This study involved uating prison-based sex offender treatment programs and developing a Korean version of Multidimensional Inventory of Development, Sex and Aggression(MIDSA; Knight, 2009).

Study 1: uation of prison-based sex offender treatment programs

We adopted one group pre-post research design and collected pre-and post-treatment scores for the sex offender treatment program from 9 prisons. The total number of participants in this study was 109. All participants had taken part in a sex offender programme in Korean prison service in 2013. They completed the 13 psychometric assessment measures before and after they received treatment. Treatment change was examined using repeated measures of ANOVA. Two repeated measures ANOVAs were carried out for each measure using static risk and offence type in turn as the between-subjects factor and pre- and post-treatment scores as the within-subjects factor. Analyses found that for 7 measures out of 13 scores decreased significantly pre- to post-treatment. Scales measuring rape attitude, anger inhibition, impulsivity, child molestation, loneliness, self-esteem, and subjective well-being showed main effects for time (pre-treatment and post-treatment). Effect sizes were also examined using partial Eta squares, and for Rape attitude and Child molestation, partial Eta squares were .258 and .233 representing strong treatment effects. For Loneliness and Self-esteem, partial Eta squares were .127 and .158 representing medium treatment effects. Scales measuring coping using sex under stress, anger expression, interpersonal reactivity, hostility toward women, sexual fantasy did not show any significant results.
When examining the scores on measures by risk level, for Anger Expression, there was a difference between static risk categories. One-way ANOVAs were performed to investigate these differences. A difference between risk categories was found both pre-treatment and post-treatment. Examination of the mean scores and post hoc tests suggest that as risk category increases from low to very high, scores on the Anger Expression increase. For Impulsivity, there was no significant difference between risk categories pre-treatment, but there was a significant difference between medium risk and high risk post-treatment. For Loneliness, although there was no significant difference between risk categories, high risk offenders significantly scored higher than low risk offenders. This suggests even after the sex offender treatment, high risk offenders might have greater level of problems in areas such as emotional loneliness, anger control, impulsivity.
When examining scores on the measures by offense type, the only significant finding was found on the Child Molestation. Child molesters scored significantly higher than rapists in Child Molestation.

Study 2: Development of K-MIDSA

Last year, we had reviewed the full items of original MIDSA and validated the basic section(33 attitude scales and 4 lie scales) of the entire inventory.
We attempted to cross-validate the findings from the former study and this year again, we administered the K-MIDSA for 103 sex offenders and 300 community people. We incorporated the subjects of last year into current research and thus the final samples including ones from the previous year consisted of 391 sex offenders and 370 community people.
The psychometric properties of K-MIDSA were as follows. The internal consistencies for K-MIDSA subscales were acceptable. Cronbach’s αs for most subscales were higher than .70. Especially, sexualization scales, sexual perversion scales, child molestation scales, social self-esteem scales had highest internal consistencies as seen from Cronbach’s αs. But, masculine adequacy-related scales, psychopathy-related scales, hypermasculinity-related scales showed relatively low internal consistencies.
The convergent validity was examined through comparing K-MIDSA scales and measures currently used in prison. Sexualization scales of K-MIDSA had a high convergent validity with Coping Using Sex Inventory and Sexual Fantasy Questionnaire. Sexual Perversion Scale had a good discriminant validity with all measures currently used in prison. Expressive Aggression Scale had a good convergent validity with Anger Expression scale and good discriminant validity with Hostility toward Women Scale. Among Psychopathy-related Scales Lack of Empathy Scales had a good convergent validity with the IRI, and Hostility toward Women with Rape Attitude Scale. Pervasive Anger Scale had a good convergent validity with Anger Expression Scale. Cognitive Distortion about Child Sexual Assault had a good convergent validity with Child Molestation Scale. Lastly, Social Self-Esteem Scale had a good convergent validity with Loneliness Scale and Ronsenberg Self-Esteem Scale.
Group differences between sex offenders and community people were examined. Most of results were in the same direction as results of last year. For instance, for sexual compulsivity, sex offenders scored significantly higher than community people. Paraphilia-related scales, child molestation-related scales produced the statistically significant differences between those two groups. Sex offenders showed higher scores in those scales. But, for other scales, we could not find statistically significant differences. Group differences between recidivists and non-recidivists were also examined, and mot of results were similar to results of last year. Recidivists showed higher scores in sexual compulsivity, paraphilia-related scales, child molestation-scales. We divided the sex offenders into 3 groups: rapists, child molesters, mixed type. Among the 3 groups, mixed typed sex offenders showed the highest scores in sexual compulsivity and paraphilia-related scales. Implications for these results were discussed with limitations of this study and improvements that should be made in the future.