An examination of potential biases in research designs used to assess the efficacy of sex offender treatment