determined by conducting a power analysis for a repeated measures analysis of variance
(ANOVA) and a multiple regression analysis. Degrees of freedom for the F-ratio (u) was
determined by calculating (k-1) (r-1) where k and r are the number of levels of interacting
main effects (Cohen, 1988). In the present study, u= (5-1) (2-1) = 4. The number of
participants required to find a small effect size f=. 10 (Cohen, 1988) with u=4 with power
of .80 and an alpha level of .05 is 240 (Cohen, 1988).
The regression analysis was designed with 3 predictors in the model. A sample size
of 51 is required to detect a squared multiple correlation coefficient of R2=.20 with power
= .80 and a type-II error rate a = .05. However, the .95 confidence interval with a sample
size of 51 is R2 +/- .20, which means there is a 95% chance the squared multiple
correlation will be between .00 and .40, which is too wide. The sample size required to
achieve a 95% confidence interval +/- .10 is 200, which means there is a 95% chance the
squared multiple correlation will be between .10 and .30. A sample size of 240 is
sufficient to achieve this confidence band.
A brief overview of the nature of the research, type of instruments and time
required was made to all prospective participants. Law Enforcement Officers reviewed
the informed consent form from the Institutional Review Board (IRB) and signed consent
forms were obtained prior to participants taking the instruments.
Instruments
The instruments used in this research study included: 1) the Modified Hate Crime
Survey (MHCS), a survey designed to measure attitudes towards hate crime scenarios
from each of 5 protected categories (race, religion, sexual orientation, gender and
disability); 2) the ATDP-A, a self-report measure of attitudes towards people with