What is a Valid Evaluation or Study of Recidivism

Over the years a lot of studies and evaluations have been done on everything from the safety of automobiles, airplanes and ocean liners to the type of people that buy fast food rather than cook at home. So what truly is a valid study and what Is required for it to meet scientific standards? Please understand I am not talking about pseudo scientific standards but about mathematical statistical analysis standards or true scientific standards, the type of standards that are required within a court of law.

When people are on the sex offender registry and when justifying the reason for the registry itself is based on the belief that registrants have a high propensity to reoffend in a similar crime i.e. a new sex crime, then naturally the study must look at the fact that they are involved in a new sex crime. The study must also recognize that rearrest does not mean that a person actually committed another sexual crime, only that they are suspected of having committed another sexual crime. After all in this country you’re innocent until proven guilty in a court of law so rearrests should never come into play.

In most studies that we have looked at, we have found that the rearrest rate and the re-conviction rate are vastly different, statistically usually 25% to 50% of the people that are arrested are not re-convicted.  If there was no crime committed by these people then only being arrested for one doesn’t count. So the first requirement of a ligitimate study on sex offender recidivisim is that only re-conviction rates for a new sex crime can be considered accurate.

Next if the study is to evaluate the re-offense rate of people who have been previously convicted of a sex crime and have been placed on the registry because of their supposed dangers they present, then naturally all of the people that are on the registry must be evaluated and not just subgroups such as people coming out of treatment programs or people being released from prisons, mental hospitals or other specific subgroups such as rapists or sexual assaults. The laws and rules and regulations affect everyone that is on the registry no matter what crime they were convicted of, therefore the only effective and accurate studies will be those where the entire registry is looked at and not just studies that are using one or more specific subgroups as their sample base. Although some studies of individual subgroups may help in making decisions about members of those groups, and over the past decade there have been some subgroups studies that have shown extremely low re-offense rates, these types of studies still are not truly an accurate depiction of all of the people that are affected by the sex offender registry laws.

Studies with only small samples, usually involve subgroups and as it has been stated above, those are not valid studies for determining the re-offense rate for all of the people that are affected by the sex offender registration laws. The only acceptable study would be one that includes the entire registry within a state, if for example one truly wanted to discover the re-offense rate within that state and see if that rate justifies laws and rules and regulations within that state.

Nearly all states have within their preamble of their sex offender laws, a justification of the laws, that state that the reason for their registry laws is for the public safety because of high re-offense rates of the registrants. So naturally if there is no high re-offense rate, than the justification for the laws evaporate. Of even more value would be if a true study were to be done that would include all registrants within our country to discover what the true yearly recidivism rate is of new sexual convictions for all registrants on the National Registry.

There are of course a few mega studies out there that look and evaluate information gathered in multiple smaller studies. Many of the studies within the mega studies are questionable in fact some mega studies have used information from studies that have been proven to be completely debunked.  Before those mega studies can even begin to be considered accurate or concise, each one of the included smaller studies must be looked at separately to see if they fall into the proper guidelines set forth to represent the re-offense rate for everyone on the registry. If any one or more of those included studies fall short then the entire mega studie’s conclusion must be considered invalid.

To give you an example, if you were to do a comparison between the safety of driving the car to a destination or taking an airplane to the same destination, you would not subdivide those groups such as driving a specific type of vehicle or taking a specific type of airline you would include all of automotive transportation and all of the air transportation to get an overall picture of which is safer. You can easily see that when you start subdividing for the make or type of vehicle or a specific airline that the numbers could easily be skewed in favor of one or the other general forms of transportation. This is exactly what has happened when the pseudo-scientists that do studies to evaluate the re-offense rates for people on the registry, are not evaluating people on the registry but are only evaluating a specific group or subgroup. Often these same pseudo-scientists have a fiduciary interest in keeping the myth of high recidivisim alive or they fall into the adversarial alliance category because of who is paying to have the studies done. These factors must be taken into account when evaluating a study. it is common knowledge that a study can be skewed simply by the questions that are asked and how they are asked. Another thing to watch for is mixing hard numbers and percentages.

Now that we have laid out the groundwork; that only re-convictions are accepted and only studies that evaluate the entire registry are accepted, the next issue is the length of time that the study must be done for it’s statistically results to be correct.

Considering the fluidity of the registry one-year periods are the only acceptable answer. Those that choose to do studies for longer periods three, five, or ten years, must take into account the statistical anomaly that if you add up one side of the equation you must also equally add up the other side of the equation. Simply put, if you add the numbers of re-offense each year over a period of time you must also add up the  numbers of those that did not re-offend. example 10 people re-offended the first year and 11 people re-offended the second year, the first year there were 750,000 people on the registry the second year there were 810,000 people on the registry. Most of the studies would say 21 people out of 750,000. When in actuality comparing people that do reoffend to those that do not it would be 21 people out of 1,560,000 to arrive at the reoffend rate percentage over the two year time..  Some people who do not understand statistical analysis, might say this is not fair because you’re adding up the people that do not reoffend over and over again but to be statistically correct what happens to one side of the equation must also happened to the other side. Also, as I mentioned before, the fluidity of the number of people on the registry varies, because there are people being removed, or moving out of state, or dying. As well as large numbers of new people being added at a high rate each year. An example the Nebraska Registry increased in size by a rate of 776% over a twenty year period. So it appears that the only accurate way to judge the re-offense rate is on a yearly basis. This is the same way of evaluating that is used with all other criminal classes such as DUI or DWI.  These studies are done on a yearly basis and have a re-offense rate of over 50% yearly.  It would make no since if they were to say that there was a 250% reoffense rate over a five year period.

To my knowledge there are only two evaluations that come close to meet the criteria above where all people on the registry’s re-offense rate in the new sexual crime for the entire registry and are broken down into yearly increments have been met. The first one is the Nebraska registry study that looked at the people on the registry in Nebraska for their re-offense rate not only going back to its inception in 1996 but also evaluating people that had crimes that happened before the registry exist going back to 1971 they came up with the yearly re-offense rate average of 6/10 of 1%. The second study was one the Ohio study that looked at 21,000 people that were all dumped onto the registry at the same time and then tracked them for ten years. That study came up with an average reoffense rate of 3/10 of 1% yearly. I don’t believe that anyone would call those numbers frighteningly high.

At the present time I understand that there is a an attempt by Paul Dubbeling, a North Carolina attorney, who recently won the Doe v Cooper case in the NC 4th district circuit court, challenging NC premise for their sex offender laws. He used the state’s own data to show how extremely low the true re-offense rate was. He also stated that the method they used to gather and process the 40,000 data points (registrants) over the last 20 years of NC’s own state registry. His data on and how he did this will be shared with anyone to gather the true recividism rates in their respective states once he gets everything verified through a statistician at Duke University. It will be interesting to see what that percentage is when his research is done. You can see a further explanation of how he gathered his data and the problems of presenting this information to the court’s, in the lecture that he gave at the national RSOL conference this past year.


The bottom line of all of this is that if you want to make laws that affect a group of people especially a disfavored group of people. and you want to point to studies to justify these laws than the studies had better include every member of that disfavored group not just some sub categories of it. When laws are passed to add punishment or restrictions to a disfavored group they are in violation of our Republics Constitution. they are in fact bills of attainder or lesser pains and punishments prohibited by our Republics Constitution. Legislators do not have the ability to create new laws that add punishments to a persons or group prior crimes. Punishments can only come from the judicial branch in each individual case and with the due process of law.

3 comments for “What is a Valid Evaluation or Study of Recidivism

  1. Tim Lawver
    September 16, 2017 at 2:19 pm

    The bottom line is they did it anyway. The Whetterling act was on its face an ex post fact law. The people applauded! However, it cost them way more than they realised. Liberty is fragile and easily squashed by tyrants. I will go to my grave knowing that the very instant they sold out the child molester the American people sold out themselves and their own children’s future to be free of tyrants.

  2. caring
    September 17, 2017 at 8:37 am

    Great article Will. Something I always wanted done, but I don’t really have the means. If the re-offense rate is the main reason for justifying the registry, then there needs to be a study to show the actual and not the perceived re-offense rate. Find ALL sex offense cases and then compare them to previous offenses by that very person. You will find that very few, if any, will have previous offenses. This should be a fairly easy statistic to prove I would think if there were more people willing to invest the time.

  3. September 18, 2017 at 6:27 am

    Since this article mentions Duke University, let’s take a look at what Duke Medical Professor Allen Frances says about the newest type of RSO “evaluation” to become popular: the Static 99, and its newest “revised” version (Static 99R). Dr. Frances and other highly esteemed colleagues have discredited the Static 99 tests:


    While states like Virginia and New York have dismissed the Static 99 and 99R as what seems to be junk pseudo science, states like California have doubled-down on the Static tests. In the case of California, the state relabeled and repackaged the same exact Static 99R test as “SARATSO.”

    Here is Virginia discrediting the Static 99R:


    Here is New York discrediting the Static and Karl Hanson, it’s “developer:”


    What’s unbelievable is that the Static 99R test claims to predict recidivism from 10 questions. The score remains the same throughout one’s lifetime — with NO adjustment made for *current* age and/or even offense-free years while living in the community (both of which, many studies suggests, are most important determinants in predicting real recidivism; yet the Static’s score does not take either/or into consideration).

    Static 99 = real junk science

Leave a Reply

Your email address will not be published. Required fields are marked *