Sex Offenders in the Community: Post-Release, Registration, Notification and Residency Restrictions

Karen J. Bachar: Today's presentation is on “Sex Offenders in the Community: Post-Release, Registration, Notification and Residency Restrictions.” The management of sex offenders in the community post-release is an issue of increasing concern to law enforcement, policymakers and the public.

In recent years, there've been many efforts made to strengthen registration and notification. At the same time, we haven't paid a lot of attention to related matters, such as how residency restrictions may impact offenders' efforts to find work and stability, how their living arrangements — what happens when they get released from prison, whether rates of recidivism have changed according to these policies and whether these policies have succeeded in increasing public safety. These are some issues that a lot of people ask.

And to talk about these and related issues to sex offenders, I am happy to have a panel of experts to share their information.

First up will be Dr. Elizabeth Letourneau, who is an associate professor in the Department of Psychiatry and Behavioral Sciences at the Medical University of South Carolina.

She's worked on Family Services Research Center since 2000 and studied various aspects of sex offending and sexual victimization for more than 20 years. Her research examines interventions for youth who engage in risky behaviors, including juveniles who have sex offended, youth with HIV who engage in unprotected sex and delinquent youths who engage in risky sexual behaviors.

Next, Dr. Kristen Zgoba is the supervisor of Research and Evaluation at the Office of Policy and Planning for the New Jersey Department of Corrections. Her research involves studies of homicide offenders, sex offenders and geographic analysis. She serves on the board of directors at the American Correctional Association and is on the editorial board for Victims and Offenders, Journal of Research in Crime and Delinquency and Criminal Justice Abstracts.

Last but not least, we have Alisa Klein, who is a public policy consultant to the Association of Treatment for Sex Abusers, an international, multidisciplinary organization dedicated to preventing the sexual abuse through assessment, treatment and management of individuals who have sexually abused or at risk to abuse.

Without further ado, I will bring you Dr. Elizabeth Letourneau.

Elizabeth J. Letourneau: Several years ago, I received funding from the NIJ to look at the effects of South Carolina's sex offender registration and identification policy on several outcomes. Those outcomes include primary prevention or general deterrence, recidivism, judicial decision making, and then we've also looked at failure to register and recidivism, whether failure to register is associated with sexual or nonsexual recidivism.

I don't have enough time to talk about that last study, but all of these studies have been published, and I give you a partial citation there. I'm happy to send the publications to anyone who cares to e-mail me.

I'm going to be talking about the first three of those four topics, and 20 minutes is not a lot of time to talk about three studies then. So I'm going to cut to the chase on these and not give you a lot of detail. Again, these are all published. The full citations will be listed, and I can send you the articles if you e-mail me or they're pretty easy to find online.

And I think we're saving questions until the very end. So we have 30 full minutes for questions. So you will have time to ask your questions, we'll have time to answer them, but if you could save them to the last speaker, that would be great.

South Carolina's SORN policy is very similar to the Sex Offender Registration Notification Act, the part of the Adam Walsh Act that legislates SORN policies nationwide. So, in South Carolina, our policy is based solely on conviction offense. There's no risk assessment, which is similar to the Adam Walsh Act, and includes a very wide range of offenses. There are almost no sex offenses in South Carolina that are not registry-eligible crimes.

We require lifetime registration, which is misspelled. That's the only place where we actually are a little more severe than what is required by the Adam Walsh Act, and then we require online registration for all registrants over the age of 16 and several who are under that age as well.

Given these similarities, it is my belief that the findings that we get from our research might forecast the effects of nationwide implementation of the Adam Walsh Act.

The first study that I'm going to talk about has to do with the effects of South Carolina's SORN policy on general deterrence or primary prevention of sex crimes. So the question we asked here and tried to answer is did first-time sex crime arrests decline after SORN was implemented, and, again, there's the citation for those of you that want to read the article.

Using adult arrest data across a 15- … 16-year period of time, we modeled the effects of two intervention years, 1995, which is when South Carolina initially enacted SORN, and 1999, which is when online registration was enacted in our state.

We looked at the rates of first-time sex crimes using ARIMA analyses, and then for comparison, first-time assault and robbery crimes because, if we saw a decline in sex crimes, but the decline was the same or occurred at roughly the same time as decline in other violent crimes, then that would not be attributable to SORN, but if we see a decline in sex crimes that corresponds with 1995 and/or 1999 and we don't see that for other violent sex crimes, that supports a SORN-related hypothesis.

We had quite a few arrestees, nearly 20,000 sex crimes, over 173,000 assault crimes and 13,000 robbery crime arrests from these data, so these are population analyses for this data, South Carolina, and not samples.

What we found, which frankly was surprising to me, is that there were declines in first-time sex crime arrest rates following 1995 for sex crimes but not for the other two types of crimes. This translated to about a 10-to-11-percent decline in first-time sex crime rates or about three sex crimes averted per month following the initiation of SORN in South Carolina. We didn't see similar effects for assault and robbery, and we concluded that this supported a general deterrent effect or primary prevention effect for South Carolina's initial SORN policy. There was no similar effect for online registration, just for the original registration policy.

These results replicate those of Prescott and Rockoff 2008, an unpublished but widely circulated study on multiple states' worth of data. They also found a deterrent effect.

Sandler and Freeman, in New York, found no evidence of deterrence, and so our results are markedly different from theirs, and they used the exact same methodology, ARIMA analyses across a 20-year period of time.

We think that our differences are due to variation in state SORN policies. For example, South Carolina registers twice as many citizens as New York does. New York has a much more — has a risk-based SORN policy, and consequently, they register fewer people.

My working hypothesis right now is that when you register a lot of people for a wide variety of crimes, the message might actually get out to the general public, and I don't know if the message is “hey, this is a sex crime, you may not have known that, but now you do,” or if the message is simply “don't do this because your face is going to — you're going to be on a registry at some point.”

What we need to do, however, is compare some state policies in the same study to find out if it's really policy variation that's driving the differences in the outcomes between our studies and, for example, those of the New York folks.

Moving on to our second study, here, we looked at the relationship between South Carolina's SORN policy and recidivism, and we asked the question did SORN reduce sexual recidivism, which is its main function. It's to reduce recidivism, right? And we also looked at other forms of recidivism, and, again, there's the full reference. It's still in press, and I keep checking, and it's not out yet, but it should be soon, I'm sure.

Again, we did population analyses. We took all men convicted of one or more sex crimes between a 15-year period of time, and at some point a little over half are registered during follow-up. The mean follow-up time frame was 8.4 years with an average standard deviation of about four years. So, basically, anywhere from four to 12 years, we were following recidivism.

We looked at new sex offense charges and then separately we looked at new sex offense convictions, and we also looked at other violent offenses, any kind of person-related crimes, and nonviolent offenses.

Regression analyses. We used regular standard Cox regression analyses, and we also used some comparative regression analyses where you can include all types of outcomes, but the results did not vary. In the models, we examined several covariates, including whether the offense was a registry-eligible offense, and then some demographic characteristics of offenders that we know influence recidivism: age at start of follow-up, offender race, number of prior convictions; and we looked at whether the victim of the index or the first sex crime was a minor or not.

There were a few new sexual charges across this average 8.4-year follow-up, only 8 percent. Registration status did not significantly predict sexual charges. Several other variables did, as predicted.

There were even fewer new convictions as we would expect, only 5 percent, and if that looks low to you, that number actually corresponds very closely with other research that has been done in South Carolina that was funded by the Bureau of Justice in a follow-up study to the three-year recidivism follow-up rate of all rapists and all child molesters. They found similarly low recidivism rates, as have recent reviews of national data.

So, when you're looking at people, not people who are referred for treatment but if you are just looking at folks who have been arrested and convicted of sex crimes, their recidivism rates are low, and so that's actually not surprising. In any event, registration status was not a significant predictor of new sex crime convictions.

All of the predictors in our model predicted violent recidivism, except registration status and the same with nonviolent offense charges, and so we concluded a couple of things. Sexual recidivism was a rare event, and SORN did not influence it, and SORN did not influence any other type of recidivism either.

The folks in New York, Freeman and Sandler, Naomi Freeman and Jeffrey Sandler, also did not find that SORN influenced recidivism, they ran very similar analyses to ours.

Other researchers have found conflicting results. Duwe and Donnay, who are in Minnesota, found that the Minnesota notification process actually reduced recidivism. That process is a risk-based process that targets very high-risk offenders with the highest level of notification. They published a study in 2008, I think it was, where they found significant reductions in recidivism in relation to a comparison group of offenders who weren't subjected to notification. On the other hand, Prescott and Rockoff, who I just cited a moment ago, found increased recidivism rates for men who were subjected to notification, and so the results really vary widely, and, again, I suspect strongly that that is due to differences in state policies. But, again, there's been no research that has specifically compared different state policy characteristics to see if some characteristics are actually associated with increased recidivism, a criminogenic effect; whereas, others might be associated with decreased recidivism and others, like South Carolina's, are just plain ineffective at changing recidivism rates.

Our third study looked at the effects of SORN on judicial decision making, and, specifically, we looked at whether the likelihood of pleading from a sex crime to a nonsex crime changed after SORN was implemented and whether the likelihood of a final guilty determination changed after SORN was implemented. And, again, we looked at two different time periods, 1995 time period, 1995 to '98, which is when SORN was implemented, and 1999 to 2004 when online SORN was implemented. Again, these are population analyses with a very large number of defendants who have been charged with at least one sex crime in a 15-year period of time.

And we included other covariates in the models, including whether the arrest was for a registry-eligible offense, whether the victim was a minor, offender age, race, number of priors and whether the arrest was a no-parole offense, and that's a truth-in-sentencing designation, so if this was an offense that would get you some extra time per truth-in-sentencing laws, which were passed at almost the same time as registration laws were passed in South Carolina.

What we found is that the rate of pleading from a sex to a nonsex crime doubled over time, and so we went from about 10 percent of defendants pleading from a sex to a nonsex crime pre-SORN, so 1990 to 1994, before we had any SORN policy, to 20 percent after online registration was implemented, and those changes were all statistically significant.

Variables that increased the likelihood of pleading included registry-eligible charge and having minority race, and then if you had a minor victim or you were older at the time of your charge, you were less likely to be permitted to plead to a nonsex offense. But so here we see that SORN has actually doubled, responsible, I think, for doubling the rate of pleading. We did not see similar increases in pleading for other crime types.

There are approximately 1,100 sex crime charges against adults per year in South Carolina, and that means we went from allowing 110 of those to plead to allowing 220 of those to plead. That's a lot of people who are being permitted to plead to a nonsex crime, and when they plead, pleas are almost universally associated with guilty determinations, of course, so you're being found guilty of aggravated assault and battery. Almost all the offenses are pled to that particular crime, which means you will not get sex offender treatment, and you will avoid registration notification, which is the whole point of pleading in the first place or it's a big part of it. So these folks are effectively opting out of any kind of treatment that might have been offered to them.

Results for guilty dispositions were also interesting. Sixty percent of all sex charge cases resulted in guilty dispositions, and the likelihood of getting a guilty finding increased after SORN was initially implemented and then declined after online notification was implemented, and when we removed the pleaded cases, which you will remember increased over time and they're almost all associated with guilty determinations, when you take those out of the equation and you just look at people who were charged with a sex crime and adjudicated for a sex crime, the likelihood of them getting a guilty determination declined after online registration went into effect to below pre-SORN levels.

So we started out at about 48 percent of these individuals being found guilty for a sex crime that increased for a brief period of time after SORN was initially implemented, and then it declined below the pre-SORN levels to 45 percent. And although that doesn't seem like a big decline, 48 percent versus 45, it is statistically significant, and it indicates that after online notification, we actually had more people being acquitted, so we have more people pleading, a significant increase in pleading to nonsex offenses. If you don't plead, you're actually now much more likely to simply be acquitted outright.

In combination, the results suggest that South Carolina's policy did achieve some modest effects in deterring new sex crimes, as I mentioned earlier, about three per month, but you have to look at that positive effect in light of the entire effect of this policy, I think. There's no effect on recidivism, which is what it was designed to reduce in the first place, and there are real significant changes in how these sex crime cases are handled. Many more are allowed to plead to nonsex crimes, and then significantly more now are simply not found guilty at all.

We did not find similar changes, by the way, for other types of violent crimes, and so we did do our comparison analyses.

As I mentioned, sex offenders convicted of nonsex crimes are going to not receive specialized treatment. They avoid registration and the notification altogether. Exonerated sex offenders are going to avoid any kind of judicial consequences, and, potentially, the effects of that are that they may not view their actions as having been particularly detrimental. We also know that victims are hurt when offenders are not held accountable. There's a long history of research in that area.

South Carolina's public registration policy relies entirely on the conviction offense. There's no risk assessment. It targets a very broad category of offenders. There are people who are on our lifetime online registry for voyeurism and for indecent exposure, as well as for much more serious sex crimes, but because our policy is so broad, it overwhelmingly targets low-risk offenders. But I think that the results from our research suggest two policy modifications.

I think there's a strong argument to be made for reserving registration and particularly for reserving notification for high-risk offenders. These are costly interventions. We're starting to look into the costs. My colleagues, actually all the colleagues here, Kristen and Alisa and some others are starting to look into the costs of registration and notification, and it's really quite astronomical, not to mention when you have so many offenders on the registry, it, I think, does the community a disservice.

I'm not sure what you would do to protect yourself from someone who is an indecent exposure versus someone who is a child molester with repeated victims. When you treat those two people the same, I think it makes it difficult for the community to use these tools in any meaningful way.

I think online notification, in particular when there is no discretion, there's no judicial discretion, there's no risk assessment, we see some very significant changes in case processing, and these changes were completely unanticipated. Nobody expected more sex offenders to be let go, and nobody expected more sex offenders to be permitted to plead to nonsex offense crimes.

Those are some effects of South Carolina's policy, and I suspect those will be the effects of the Adam Walsh Act when it's implemented nationwide because the two policies share so many characteristics, and so I think the second policy recommendation is that online notification should be curtailed significantly, if not eliminated, but certainly curtailed.

In several cases, our results conflicted with other researchers, and as I've said, I think that policy characteristics probably account for those changes.

New York registers relatively few offenders compared to South Carolina. That might account for their differences in general deterrence. They found no effect; we did find a general deterrent effect, but both states found no evidence of reductions of recidivism.

On the other hand, Minnesota, which has a different tiered and risk-based policy, did find reductions in recidivism. Prescott and Rockoff, using national data, found increases in recidivism.

We really have to move beyond single-state studies, and this is a note that I have been sounding for the last couple of years. We have got to have some comparison studies that take data from several states, not just aggregate data. You really need individual-level data, so you can identify offender characteristics and specific offender recidivism and see how those are influenced by policy characteristics, such as whether registration is conviction-based or risk-based, whether duration is 10 years or life, whether notification involves online notification or some much more targeted notification or no notification. I think those are the three main areas of variation across state policies that probably are having influences on these outcomes, in particular the outcomes in judicial decision making.

I believe that when judicial actors perceive that they have judicial decision, that they have some control over what happens, that you wouldn't see the increases in plea bargains or the reductions in guilty determinations that you see in our state when judicial actors have zero control over the process.

And that's the end of my talk. Thank you very much for your time.

[Applause.]

Kristen M. Zgoba: Good afternoon, everyone. I'm here to talk to you today a little bit about NIJ's funding of a study that came out of New Jersey. As I'm sure you're all familiar with, the birthplace of Megan's Law was in New Jersey, and we thought given that fact, that coincided with the 10-year mark of the implementation of Megan's Law, that it had been a good time to really apply for some funding that would track some long-term outcome of what we were seeing in that state. And when we started to look at the literature, we saw that presently all 50 states had some version of SORN, some version of Megan's Law, as well as the District of Columbia and many countries abroad.

So we started to ask ourselves, well, all of these states have these laws, what really is the effect, what's happening, what can we see as the likely outcome of what we've implemented. So we realized that despite this widespread support for these registration and notification laws, there was little evidence, if any, that supported any type of reduction in first-time offenses, which we saw through protective measures or general deterrence, or any literature on sexual re-offense reductions, which would be through protective measures and specific deterrence.

So, to provide you a little background of the study, the National Institute of Justice awarded myself, as well as a number of colleagues, this grant in 2006, and what we're looking to do within this grant that has since ended was to compare a potential pattern of sexual offending rate specific to New Jersey, so before and after the implementation of Megan's Law.

What I should also note here is that within the context and within the interest of time, I'm presenting to you two phases of this study. There are actually three phases to this NIJ study, the last phase being a cost-effectiveness analysis of Megan's Law specific to New Jersey, which, again, in the interest of time I'm leaving out, but, as Elizabeth has mentioned, Alisa, Elizabeth and I are working on an analysis of that.

Also, we've been able from the results of this study to parlay a number of different analyses specific to sex offender registration and notification, particularly whether or not sex offending characteristics and predictors of recidivism have changed over time. So we were able to have this 10-year time span of sex offending recidivism information, and what we were able to also do outside the scope of this grant was to analyze whether or not those risk predictors of recidivism changed from when we started to see Megan's Law implementation and risk assessment instruments being utilized to where we are now because, as we know, our risk assessments haven't changed all that much. OK.

We were happy to see that risk predictors of sexual recidivism remained stable over the 10-year time frame. So they haven't changed a whole heck of a lot from what we saw from our research.

So there are two different phases to this research, and I'm presenting you the methodology on both of them. The first was a simple pre- and post-research design that determined whether or not there were significant changes in the rates of offenses reported by law enforcement agencies prior to the implementation of Megan's Law and after the implementation.

We looked at rates for sexually based offenses, nonsexually based offenses and drug-based offenses from 1985 through 2005. We utilized data from all 21 counties in New Jersey as well as the aggregate form of New Jersey as a whole, utilizing, again, Uniform Crime Report data.

What we did was look at the prevalence rates for re-arrest, re-conviction and re-incarceration over the time frame, and we compared population estimates from the Department of Labor's Bureau of Labor Statistics. These population estimates from New Jersey were cross-referenced with the Sourcebook of Criminal Justice Statistics, and there was no significant difference between the population estimates between the sources. So we utilized UCR data for the data that was significant, obviously, for the study.

Again, as we all know, one of the limitations that I'll discuss in the last section are the limitations of utilizing official data and arrest report.

We also thought it was necessary to not only take sex offending data but also to look at nonsexual offending patterns over this 20-year time frame and also drug offense patterns. The reason that we chose drug offense patterns was because we've seen, specific to New Jersey, very similar legislative pathways. So we thought it would be interesting within these time frames if we could take all of this information and sort of fit it into the context.

Phase two of this grant had a very, very different methodology. We utilized a sample of sex offenders that were released from the New Jersey Department of Corrections, so all of the offenders in phase two of the study were convicted sexual offenders. They were released either from our adult diagnostic and treatment center or the general population.

In New Jersey, we are one of only two states in the country that have a prison facility specifically devoted for the treatment of sexual offenders. Those sex offenders are incarcerated just like they are in general population, with the exception that they receive treatment specific to sex offending. OK.

So we utilized 50 sex offenders per year, 25 from the ADTC and 25 from the general population. We randomly selected them for these 11 years, from 1990 through 2000. This yielded us a sample of 550 cases of sex offenders. Within these 550 cases, we collected quite an assortment of data on each individual offender, approximately 100 variables on each offender.

They ranged from historic information to demographic information, sentencing and re-offense patterns for each individual sex offender in the sample.

Phase two analyzed these pre- and post-group differences on three specific outcomes that we were interested in.

The first was the most obvious: the reduction of recidivism. We thought it would be in our best interest to utilize a sort of triangulated methodology here and look at re-arrest, re-conviction and re-incarceration.

We were also very interested in increased community tenure. What we meant by increased community tenure here was the number of days it took to the first arrest for general offending and the number of days it took to arrest for sexual offending.

The reason that's important is because whether or not you see a statistically significant reduction in recidivism is, of course, the hallmark of your outcome study, but if an offender is also staying in the community at a longer time span prior to recidivism, that, in some ways, is still a very significant outcome because we start to see cost benefits, a number of different benefits that can be gauged from that.

We also looked at reduced harm. Within the context of reduced harm, we were interested in whether or not the obvious happened. Are there fewer sex offenses occurring after the implementation of Megan's Law?

Within that, we were also interested whether or not offenders in general were less violent, OK, so whether or not this sample of sex offenders after the implementation of Megan's Law, whether or not they just happened to be less violent offenders and whether or not we saw offenders who had fewer child victims, because we know through the majority of research, Megan's Law is going to focus primarily on those individuals that have those child victims. So it was important for us to really gauge whether or not we were seeing a reduction there.

The first results you see here are for phase one, and the next few slides, I'll point out to you, were all for phase one. The rates varied here, so you'll see on the bottom portion, the X axis of your scale here starts you off at 1985 and then brings you up to 2005. So what you see here at 1994 is the point of implementation for Megan's Law. The rates varied from a high of 51 offenses in 1986 to a low of 29 offenses per 100,000 individuals in the population in 2005. In general, we saw a consistent downward turn of sexual offending rates over this 20-year pattern.

What you are looking at here is the rates before and after the implementation of Megan's Law. The upper line represents sex offenses through the years 1985 to 1994, and the lower line represents sex offenses for the year 1995 to 2005. Superimposed over these data points is a linear trend analysis that will ultimately provide us the information to see whether or not the slope has increased or decreased over time.

The first thing that should be noted from what you're looking at here is that beginning in 1995, which is your bottom line, the rate of sex offenses have never again approached the pre-1994 levels. The intercept and the average of these two lines are different.

And the second important thing to take home is that the slope is deeper in the post-Megan's Law period. This is particularly notable for us, for the facts that these are sex crimes, and we all know that sex crimes have a low base rate of offending, so one wouldn't necessarily expect to see a steeper dropoff of these rates.

Now, when we do the same with our nonsex offenses, just to sort of get it into context, we see that the average number of crime has been changed to per 1,000 population, and the average is 50 crimes per 1,000 individuals and the highest is 56 in 1989. We see here at 1994, again, is the point at which Megan's Law was enacted. The lowest rate we see here is in 2003 at 45 nonsex crimes per 1,000.

For the last five years of these data, we ultimately see that the crime rate has remained stable. There is a significant change point, however, if you look four years past the 1994 at 1998. For these data analyses, we utilize what's known as a Mann-Whitney U test, which rank-orders all of your data and is particularly useful for individuals when you don't know where your data points change. OK. So, within the nonsex offending rates, we saw a significant data change point at 1998.

These are what our statewide drug offenses looked like for the same time frame. Rates varied from a high of 89 in 1989 to a low of 52 in '85. Again, so we can see that in 1989, our drug rates spiked, there is no significant change point with our drug offense rates during that same time frame.

This moves us over to our recidivism information for phase two of the sex offending study. What this does is present pre and post contrasts that are controlled by time at risk. So these are year-by-year graphs that analyze our different data points for arrests, convictions and incarcerations for the 10-year time frame, so this is phase two.

OK. So, going from 1990 to 2000, five years prior to Megan's Law, five years post-Megan's Law, and, again, it represents the rate of sex offenders released during these time frames. Overall, 46 percent of offenders were re-arrested, 41 percent were convicted and 35 percent were re-incarcerated, and, again, this is hard to gauge simply from this, and you'll see this in the next table that I'm about to show you. This is for general recidivism by year, so this is not broken down for sexual or for drug offenses.

This is what became particularly important for us with our sex offending study, and this represents the outcomes pre- and post-Megan's Law. You'll see that at the top of the table, we have recidivism information. You see it broken down by pre and post total, utilizing either an ANOVA or a chi-square.

Of particular interest are the top three. This weighs general recidivism. So, within general recidivism, we took sexual recidivism out. OK. So this is just general types of recidivism here. We did see a significant pre/post effect on general recidivism after Megan's Law. So, on all three categories post-Megan's Law, we do see a significant reduction for general recidivism.

When we look, then, at community tenure, community tenure, as I explained to you, was days to arrest for any crime. So, for our general recidivism or days to arrest for our sex crimes, we see no significant difference within these two categories. So sex offenders are spending approximately 772 days in the community prior to Megan's Law before they commit any crime, and 726 days in the community prior to committing any offense after Megan's Law. OK. We see slightly higher time frames pre-Megan's Law for sex crimes, and then post, 765. This didn't achieve any sort of statistical significance.

When we look at harm, this is what, again, is the primary interest particularly for us because we know that our Megan's Law study is supposed to have an effect on sex crimes. So we show here the percent re-arrested for any type of sex crime. Prior to Megan's Law, 10 percent of the sample was re-arrested for any type of sex crime. After Megan's Law was implemented, 7.6 percent of the population had been arrested for any type of sex crime. Again, no statistical significance. When we looked at re-conviction and re-incarceration, we also did not see any difference.

Also, of importance was the type of sex crime. As I explained to you in the first few charts, we were interested in whether or not the type of offenses that sex offenders were committing actually changed over time, and, again, we didn't see any change with that. We didn't see that sex offenders were committing crimes against a smaller amount of children or larger amount of women, adult women, anything like that. The only place that we saw a reduction was in the percent of violent offenses. We saw that post-Megan's Law implementation, sex offenders were generally less violent offenders, OK, and that reached a statistically significant level.

Our survival functions, you'll see before you, considered are censor cases or time at risk here. Our offenders in our previous analyses were capped, so all offenders had the same amounts of time at risk, which ultimately became six and a half years, and you'll see from the graph that there's very little difference between the offenders pre- and post-Megan's Law implementation in terms of those staying within the community. This, of course, did not reach a statistically significant level either.

So the results tell us that New Jersey has experienced a downward trend in sexual offense rates with the change point at 1994, and this became a very important issue for us because, ultimately, what was happening was phase one of our study was indicating that sex crime rates were down, that they were down from the 20-year mark, and they were consistently going down, OK, which led many individuals in the community to surmise, “Megan's Law is having an effect,” because your sex crimes are down. OK.

We then, when we moved on to phase two, did not see that same result. When we took our aggregate data and sort of de-aggregated it, we no longer saw that effect over time. OK.

The majority of the counties, sex offense rates were higher prior to 1994 and were lower after 1995, which is ultimately when we know Megan's Law came into effect, but we couldn't in any way point to the fact that Megan's Law was having that effect.

While sex offenses have rebounded somewhat, they are far lower than they were in the '80s and early 1990s. In terms of sex offenses, the general decline is similar to that of nonsex crimes; however, the statewide change point for sex offenses occurred during the time frame that we would have expected it to, around 1994, while our nonsex crime rate changed in 1998.

The wide year-to-year fluctuations in drug crimes were, we assume, a reflection of our drug policies and our practice efforts, although those efforts have not been sustained over time, and in the case of sex offenses, we know the statewide change occurred when it was predicted to change, and we have seen that it's maintained that power, primarily the statistical power, over time.

Again, when we look at the pre- and post-Megan's Law analysis, though, we see that Megan's Law was not effective in increasing community tenure, so they're not spending longer amounts of time in the community. It has not reduced the number of sex crimes. It has not changed the type of sexual re-offense or first time sexual offense, and it has not reduced the actual number of victims involved with sexual offending.

Before I move on to the limitations very shortly, just to say again this is something that we struggled with in New Jersey and had to be very careful with how we worded it in our final report to the Department of Justice because we knew that this information could easily be misconstrued particularly within the media and the legislature saying that Megan's Law was, in fact, effective. And we knew primarily that was coming from individuals that really didn't have a strong concept of the statistical analyses when, in fact, this actually did happen, particularly with one of the state senators who drafted Megan's Law ultimately came out saying that, “Look, Megan's Law is effective, you know. Bottom line, are sex crime rates lower after Megan's Law was implemented, or were they higher?” you know, and your obvious answer is, “Well, they're lower,” and he said, “Well, you see, it's effective, then.” So, again, we had to be very, very particular in how we drafted these results.

The limitations for this study are those limitations that we would see in most research, including official reports, particularly that of utilizing UCR data. We know about the underreporting, and we know about the low base rate of reporting.

Particularly with sex offending, in all of our studies, it's become very difficult for us to achieve statistical power because we simply don't have enough sex offenders committing re-offenses of a sexual nature, and that is in no way advocating that sex offenders have to commit more sexual re-offenses for us, but what it does is make our jobs harder because of obvious statistical power. OK.

Thank you very much.

[Applause.]

Alisa Klein: Hi, everybody. Thank you for being here. This is a topic that isn't so pleasant to listen to some of the time.

So I'm curious just to know, if you could just raise your hands and let me know how many of you are researchers? OK. And how many of you do policy-related work or make policy, influence policy? OK.

So I'm going to take this discussion in a little bit of a different direction and look at some of the kind of policy questions that come up for me, and I'm going to pose a lot of questions more than I'm going to offer answers because these are the kinds of questions that I'm constantly grappling with and kind of keep me up at night, and I'm going to share some of them with you and also see if we can get a little bit of feedback from the other presenters about their thinking about how their research can translate to sound policy.

I actually want to pose a question and just hear briefly from each of you, Kristen and Elizabeth — how have your studies been received? How has your state, how has the federal government responded, because that's really going to lead us, I think, down a path of kind of talking about what we can do in terms of policy.

Elizabeth Letourneau: So that's easy for me to answer. I don't think my research has had any effect. It certainly has not in my state, and I don't think it has at a larger level, and part of that, I think, is because I haven't effectively targeted my findings to policymakers in my state. It's actually a question I was asked in Australia recently when I was over there presenting, and they said, “Well, you know, what have you done for the U.S.?” And the answer was nothing in terms of the effects that I'm having.

So, yeah, I haven't figured out how to translate. I'm so delighted to see so many hands raised when Alisa asked how many policymakers, policy analysts were in the audience, because that's probably the largest number that has heard anything that I've had anything to do with.

Kristen Zgoba: It's been a very interesting situation in New Jersey. The federal government was very supportive of our findings, certainly promoted the findings. The state, unfortunately, was not as supportive, if I can say that.

[Laughter.]

Zgoba: We began reaching out to the governor's office early on when we saw that our research was indicating that there was no reduction in sex crimes consistent with Megan's Law implementation, and, ultimately, we got no response.

We got no response for many months until our final warning that our deadline was fast approaching to NIJ, and that the findings would become public. And when that occurred, we were told we could give interviews, and the Associated Press was the first interview at which point it was then picked up by many, many newspaper articles. And it just kind of became a firestorm in the state, and, at the same time, everything sort of exploded, and Maureen Kanka came out with a statement and state senators came out with statements, and then the governor at that time immediately issued a statement within days saying that Megan's Law would never be changed and that no study would have an effect on changing the policy.

I can't imagine that given the sort of political background of our new governor that we'll see much of a change consistent with that either, so, unfortunately, that's the case.

Klein: Also thinking about it in terms of timing, the Adam Walsh Act with the Title I, which is the Sex Offender Registration and Notification Act that is mandating to all 50 states, 197 Indian tribes and the five U.S. territories, how to create a uniform system for public notification, and, as Elizabeth was discussing, how to tier offenders based on offense, not based on assessed risk and so forth.

So that was passed in 2006. We're coming up on a deadline, the final deadline of July 2011 for the implementation of SORNA. In the meantime, these studies have come out, several other studies have come out, and I've just been thinking about how this kind of information does either get kind of sidelined or ignored because we now have a major federal project on our hands of implementing SORNA, which essentially is going against what the data tells us is, in fact, effective.

And we see this around sex offender management policy community, management of sex offenders, that kind of policy, we see this all the time, that policymakers essentially often find themselves in a dilemma. We're seeing more and more research that's showing that a lot of the things that we are turning into policy, that we're seeing being turned into policy, are not effective, residence restrictions being the other significant way that sex offenders are managed when they return to communities.

So, you know, we have to think about what is compelling — not sex offenders — what is compelling policymakers to legislate in these ways when data is telling them otherwise.

So one of the things we do know — and this is something for all of you researchers to think about certainly, but also the policymakers here — we do know from a lot of research, ironically enough, that policymakers don't make decisions based — I mean, I'm saying something fairly obvious, I guess — based on research, the decisions that are made about laws that are implemented, especially around things that are very emotional, very painful for communities, individuals, families to grapple with. They're not making decisions necessarily based on what research tells us.

So one of the things we need to be thinking about is how can research play a more influential role in influencing policy, and how can we provide policymakers — this is the slightly more complex piece of this — how can we provide policymakers with the kind of cover that they need to make the decisions that might not be perceived as popular when it comes to sex offender-related policy.

We know that sex crime issues have little to do for people with facts and research, so we're talking about kind of communities. Common sense says if you want to protect kids from predators, you need to know where they are, thus community notification, thus residence restrictions. If we just move them away from where children congregate, we can protect them; it seems to make sense. And the terms that are used, you know, don't sound so bad. And so one of the things that policymakers need to be thinking about is how do you change or reframe those terms and the concepts behind them.

And then another very interesting piece of this community notification discussion is, again, what the public thinks about it. The Washington State Institute for Public Policy did some interesting research in 1998. They did — just in Washington, in the state of Washington, they asked people about their perceptions of community notification. And what they found was that almost 75 percent of respondents reported that they learned more about sex offenders, sex offenses, how sex offenders operate because of community notification; their awareness was raised, and over 60 percent said that community notification makes released sex offenders behave better. So they had a sense that it actually would influence sex offenders to feel like they're being watched. People know who they are, so they are going to be less likely to re-offend.

And the vast majority — and I can't remember what the exact percentage is; I think it was in the high 80s — felt safer knowing about convicted sex offenders living in their communities, and they indicated that they were more safety conscious, they took precautions, that kind of thing.

The question here is, you know, are policymakers making these decisions based on what they think the public wants to hear; if so, how do we educate the public properly, and then how do we get the policies to shift based on research, of course.

And then one of the things that just has come up for me in listening to these two folks or reading their material is, you know, all of the policies historically since the early '90s that have legislated these registration notification laws have been named after three different children, nationally known situations of extreme, deep pain of children who have been abducted, have never been recovered. In one case, we know, Megan Kanka was, in fact, raped before she was killed. Adam Walsh was never found; we don't know what happened to him. Jacob Wetterling was never found. These are the three children upon which federal policy has been set around community notification.

And so I just pose this question, you know, does this make it psychologically difficult for anyone to oppose or question these policies because somehow it might be detracting from the pain, the very real pain that we all feel around what has happened to these particular kids, is it an affront to the families, the bereaved families.

I also am struck by something that isn't talked about quite as much when we talk about notification laws, the sex offender registration and notification laws, and that is this difference between reportage and incidence.

So one of the things that some of us that do this work have a sense of is that the policies that we're increasingly legislating around sex offenders returning to communities are policies that may be having a deep influence on the decision to report, to bring to light situations of child sexual abuse and adult sexual assault.

We know that the vast majority of sexual violence is perpetrated within intimate circles, communities. With child sexual abuse, 93 percent is perpetrated by someone known to the victim. About — what is it? — 35 percent is actual family members of the victim, and with adult sexual assaults, it's about 64 percent of women we know have been raped, physically assaulted or stalked by an intimate partner.

So we know that this is a crime that is perpetrated within intimate circles, and the reportage rates are kind of in keeping with those statistics. We know that only about 12 percent of child sexual abuse is ever reported to the authorities, only about 16 percent of rapes, 20 percent is what Tjaden and Thoennes tell us. So, with sexual violence happening in these intimate circles, are we somehow, with these laws, keeping people from feeling like they want to bring forward these members of their families, their communities to — are we somehow keeping situations underground?

Let's see. And the last thing that I wanted to bring up is something that these two studies that we heard about don't focus on, but that is particularly significant, are the unintended consequences that are associated with community public notification on sex offenders.

We know that sex offenders returning to communities who are publicly notified upon suffer from loss of jobs and unemployment, employment instability. They suffer often from harassment and physical assault. They have chronic difficulties finding places to live, finding jobs, and they are frequently forced to the outskirts of communities into increasingly rural areas where they're not going to be able to access the kind of specialized supervision and services that they may need to not re-offend.

These kinds of stressors on sex offenders' lives are shown to actually raise their risk for recidivating. We know from the general criminogenic literature and we know from the sex offender-specific research that when sex offenders have stable jobs, housing, social bonds to the community in which they live and they are able to continue their family relationships, they are going to be less likely to re-offend.

So, certainly for policymakers, we need to be thinking about those kinds of questions, you know, what are the unintended consequences, what are the collateral consequences of the decisions we're making, especially when they're not based on the research that we're seeing.

So that is what I wanted to share, and I know that we're all open for questions now.

Karen Bachar: Please join me in thanking the panel.

[Applause.]