Changing Time Frames on Symptom Inventories: Effects on the Psychiatric Epidemiology Research Interview

The effect of changing a time frame used in the Psychiatric Epidemiology Research Interview (PERI) from 1 year to 1 month was studied by randomly assigning year and month versions of the PERI to a sample of respondents from an urban community. Although the response format (never, almost never, sometimes, fairly often, very often) of PERI items requires an implicit averaging of symptom frequencies over time, we found that the mean values of the scale scores were consistently larger for the year time frame than they were for the month time frame for all 14 symptom scales. The variances of the scales in the year version were also larger than those in the month version, and the covariance pattern of the scales within the versions was different, although this latter difference seemed to result from differences on only 4 of the scales. The internal consistencies of the scales in both versions were good to very good. Neither version was consistently better in distinguishing persons with known psychopathology (sampled separately from psychiatric treatment facilities) from persons with no known psychopathology.