Dear All
I am emailing the list in the hope that one of the members may be able to
advise me on a statistics query. I am conducting a PhD part of which is
investigating whether juvenile sex offenders are consistent in their
offence behaviours across series of offences. Previous literature in the
area has coded offence behaviours dichotomously (1 for present and 0 for
absent). Past authors have therefore used this information to calculate a
Jaccard's coefficient to indicate similarity between pairs of
offences. Often crime analysts are concerned with assessing similarity
between more than two offences therefore my problem relates to how to
assess similarity between more than two offences. I am wondering whether
it is possible to calculate a Jaccard's coefficient for each pair of
offences within a series (for example between offence A and B, A and C and
B and C) and then to calculate a mean Jaccard's coefficient to give an
overall idea of the degree of consistency within a series. Any indication
as to whether this is an appropriate procedure to follow or suggestions for
alternative methods would be greatly appreciated via my email below.
Best wishes.
Jessica
Ms. Jessica Woodhams
Email: [log in to unmask]
|