eferred to as “out-<strong>of</strong>-State” arrests).Data on arrests came <strong>from</strong> State RAPsheets and FBI RAP sheets. RAPsheets (Records <strong>of</strong> Arrest and Prosecution)are law enforcement records<strong>in</strong>tended to document a person’s entireadult crim<strong>in</strong>al history, <strong>in</strong>clud<strong>in</strong>g everyarrest, prosecution and adjudication fora felony or serious misdemeanor<strong>of</strong>fense. Arrests, prosecutions andadjudications for m<strong>in</strong>or traffic <strong>of</strong>fenses,public drunkenness, and other pettycrimes are not as fully recorded asthose for serious crimes. The “percentrearrested” is calculated by divid<strong>in</strong>g thenumber rearrested by the numberreleased <strong>from</strong> prison <strong>in</strong> <strong>1994</strong>.All measures <strong>of</strong> recidivism based oncrim<strong>in</strong>al records are subject to twotypes <strong>of</strong> errors. Type 1 errors arisewhen the arrest or the conviction <strong>in</strong> thereleased prisoner’s record is for acrime that person did not commit.Type 2 errors arise when the releasedprisoner commits a crime but he is notarrested for it, or, even if he is, thearrest does not result <strong>in</strong> his conviction.Some amount <strong>of</strong> type 1 and type 2error is <strong>in</strong>evitable, however recidivismis measured. But that does not meanthat all recidivism measures are equallysuitable, no matter the purpose theyare <strong>in</strong>tended to serve. The ma<strong>in</strong>purpose <strong>of</strong> this recidivism study was todocument the percentage <strong>of</strong> sex<strong>of</strong>fenders who cont<strong>in</strong>ued their <strong>in</strong>volvement<strong>in</strong> various types <strong>of</strong> crime aftertheir release <strong>from</strong> prison <strong>in</strong> <strong>1994</strong>. Themore suitable measure for that is theone with the fewest type 2 errors: theone, <strong>in</strong> other words, less prone tosay<strong>in</strong>g someone is not committ<strong>in</strong>gcrimes when he actually is. Betweenrearrest and reconviction as the recidivismmeasure, the one less likely tomake that type <strong>of</strong> error is rearrest.One reason is that the rigorousstandard used to convict someone —“pro<strong>of</strong> beyond a reasonable doubt” —makes it certa<strong>in</strong> that guilty persons willsometimes go free. Another reason isrecord keep<strong>in</strong>g: the justice systemdoes better at record<strong>in</strong>g arrests thanconvictions <strong>in</strong> RAP sheets. For suchreasons, this study uses rearrest more<strong>of</strong>ten than reconviction as the measure<strong>of</strong> recidivism.Rearrest forms a conservative measure<strong>of</strong> re<strong>of</strong>fend<strong>in</strong>g because manycrimes do not result <strong>in</strong> arrest. Not alltypes <strong>of</strong> crime are alike <strong>in</strong> this regard.Crimes committed <strong>in</strong> nonpublic places(such as <strong>in</strong> the victim’s home) by onefamily member aga<strong>in</strong>st another (suchas by the husband aga<strong>in</strong>st his wife, orby the father aga<strong>in</strong>st his own child) area type that is less likely than manyother types to be reported to policeand, consequently, less likely to result<strong>in</strong> arrest. <strong>Sex</strong> crimes, particularly thoseaga<strong>in</strong>st children, are a specificexample <strong>of</strong> this type. While some sex<strong>of</strong>fenders <strong>in</strong> this study probably committeda new sex crime after theirrelease and were not arrested or convicted,the study cannot say how many.As mentioned above, one reason whysex <strong>of</strong>fenders are not arrested is thatno one calls the police. Results <strong>from</strong>the National Crime Victimization Survey<strong>in</strong>dicate that the <strong>of</strong>fenses <strong>of</strong>rape/sexual assault are the least likelycrimes to be reported to the police.(See Report<strong>in</strong>g Crime to the Police,1993-2000, March 2003, .)Reconviction Except where statedotherwise, this recidivism measureperta<strong>in</strong>s to State and Federal convictions<strong>in</strong> any State (not just convictions<strong>in</strong> the State that released them) <strong>in</strong> thethree years follow<strong>in</strong>g release. Informationon convictions came <strong>from</strong> Stateand FBI RAP sheets. RAP sheets are<strong>in</strong>tended to document every convictionfor a felony or serious misdemeanor,but not every conviction for a m<strong>in</strong>or<strong>of</strong>fense. “Percent reconvicted” is calculatedby divid<strong>in</strong>g the number reconvictedby the number released <strong>from</strong>prison <strong>in</strong> <strong>1994</strong>. (It is not calculated bydivid<strong>in</strong>g the number reconvicted by thenumber rearrested.)Return to prison Two recidivismmeasures are returned to prison —with a new sentencewith or without a new sentence.<strong>Recidivism</strong> def<strong>in</strong>ed as Returned toprison with a new sentence perta<strong>in</strong>sexclusively to sex <strong>of</strong>fenders who, with<strong>in</strong>3 years follow<strong>in</strong>g release, were reconvictedfor any new crime <strong>in</strong> any Statefollow<strong>in</strong>g their release and received anew prison sentence for the new crime.<strong>Recidivism</strong> def<strong>in</strong>ed as Returned toprison with or without a new sentence<strong>in</strong>cludes resentenced <strong>of</strong>fenders plusany who were returned to prison with<strong>in</strong>3 years because they had violated atechnical condition <strong>of</strong> their release.Technical violations <strong>in</strong>clude th<strong>in</strong>gs suchas fail<strong>in</strong>g a drug test, miss<strong>in</strong>g anappo<strong>in</strong>tment with their parole <strong>of</strong>ficer, orbe<strong>in</strong>g arrested for a new crime. Offendersreturn<strong>in</strong>g to prison for such violationsare sometimes referred to as“technical violators.”<strong>Prison</strong>s should not be confused withjails. A prison is a State or Federalcorrectional facility reserved forconvicted persons with relatively longsentences (generally over a year).A jail is a local correctional facility forconvicted persons with short sentencesor for persons await<strong>in</strong>g trial. Returns toprison refer to any prison, not necessarilythe same prison that released the<strong>of</strong>fender <strong>in</strong> <strong>1994</strong>.The “percent returned to prison with anew sentence” is calculated by divid<strong>in</strong>gthe number returned to prison with anew sentence by the number released<strong>from</strong> prison <strong>in</strong> <strong>1994</strong>. The “percentreturned to prison with or without a newsentence ” is calculated by divid<strong>in</strong>g thenumber returned to prison with orwithout a new sentence by the numberreleased <strong>from</strong> prison <strong>in</strong> <strong>1994</strong>.Data on returns with a new sentenceare based on State and FBI RAPsheets. Data on returns with or withouta new sentence are based on Stateand FBI RAP sheets plus prisonrecords.6 <strong>Recidivism</strong> <strong>of</strong> <strong>Sex</strong> Offenders <strong>Released</strong> <strong>from</strong> <strong>Prison</strong> <strong>in</strong> <strong>1994</strong>
Demographic characteristicsAll sex <strong>of</strong>fendersOf the 9,691 released sex <strong>of</strong>fenders,approximately —• 6,503 (67.1% <strong>of</strong> the 9,691) werewhite males (table 1)• 3,053 (31.5%) were black males• 136 (1.4%) were males <strong>of</strong> other races(Asian, Pacific Islander, AmericanIndian, and Alaska Native).The vast majority <strong>of</strong> sex <strong>of</strong>fenderswere non-Hispanic males (80.1%).Half were over the age <strong>of</strong> 35 whenreleased.Rapists and sexual assaultersAs def<strong>in</strong>ed <strong>in</strong> this report, all sex <strong>of</strong>fendersare either “rapists” or “sexualassaulters.” Of the 9,691 released sex<strong>of</strong>fenders, 3,115 were rapists and therema<strong>in</strong><strong>in</strong>g 6,576 were sexualassaulters.Of the 3,115 rapists, 1,735 (55.7% <strong>of</strong>3,115) were white males and 1,327(42.6%) were black males. Of the6,576 sexual assaulters, 4,768 (72.5%<strong>of</strong> 6,576) were white males and 1,723(26.2%) were black males.Table 1. Demographic characteristics <strong>of</strong> sex <strong>of</strong>fenders released<strong>from</strong> prison <strong>in</strong> <strong>1994</strong>, by type <strong>of</strong> sex <strong>of</strong>fender<strong>Prison</strong>ercharacteristicTotalRaceWhiteBlackOtherHispanic orig<strong>in</strong>HispanicNon-HispanicAge at release18-24*25-2930-3435-3940-4445 or olderAge at releaseAverageMedianTotal releasedAll67.1%31.51.419.9%80.112.2%16.420.019.113.319.036.8 yrs35.39,691Percent <strong>of</strong> released prisoners<strong>Sex</strong>ualRapists assaulters100%100%55.7%42.61.722.6%77.410.6%17.322.420.913.315.536.1 yrs34.93,115100%72.5%26.21.318.9%81.113.0%16.018.818.313.320.637.1 yrs35.56,576Note: The 9,691 sex <strong>of</strong>fenders were released <strong>in</strong> 15 States. Data identify<strong>in</strong>grace were reported for 98.5% <strong>of</strong> 9,691 released sex <strong>of</strong>fenders; Hispanicorig<strong>in</strong> for 82.5%; age for virtually 100%.*Age at release 18-24 <strong>in</strong>cludes the few who were under age 18when released <strong>from</strong> prison <strong>in</strong> <strong>1994</strong>.Rapists and sexual assaulters wereclose <strong>in</strong> age at time <strong>of</strong> release: over70% were age 30 or older. Median ageat time <strong>of</strong> release was about 35 yearsfor both rapists and sexual assaulters.<strong>Recidivism</strong> <strong>of</strong> <strong>Sex</strong> Offenders <strong>Released</strong> <strong>from</strong> <strong>Prison</strong> <strong>in</strong> <strong>1994</strong> 7