older' (coded as '1'), 'seventeen' (coded as '2'), 'sixteen' (coded as '3'), 'fifteen' (coded
as '4'), 'fourteen' (coded as '5'), and 'thirteen or younger' (coded as '6'). Similarly,
response options for the second question included: 'I have never had sexual intercourse'
(coded as '0'), '1 partner' (coded as '1'), '2 partners' (coded as '2'), '3 partners' (coded
as '3'), '4 partners' (coded as '4'), '5 partners' (coded as '5'), and '6 partners or more'
(coded as '6'). The two questions were summed and coded as a sexual risk taking index
(SEXRISK) and have a scale reliability of .81. Scale scores ranged from 0 to 12 whereas
a higher scale score indicates a higher level of sexual risk taking (meaning younger age at
first sexual intercourse and a higher number of total sexual partners).
Analytic Plan
Bivariate correlations were used to determine how variables are related in terms of
multicollinearity. Several statistical models were estimated in order to explore
relationships between the dependent variables, stalking perpetration and victimization,
and eleven independent variables. Separate analyses will be conducted for stalking
victimization and stalking perpetration. Logistic regressions will be used because both
dependent variables are dichotomous and because this study aims to predict the presence
of characteristics based on values of a set of predictor variables. Odds ratios will be used
to predict the probability for which participants are either stalking victims or perpetrators.
Logit regression models for stalking victimization and perpetration will be
estimated for the full sample. If gender has a significant influence on stalking
perpetration or victimization in the full sample models, the data will be split so that male
and female models can be estimated separately. The six models would include three for
victimization (full model, female model, and male model) and three for perpetration (full
model, female model, and male model). If the full model for either dependent variable