In this study we intended to analyse if the REF Scale is a good indicator to differentiate the two subtypes of paranoia “Bad Me” and “Poor Me” (Trower & Chadwick, 1995).

Participants: We analyzed data from a different sample of previous studies with 326 participants (64.11% women, mean age 30.8, SD = 10.84), of whom 212 were patients.

Design, materials and procedure: We used the REF-scale (Lenzenweger et al., 1997) adapted to Spanish language, of which we deleted two items because of psychometric criteria, resulting 32 dichotomic items. We used SPSS 15.0 to conduct a principal-components factor analysis with a varimax and oblimin rotation, retaining two factors.

Two factors explained 31.32% of the variance (rotated solution). We interpreted factor through factor loadings higher than .42. Factor 1 accounted for 18.28% of the variance and it's associated with referential laughter, commentaries and guilt. Factor 2 accounted for 13.05% of the variance and it's associated with referential concerns related to the media.

Since the inter-factor correlation is moderate (.44) and there are no relevant clinical differences about the content between the two factors, the REF scale is a one-dimensional measure. Therefore, two big factors don’t emerge from the REF scale related to referential concerns about laughter-commentaries and guilt that correspond to “Poor Me” and “Bad Me”, respectively.