Asking questions about urges or cravings for cigarettes.

Abstract

Smokers (N = 116) were administered the Questionnaire of Smoking Urges (QSU; S. T. Tiffany & D. J. Drobes, 1991) to explore the measurement of drug urges or cravings. Confirmatory factor analysis replicated the 2-factor structure, using the 6 best items on each of the QSU factors, although further analyses indicated that 1 conceptual factor may be a better fit. Three different categories of internally consistent items were identified within the QSU: urges to smoke, expectancies from smoking, and intentions to smoke. Path-modeling techniques were used to demonstrate patterns of interrelationships among these categories. Despite the widespread criticism of single-item scales, the present approach indicated that they are useful. In this sample, a 2-item or 3-item "desire" scale effectively measured urges to smoke. Complex scales can obscure the direct measurement of urges or cravings for a cigarette. Kozlowski and Wilkinson (1987a, 1987b) proposed that the study of drug urges ("cravings") necessarily involves the study of selfreports of the intensity of the desire to use drugs and that ideal studies should include behavioral measures of drug taking and measures of biological events as well as subjective reports. Self-report measures are commonly used in drug research to establish urges to use drugs, but there has been relatively little work on how best to ask questions about drug urges.In the smoking literature, self-report measures of urges vary in length, content, and complexity and frequently include wording such as "craving a cigarette," "desire to smoke," "urge to smoke," "missing a cigarette," "need to smoke," and "want to smoke." Numerous measures have included one or two face-valid items rated with fixed-point Likert-type scales (e.g.,

Revisiting the Factor Structure of the Questionnaire on Smoking Urges.

“…It is possible that the original two-factor structure reported by Tiffany and Drobes (1991) and replicated by Davies et al (2000) cannot be confirmed with older, treatment-seeking individuals who have been smoking larger quantities of cigarettes over a longer period of time. It will be important for future studies to test whether or not this original two-factor structure using all 26 items can be confirmed in such samples.In the present study, an alternative model, initially proposed by Kozlowski et al (1996), using the 12 most robust items from the original analysis appeared to represent an adequate fit for the data. Structural equation models, because they are merely an approximation of reality, are not generally expected to fit the data perfectly (Floyd & Widaman, 1995).…”

“…To address this problem, several studies have developed longer urge scales with as few as 5 items or as many as 30 items, using more traditional psychometric approaches (e.g., Flannery, Volpicelli, & Pettinati, 1999;Tiffany, Singleton, Haertzen, & Henningfield, 1993). These scales, designed to assess either a single unidimensional factor (e.g., the Penn Alcohol Craving Scale; Flannery (Cox, Tiffany, & Christen, 2001;Kozlowski et al, 1996). Therefore, a questionnaire that is short enough for practical use but long enough to provide a reliable measure of craving is ideal.…”

“…First, the original factor analyses of the QSU used such items. Second, data had not been gathered for positively worded items like that collected by Kozlowski et al (1996). Whether negatively worded items should be used is an issue that deserves further investigation.…”

“…For this latter model, they replaced the negatively worded items with identical items that were positively worded. Kozlowski et al (1996) tentatively suggested that this one-factor model might be a better fit for their data.Thus, little previous research has focused on attempting to replicate the factor structure of the QSU. Moreover, the two studies that tried to do so used relatively small samples as compared with the number of variables being analyzed.…”

“…The present study assessed both of the two-factor models (Kozlowski et al, 1996; that were assessed in previous research. Additionally, the one-factor model suggested by Kozlowski et al (1996) was tested.It should be noted that it was not possible to conduct the exact same analyses as in Kozlowski et al (1996) when evaluating this latter one-factor model because of the wording changes made by these researchers (i.e., Kozlowski et al, 1996, replaced negatively worded items with identical positively worded items). However, it was deemed prudent to test this third model as this issue (i.e., the number of factors related to craving) has been debated in craving research (Tiffany, 1997).…”