Shortened versions of the Dogmatism Scale failed to represent the full scale

Evidence

The ten-item version correlated only . and . in two samples with the fortyitem version (which had the same ten items in it, providing a perfect match with one quarter of the criterion). Such a low overall figure strongly implies that the ten items chosen did not adequately represent whatever the forty items measure. Indeed, the correlation of the Chosen Ten with the Rejected Thirty was given for the second sample and equaled only .46, a terribly low figure. — Bob Altemeyer, The Authoritarian Specter, 4251