Measuring drug urges by questionnaire: Do not balance scales

Measures designed to assess self-reported drug urges are routinely balanced; that is, they contain a relatively equal number of positively and negatively worded items. Using smoking urges as an example, we explored the effect of item wording on responses to self-report items. One hundred sixteen cigarette smokers responded to 32 positively worded statements and their 32 negatively worded counterparts adapted from the Questionnaire of Smoking Urges. For 59% of the paired statements, the negatively worded item was scored higher than the positively worded item. Significant mean differences (item means for 56% of the paired statements were significantly different at the .01 level) and only moderate correlations (mean = .54; range = -.04 to .79) between item pairs were found.