areas: motivation for treatment, ability to make progress in treatment, and tendencies that might undermine treatment. At this time, the treatment readiness scale of the Personality Assessment Inventory is probably the best empirical measure of prognosis.

Daubert Challenges

As you might expect from the extensive literature cited to this point, the Static-2002 and the LSI-R fare well when facing a Daubert challenge. Below you will find sample responses to each of the four questions of the Daubert challenge.

Since the Daubert challenge specifically asks about error rates, the discussion below includes a description of statistics. For those not familiar with the statistics used in the discussion below, here is a brief explanation of the Area Under the Curve (AUC) statistic, which is the statistic of choice when responding to Daubert questions about error rate.

The AUC statistic is from a family of statistics known as “signal detection” statistics. These statistics were originally developed by scientists trying to answer questions about the accuracy of radar signals (Marcum, 1947). Psychology borrowed the signal detection statistics and used it to measure a variety of factors, including human judgment (Tanner & Wilson, 1954). As the years passed, signal detection theory became more sophisticated and the AUC statistic was developed.

Even though the AUC statistic is sophisticated, its interpretation is simple. For example, an AUC statistic with a value of .50 equals chance—e.g., when flipping a coin, you have a 50% chance of being right, or AUC = .50 of being right. If on the other hand, you were correct in guessing the occurrence of an event 70% of the time, your AUC statistic would be equal to about .70.

Static-2002 Daubert Challenges

1.  Has the theory or technique been tested or is it subject to being tested?

  • The authors of the Static-99 developed the Static-2002.
  • These authors felt it necessary to improve upon the static-99 so there would be:
    ▶ Increased conceptual clarity
    ▶ Improved inter-rater reliability
    ▶ Elimination of paradoxical scoring
    ▶ Improved predictive accuracy
  • There were 4,596 subjects used to validate the Static-2002, with 1724 subjects coming from the United States (SOTEP California = 1137; Washington State = 587)

2.  Has the theory or technique been subjected to peer review and publication?

  • There have been at least 20 studies in which the reliability and validity of the Static-2002 has been assessed.
  • Langton et al (2007) did correlational studies of the Static-2002 and other sex offender risk assessment instruments and found that the Static-2002 correlated very well with those instruments: Static-99 (r = .80), SORAG (r = .71), RRASOR (r = .69), MnSOST-R (r = .58), and VRAG (r = .54)
  • Langton et al. (2007) found cut scores did a good job of identifying low-, medium-, and high-risk offenders. For a 10-year follow-up period, those with a score of 4 or less recidivated at a rate of 4.2% or less; those with scores of 5 or 6 recidivated at a rate of 15.1%; and those with scores of 7 or higher recidivated at a rate of 27.6%.

3.  What is the known or potential rate of error in applying the particular scientific theory or technique?

  • In the normative sample for the Static-2002, the authors found that it had an AUC = .716 for predicting sexual recidivism.
  • Langton et al. (2007) followed 464 sex offenders for an average of 5.9 years and obtained a AUC = .71.

4.  To what extent has the theory or technique received general acceptance in the relevant scientific community?

  • It is becoming the risk assessment instrument of choice among mental health professionals conducting risk assessments.
  • It might take longer for probation and parole departments in Texas to adopt the Static-2002 and replace the static-99 because it would cost money to train supervision officers to use the Static-2002.

LSI-R Daubert Challenges

1.  Has the theory or technique been tested or is it subject to being tested?

  • The Level of Service Inventory was created in the late 1970s and was revised and published in 1995 as the Level of Service Inventory-Revised. The Level of Service Inventory-Revised had been the subject of intensive study since that time, and hundreds of peer reviewed research studies have been published regarding the Level of Service Inventory-Revised. The Level of Service Inventory-Revised manual contains a listing of approximately 50 peer reviewed works about the Level of Service Inventory-Revised.

2.  Has the theory or technique been subjected to peer review and publication?

  • The Level of Service Inventory-Revised manual contains a listing of approximately 50 peer-reviewed works about the Level of Service Inventory-Revised.

3.  What is the known or potential rate of error in applying the particular scientific theory or technique?

  • Simourd (2004) reports that with respect to psychometric properties, the Level of Service Inventory-Revised has shown to have good internal consistency (coefficient alpha = .72), inter-rater reliability (r = .94) and temporal stability (r = .80).
  • “The false negative rate for the Level of Service Inventory-Revised is usually found to be approximately 2 to 3 percent. This means that when an individual is placed in low security based on a Level of Service Inventory-Revised score, there will rarely be any major problems with that individual. Andrews (1982) also found that even when an individual with a low Level of Service Inventory-Revised score transgresses, it was usually a minor incident” (Level of Service Inventory-Revised Manual, pg. 47).
  • When assessing female offenders, an AUC = .63 for total LSI-R score and general recidivism was obtained (Rettinger and Andrews, 2010).
  • When assessing male offenders, an AUC = .78 was obtained for total LSI-R score and general recidivism (Loza & Green, 2003).

4.  To what extent has the theory or technique received general acceptance in the relevant scientific community?

  • Simourd (2004) reports that the Level of Service Inventory-Revised had been successfully employed as a classification/management tool among an array of offender groups, including probationers, male inmates, female offenders, Native American inmates, juvenile offenders, and sex offenders.
  • The Level of Service Inventory-Revised is used in the Canadian Criminal Justice System and in many states in the US. It is also used in many European nations.

Direct Examination Questions

The following are suggested direct examination questions for an expert testifying about risk in the punishment phase of a sex offense case. The questions listed below are sufficient for eliciting all the relevant information from the risk assessment. Of course, you may want to add questions and explore additional matters during direct examination depending on the specific facts of your case.

1.  Please state your name.
2.  How are you employed?
3.  As a licensed psychologist what kind of work do you do?
4.  As a Licensed Sex Offender Treatment Provider, what kind of work do you do?
5.  Describe your educational background?
6.  What experience do you have that is relevant to this case?
7.  What did I ask you to do in this case?
8.  Have you done that kind of work before?
9.  Have you ever been designated as an expert and testified in court about your analysis and findings when you have done this type of work?
10.  Have you testified for the state as well as defense attorneys?
11.  What methodology did you use to conduct your risk assessment of Mr. Upshot?
12.  After reviewing all the evidence, do you have an opinion regarding Mr. Upshot’s risk for future misconduct?
13.  What is your opinion?
14.  What do you base that opinion on?
15.  Could you identify the first risk assessment instrument that you used and tell us what you found?
16.  Could you identify the second risk assessment instrument that you used and tell us what you found?
17.  Did you assess Mr. Upshot’s prognosis?
18.  What is prognosis?
19.  What is Mr. Upshot’s prognosis?
20.  What is your overall opinion regarding Mr. Upshot?
21.  Pass the witness.


Experienced experts who testify about risk and treatment will tell you that even though there are an infinite number of possible cross-examination questions, there are a finite number of categories from which cross-examination questions emanate. In preparing for cross-examination, it might be useful if the expert has reviewed the scientific literature and is prepared to respond to cross-examination questions that will likely come from one of four categories.

  • Group vs. Individual Statistics: The attorney conducting cross-examination of the expert may ask: “These risk assessment instruments yield reoffense rates for groups of offenders. How can you be sure that these group statistics apply to this individual?” Response: The expert should explain that the AUC statistic elucidates just how accurate the risk rating is. The AUC statistic would equal .50 if you were trying to guess the flip of a coin. For both the Static-2002 and the LSI-R, the AUC statistic is around .70, which is a 40% increase over just guessing.
  • Hypothetical: Some attorneys get into the guts of the Static-2002 and create a hypothetical about a sex offender whose offenses don’t match up with the scoring criteria on the Static-2002. For example, the attorney might ask: “Let’s say that the defendant had 1,000 sex offenses for which he was never caught. What would his risk level be on the Static-2002?” Response: Aside from being an improper hypothetical because it is not based upon the facts of the case, this hypothetical should not be a problem for the expert. The expert should agree that the Static-2002 cannot count events that are unknown, but once they are known, the Static-2002 does a good job of incorporating them into the risk assessment.
  • Soft Science: The attorney conducting cross-examination might denigrate the reliability of the risk assessments because risk assessments are soft sciences and as such they are not as trustworthy as hard sciences like medicine. Response: The AUC statistic used to measure the reliability of the risk assessment is also used to measure the reliability of medical techniques, like the x-ray. The AUC the curve statistic for x-rays of the hip is around .74 to .77 (Kirby & Spritzer, 2010), which is about the same as the AUC statistic for the Static-2002 and the LSI-R. The correlation coefficient is often used to explain the efficacy of risk assessments and medical procedures. For example, the LSI-R routinely has correlation coefficients of .30 to .50, between total LSI-R score and reoffense (Andrews & Bonta, 1995). This correlation is many times larger than the correlation for some familiar and trusted “hard science” events: correlation = .02 between aspirin and reduced risk of death by heart attack; correlation = .08 between ever smoked and lung cancer in the next 25 years; correlation = .11 antihistamine use and reduced runny nose and sneezing; correlation = .14 between non-steroidal anti-inflammatory and reduced pain; and correlation = .38 between Viagra and sexual performance (Meyer, Finn, Eyde, Kay, Moreland, Dies, Eisman, Kubiszyn, & Reed, 2001). All of these “hard science” correlations are a fraction of the correlations found for the “soft science” risk assessment techniques.
  • Moral Outrage: An attorney may question the expert about the inability of the risk assessment to measure the pain of the victim or the outrage of the community about the defendant’s misdeeds. Response: The risk assessment instrument does one thing and it does it very well. It measures risk. If we could only function as efficiently as the risk assessment instrument, we would be doing very well.


“After all, facts are facts, and although we may quote one to another with a chuckle the words of the Wise Statesman, “Lies—damn lies—and statistics,” still there are some easy figures the simplest must understand, and the astutest cannot wriggle out of.”

Mark Twain has attributed this quote to the famous British politician Benjamin Disraeli, even though careful review of Disraeli’s public speeches suggests that he never uttered those words. According to Martin (2010), the earliest use of this quote can be traced to a speech made by Leonard H. Courtney in New York, circa 1895. This quote is often shortened to “lies, damned lies, and statistics” when describing the testimony of the risk assessment expert.

Those inclined to see risk assessments as either “lies” or “damned lies” fail to appreciate the risk assessment zeitgeist, which is not limited to the area of sex offenders. Many risk assessment instruments have been empirically derived and empirically validated for a broad number of target behaviors. A sampling of some of the more popular risk assessment instruments includes the following:

  • History Clinical Risk—Twenty Factors (HCR-20): This instrument assesses the risk of adult males and females for acts of physical violence (Douglas & Webster, 1999).
  • Spousal Assault Risk Assessment (SARA) Guide: If you want to know if a man or a woman is a risk for domestic partner violence, you would use this instrument (Kropp & Hart, 2000).
  • Hare Psychopathy Checklist-Revised: The psychopath is the individual who produces the most crime and the most dramatic crimes, and with over 550 articles to its credit, this instrument does an unimpeachable job of classifying psychopaths (Hare, 2003).
  • Risk Assessment Scale for Prison—Capital Offender: It has been proven that all other risk assessments do not accurately assess capital offenders for future risk, so you should use this instrument if you need to know if an individual convicted of capital murder will be violent if the individual is not given a death penalty and placed in the general prison population (Cunningham & Sorenson, 2007).
  • Child Abuse Potential Inventory: For those cases where a parent is suspected of physically abusing a child, the CAPI can be used to identify those adults who are a high risk for physically abusing children (Milner, 1990).
  • Estimate of Risk of Adolescent Sexual Offense Recidivism (ERASOR): This instrument assesses an adolescent sex offender’s risk for new sex offenses (Worling & Curwen 2001).
  • Structured Assessment of Violence risk in Youth: Version 1.1 (SAVRY): Given the notoriety of some of the violence perpetrated by youth, you can use this instrument to determine if a juvenile is at risk for physical violence (Bartel, Borum & Forth, 2002).
  • Psychosocial Evaluation and Threat Risk Assessment (PETRA): When assessing school violence situations like the Columbine event, PETRA provides an actuarial assessment of threats made by students in school settings (Schneller, 2005).
  • Workplace Assessment of Violence Risk—Twenty-One Factors (WAVR-21): This instrument is designed to determine if an individual poses a threat of violence in the workplace, against coworkers or employers (White & Meloy, 2007).

The primary limitation to the use of risk assessment during sentencing is that it is a highly intellectual approach. In general, people use intuition and emotion more often than intellect when making decisions. There is research that shows we make up our mind based upon feeling, intuition, or emotion and then use intellect to justify the decision (Goleman, 1994). So, the risk assessment approach is limited because it does not appeal to the modal manner in which almost all decisions are typically made.

The only way to compensate for the limitation of risk assessment data is to couple it with emotional testimony. The defense attorney would want to pair risk assessment testimony with character witness testimony to get the most effective punishment testimony. The prosecuting attorney would want to pair risk assessment testimony with testimony from the victim or the victim’s family.

Having completed thousands of risk assessments, I would like to close this article by telling you something that only an old hand at risk assessments could pass along. In my opinion, a risk assessment is like the unthinking millstone that blithely crushes all in its path. Like the milestone, the risk assessment doesn’t care who the defendant is or who hired the expert to crunch the numbers. The risk assessment yields invariant results that do not cater to anyone’s bias, so it is up to the discerning attorney to use these assessments wisely.


Abel, G. G., Becker, J. V., Mittelman, M., Cunningham-Rathner, J., Rouleau, J. L. & Murphy, W. (1987) Self-reported sex crimes of non-incarcerated paraphiliacs. Journal of Interpersonal Violence, 2, 3–25.

Andrews, D. A. & Bonta, J.L., (1995) The Level of Service Inventory-Revised: Users Manual. Toronto: Multi-Health Systems, Inc.

Barrett, M., Wilson, R. J., & Long, C. (2003) Measuring motivation to change in sexual offenders from institutional intake to community treatment. Sexual Abuse: A Journal of Research and Treatment, 15, 269–283.

Borum, R., Bartel, P. & Forth, A. (2002). Manual for the Structured Assessment of Violence Risk in Youth (SAVRY), Version 1. Tampa, Florida: University of South Florida.

Boer, D. P., Hart, S. D., Kropp, R. P., & Webster, C. D. (1997). Manual for the Sexual Violence Risk-20. Vancouver, Canada: British Columbia Institute against Family Violence.

Cunningham, M. D., & Sorensen, J. R. 2007 Capital Offenders in Texas Prisons: Rates, Correlates and Actuarial Analysis of Violent Misconduct, Law and Human Behavior, Volume 31, 553–571.

Diamond, J. (1993) The Third Chimpanzee: The Evolution and Future of the Human Animal. New York: Harper Perennial.

Douglas K. S. & Webster, C. D. (1999) The HCR-20-20 Violence risk assessment scheme: Current validity in a sample of incarcerated offenders. Criminal Justice and Behavior, 26, 3–19.

Finkelhor, D. (1990) Early and long-term effects of child sexual abuse: An update. Professional Psychology: Research and Practice, 21, 325–330.

Goleman, D. (1994) Emotional Intelligence: Why It Can Matter More than IQ. NY: Bantam Books.

Hanson, R. K. & Bussiere, M. T. (1989) Predicting relapse: A meta-analysis sexual recidivism studies. Journal of Consulting and Clinical Psychology, 66, 348–362.

Hanson, R. K. & Morton-Bourgon, K. E. (2005) The characteristics of persistent sexual offenders: A meta-analysis of recidivism studies. Journal of Consulting and Clinical Psychology, 73, 1154–1163.

Hare, R.D. (2003). Psychopathy Checklist-Revised Technical Manual Second Edition. Toronto: Multihealth Systems, Inc.

Kirby, M. W. & Spritzer, C. (2010) Radiographic detection of hip and pelvic fractures in ED: Results. American Journal of Roentgeol, 4, 1054–1093.

Kropp, P. R. & Hart, S. D. (2000) The Spousal Assault Risk Assessment (SARA) Guide: Reliability and validity in adult male offenders. Law and Human Behavior, 24, 101–118.

Langan, P., Schmitt, E. L., & Durose, M. R. (2003) Recidivism of Sex Offenders Released from Prison in 1994. Washington, DC: US Department of Justice Office of Just Programs Bureau of Justice Statistics.

Langton, C. M., Barbaree, H. E., Hansen, K. T., Harkins, L. & Peacock, E. J. (2007) The reliability and validity of the Static-2002 among adult sex offenders with reference to treatment status. Criminal Justice and Behavior, 34, 616–640.

Lowenkamp, C. T., Holsinger, A. M., and Latessa, E. J. (2001) Risk/need assessment, offender classification, and the role of childhood abuse. Criminal Justice and Behavior, 28, 543–563.

Lowenkamp, C. T., Lovins, B., and Latessa, E. J. (2009) Validating the Level of Service Inventory-Revised and the Level of Service Inventory: Screening Version with a sample of probationers. The Prison Journal, 89, 192–204.

Loza, W. & Green, K. (2003) The Self-Appraisal Questionnaire: A self-report measure for predicting recidivism versus clinical-administered measures: A 5 year follow-up study. Journal of Interpersonal Violence, 18, 781–797.

Marcum, J. I. (1947) A Statistical Theory of Target Detection by Pulsed Radar. The Research Memorandum: 90. http://www.rand.org/pubs/research_memoranda/RM754/. Retrieved 2009-06-28.

Martin, G. (2010) The Phrase Finder. Retrieved from: http://www.phrases.org.uk/meanings/31000.html.

Mills, J. F., Jones, M. N. & Kroner, D. G. (2005) An examination of the generalizability of the LSI-R and Vrag probability bins. Criminal Justice and Behavior, 32, 565–585.

Milner, J. S. (1990) An Interpretive Manual for Child Abuse Potential Inventory. DeKalb, IL: PsyTech.

Meyer, G. J., Finn, S. E., Eyde, L. D., Kay, G. G., Moreland, K. L., Dies, R. R., Eisman, E. J., Kubiszyn, T. W. & Reed, G. M. (2001) Psychological testing and psychological assessment.: A review of evidence and issues. American Psychologist, 56, 128–156.

Morey, L. (1991) The Personality Assessment Inventory Manual. Lutz, FL: Psychological Assessment Resources.

Phenix. A., Doren, D., Helmus, L., Hanson, R. K. & Thornton, D. (2003) Coding Rules for Static-2002. Ottawa: Department of Solicitor General Canada.

Pithers, W. D. & Gray, A. S. (1996) The utility of relapse prevention in treatment of sexual abusers. Sexual Abuse: A Journal of Research and Treatment, 8, 223–230.

Rettinger, L. J. & Andrews, D. A., (2010) General risk and need, gender specificity and recidivism of female offenders. Criminal Justice and Behavior, 37, 29–46.

Schmidt, F., Hoge, R. D., & Gomes, L. (2005) Reliability and Validity analyses of the Youth Level of Service/Case Management Inventory. Criminal Justice and Behavior, 32, 329–344.

Schneller, J. (2005) Psychosocial Evaluation and Threat Risk Assessment. Lutz, FL: Psychological Assessment Resources.

Simourd, D. J. (2004) Use of dynamic risk/need assessment instruments among long-term incarcerated offenders. Criminal Justice and Behavior, 31, 306–323.

Simourd, D. J. & Malcolm, P. B. (1998) Reliability and validity of the Level of Service Inventory-Revised among federally incarcerated sex offenders. Journal of Interpersonal Violence, 13, 26274.

Tanner J., Wilson P., John A. Swets, J. A. (1954) A decision-making theory of visual detection. Psychological Review, 61, 401–409.

White, S. G. & Meloy, J. R. (2007) Workplace Assessment of Violence Risk—Twenty One Factors. San Diego: Specialized Training Services.

Worling, J. R. & Curwen, T. (2001) Estimate or Risk of Adolescent Sexual Reoffense Recidivism Version 2.0. Ontario: Thistletown Regional Centre for Children and Adolescents, Ontario Ministry of Community and Social Services.