The controversy with regard to using reverse or negativelyworded survey stems has been
around for several decades; it is a practice of questionable utility intended to guard
against acquiescence or response set behaviors. A 2 ´ 3 design in which item stem direction
and item response pattern direction were crossed was used to determine effects on
internal consistency reliability as measured by Cronbach’s alpha. The condition having
the highest alpha was when all directly worded stems were used with bidirectional
response options. Alphawas higher and accounted for at least 10%, and in one case 20%,
higher internal consistency as compared with any of the three conditions in which negatively
worded stems were used. This would indicate that the use of all directly worded
stems and half of the response options going in one direction and half going in the other
direction may be a better way of guarding against acquiescence and response set behaviors
than the use of items with negatively worded stems.