Fig. 2.

Two-step clustering identifies groupings of categorical and continuous variables by running preclustering first and then by running hierarchical methods. Two clusters of approximately equal size could be obtained (A) which related to stimulus type (i.e., picture or video). Presentation paradigm (i.e., block design or event related) and sexual orientation were found to be weak predictors for classifications (B), while continuous variables such as machine power, age, or biological sex of study participants showed similar density distributions in both clusters (C).