Psychoticism (P) scales tend to include items that are rarely endorsed (e.g. "I am guided by voices"). Thus, we find that scores on this kind of scale are generally quite low, with even the high scorers not exceeding the mid-point on the scale. But, given the possibility that high P scorers might be expected to respond in a haphazard way, some have theorized that such persons may score higher because they are answering randomly, not because their responses reflect an accurate accounting of the way that they think and feel.
    The average P score on the Eysenck Personality Questionnaire is about 3 with a high score being deemed to be 8 or more. This scale has 24 items, so those reponding in a totally random fashion would produce scores clustering around a value of 12 (the median). Clearly then, high P scorers are not responding in a completely random fashion, but both theories (random vs content-based responding) are nonetheless tenable as explanations of the high P scores that have been observed. But if high scorers are acting randomly on the P scale, we would expect them to also produce relatively high scores on any other scale that produces average scores that are well below the median. Put another way, the difference between the observed score and the expected score would on average be less for individuals who are responding randomly than for those responding to the content of the questions. This study examined that possibility by comparing P scores with scores on a 21-item Lie scale that also characteristically produces a low mean score in contrast to the expected average (see the table below left) . The study involved 103 young adult English males who were administered a test that provided measures of P and L as well as extraversion (E) and neuroticism (N). The latter two tests did not tap rare behaviours, and had average scores that were very close to the expected values.
    The crucial correlations are shown in the table below right. The finding that the P difference score did not correlate with the L difference score indicates that those that score in the putative "random responding range" (near the median) on one scale do not tend to do so on the other. The only significant correlation in the table is that between P and E, all the rest are trivial. Overall, the findings provide no support for the "random responding" hypothesis, suggesting that variation in P scores represents a response to the content of the items.