Authors' reply

We would like to point out that we do not maintain that those in the offender group are less violent than their non-offender counterparts. Rather, we conclude that, as stated in the Results section, people in the offender group were significantly more likely to display some types of challenging behaviour but significantly less likely to display others. The results showing a reduction in the frequency of challenging behaviour during admission measured the change in rate of challenging behaviour per person per week by comparing a 4-week baseline period with the last 4 weeks of admission. Thus, these figures do not show the level of challenging behaviour exhibited in each group throughout their stay. The fact that there was no significant between-group difference in the rate of total incidents of challenging behaviour per month is shown correctly in Table 2. We thank Dr Marshall for giving us the opportunity to clarify this point.