Prediction of Homicide and Suicide: A Test in a Healthy Risk-Taking Group

Kendra (1974) and Lester, et al. (1975) have derived step-wise linear discriminant functions to predict suicidal and homicidal behavior from Rorschach data. The functions have been validated on appropriate criterion groups. In order to explore how these functions would operate in a psychologically healthy group, Rorschach data on astronaut candidates were examined.' The sample consisted of 32 astronaut candidates. When their Rorschach data were inserted into the linear discriminant functions for predicting homicide, 24 of the men were classified a s murderers and 8 as non-murdering criminals? For the nine candidates selected, five were classified as murderers and 4 as non-murdering criminals. Rorschach data were also inserted into Kendra's functions for predicting suicidal behavior, modified so as to accommodate the scoring system for the Rorschach protocols of the a~tronauts.~ Thirty of the 32 astronaut candidates were classified as non-suicidal and two as attempted suicides. These results can be seen as evidence for the validity of Kendra's procedure for predicting suicide (or alternatively as indicating that suicidal impulses play little part in the motivation to volunteer as an astronaut). The data on the prediction of homicide do not permit conclusions about the validity of the linear discriminant functions. However, if the functions are assumed to be valid, it appears that action-oriented or aggressive acting-out impulses are common in volunteers for astronaut duty, although they are not conducive to performance.