Maximizing the discriminant validity of phallometric assessment data.

Abstract

Independent data sets from phallometric assessment studies were analyzed to examine methods of maximizing the discriminant validity of phallometric testing. Community volunteers and non-sexoffender patients were compared with rapists and child molesters recruited primarily from a maximum security psychiatric institution. The results indicated that the discriminant validity of phallometric assessments is enhanced by the use of (a) z (rather than raw) scores to compensate for individual differences in responsivity, (b) deviance indexes (computed by calculating the difference in responsivity between deviant and nondeviant stimuli) rather than responses to deviant categories alone, (c) stimuli depicting brutal sexual coercion, and (d) pubescent stimuli for assessing sexual age preferences. Discriminant validity was unimpaired by including subjects with low levels of response. The implications for clinical practice and future research are discussed.

Phallometric Assessment of Sexual Arousal

“…In a study by Earls, Quinsey, and Castonguay (1987), z scores were found to describe the significantly highest proportion of variance (52.7%) in comparison to %FE (32.5%) and raw scores (30.1%). This was supported by Harris, Rice, Quinsey, Chaplin, and Earls (1992), who found z scores to be slightly superior to percentage of full erection. Only one study demonstrated the superiority of %FE to z score transformations as they did not distort the data as much (Barbaree & Mewhort, 1994).…”

“…Although Harris et al (1992) and Murphy and Barbaree (1994) found peak responses to produce more reliable and sensitive indices, Launay (1999) found that both methods provided acceptable outcomes. In the study by Harris et al (1992), better discrimination between offender types was obtained with indices than with scores based on individual categories. Quinsey and Chaplin (1984) found rape indices to be clearly superior in the discrimination of rapists and nonrapists.…”

“…There is little utility in an assessment that provides useful information on only 25% of the subjects tested. Still, the issue continues to be debated, which some writers such as Byrne (2001) maintaining the opinion that nonresponder data are typically not interpretable, while others such as Harris et al (1992) suggest that the data from low responders can be interpreted, provided that responses to sexual or violent stimuli are higher than responses to neutral (i.e., nonsexual, nonviolent) stimuli. Faking.…”

Nudity as a Disinhibiting Cue in a Date Rape Analogue

“…Standardizing the responses minimizes the effect of individual variation in overall responsiveness (Harris, Rice, Quinsey, Chaplin, & Earls, 1992).…”

“…Standardizing responses minimizes the effect of individual variation in responsiveness (Harris et al, 1992). Responses to the post-stimulus questions were averaged for each category and not standardized.…”