Behavioral Research

Table of Contents
1 Description & Theoretical Background
2 Use in Health Behavior Theories

Measures and Measurement


Similar Constructs



6 Measures Appendix
7 Published Examples

Download Full Text (PDF)

Other Constructs



Dispositional Optimism




Illness Representations

  Implementation Intentions
  Intention, Expectation, and Willingness
  Normative Beliefs
  Optimistic Bias
  Perceived Benefits
  Perceived Control
  Perceived Severity
  Perceived Vulnerability
  Self-Reported Behavior
  Social Influence
  Social Support

Kevin D. McCaul and Paul W. Goetz

<< Previous

3 Measures and Measurement

Trait Measures. Most research on worry per se has focused on the construct as a trait, and the "standard" questionnaire is the 16-item Penn State Worry Questionnaire (PSWQ; Meyer, Miller, Metzger, & Borkovec, 1990). A sample item is "I am always worrying about something;" (1 = not at all typical of me; 5 = very typical of me). Meyer et al. reported strong internal consistency (alpha = .93) and good test-retest reliability over an 8-10 week period (r = .92). The PSWQ captures the intensity and controllability of worry (Molina & Borkovec, 1994), and it discriminates between persons with GAD and post-traumatic stress disorder. Other researchers show that the PSWQ has excellent psychometric qualities, whether used with college students (e.g., Fresco, Heimberg, Mennin, & Turk, 2002) or outpatients with anxiety disorders (Brown, 2001).

The PSWQ is a "global" worry measure that is free of worry content; that is, it asks about the extent of worries without identifying the possible targets of those worriers. Tallis, Davey, and Bond (1994) developed a content-dependent worry questionnaire, called "The Worry Domains Questionnaire" (WDQ). It assesses how much people worry about the domains of relationships, self-confidence, the future, work, and finances. Sample items are "I worry that my money will run out," and "I worry that I'll never achieve my ambitions." Test-retest over 3 weeks was good for the overall score (r = .79); not surprisingly, these values were smaller for some domains for which circumstances may strongly affect worrying (e.g., test-retest for worry about relationships was r = .46). The WDQ is strongly correlated with the PSWQ (r = .67), but the two questionnaires also assess different features of worrying. The WDQ, for example, may address task-oriented constructive worrying better than the PSWQ (Davey, 1993a).

Domain Specific Measures. It is likely that people may worry about different, specific health outcomes, and such worries may be independent of or at least only modestly correlated with trait worry. One of the earlier scales in the health area was constructed by Caryn Lerman and her colleagues to study breast cancer (Lerman et al., 1991). The "Lerman Breast Cancer Worry Scale" includes three items, one measuring the frequency of worrying about "getting breast cancer some day," and two items measuring the impact of worry on mood and performing daily activities. In the original study, Lerman et al. averaged the latter two items separately from the worry frequency item, a logical approach because the impact items do not ask about experiencing worry per se.

Partly because of dissatisfaction with the "Lerman Breast Cancer Worry scale," Champion et al. (2004) recently created "The Breast Cancer Fear Scale." The authors especially wished to capture reports of physiological arousal (e.g., heart rate) and also subjective aspects of fear. The focus on physiological responding-even though measured just as self-report-makes this scale different from others that assess worry. Note, for example, the lack of items targeting physiological arousal on the PSWQ. The 8 items on the fear scale provide a scale range of 8-40, and the items hang together well (alpha = .91). The scale is reliable over a 2-month interval (r = .70). It would be interesting to know how closely this scale correlates with other measures intended to measure worry per se.

Because worry involves negative thoughts, some researchers have used the intrusive thoughts subscale from the Revised Impact of Events Scale (RIES; Horowitz, Wilner, & Alverez, 1979) to measure the concept. The intrusive thoughts subscale has strong internal consistency (e.g., alpha = .86) and good test-retest reliability (r = .87 for one week; see Sundin & Horowitz, 2002). The RIES was originally constructed as a measure of stress reactions following traumatic events, and it has been used in that manner in the cancer arena as well (e.g., Vickberg, Bovbjerg, DuHamel, Currie, & Redd, 2000). However, Wells and Papageorgiou (1995) proposed that, to the extent that worry interferes with emotional processing, worry itself could lead to the subsequent production of intrusions. In this model, then, worry actually creates higher levels of intrusive thoughts, a phenomenon that has been demonstrated by Borkovec and colleagues (York, Borkovec, Vasey, & Stern, 1987). These data suggest that intrusive thoughts are a product of different kinds of worrying rather than another measure of worry. The intrusions subscale includes seven items, and it can be reworded to measure how often during the previous week the respondent experienced intrusive thoughts about the consequences of different health outcomes (e.g., one could use lung cancer in the items, producing statements such as, "Pictures about lung cancer popped into my mind;" "Any reminder brought back feelings about lung cancer;" 0 = not at all; 1 = rarely; 3 = sometimes; 5 = often).

Finally, it is possible to construct very brief scales that will measure worry about a particular domain reliably. In our work, we have used two items that ask directly about the degree of worry (e.g., "How worried are you about developing a smoking-related medical condition?", and "How much does thinking about a smoking-related medical condition bother you?" 0 = not at all; 4 = extremely). In a study of college-student smokers, these two items were inter-correlated (alpha = .87), and the average score correlated modestly with the PSWQ (r = .31), intrusive thoughts (r = .29) and the motivation of these smokers to quit (r = .29; McCaul et al., 2005).

Dijkstra and Brosschot (2003) recently took a similar approach, creating a worry "scale" to study smoking. Their 4 items (alpha = .95) focused on the intensity of worry [e.g., "I worry about my health because of my smoking"-not at all (1) to very much (7)]. Smokers in their prospective study who worried more about the health effects of smoking also reported higher quitting activity.

Summary.. How should one measure worry? The PSWQ makes perfect sense if the researcher is interested in a relatively brief, well documented trait-worry scale that lends insight into the general intensity and controllability of everyday worrying. The WDQ also makes sense as a general measure-in this case, one assessing worrying about five everyday domains. Measuring worry about particular health events is more up in the air, and none of the existing scales is perfect. Lerman's scale includes both frequency but also "impact" items, Champion's fear scale asks about fear (rather than worry; see below) and also includes physiological arousal, which is not a part of the worry construct as envisioned by most theorists. Intrusive thoughts are related to worry but may not be the same thing, and the two-item scales we have used address worry intensity (how much do you worry?) but not frequency ("how often do you worry"?). The final scale included in the Appendix is one that we created to illustrate both dimensions of worry. It has not been tested, but it borrows from the other scales and could be a reasonable measure for researchers to try. Note carefully, however, that we do not have psychometric data for it as yet, and we do not know whether worry frequency and intensity should necessarily be combined or whether they should be considered as separate influences on health behaviors. More research along these lines would be welcomed.

<< Previous

Search | Help | Contact Us | Accessibility | Privacy Policy

DCCPSNational Cancer Institute Department of Health and Human Services National Institutes of Health

Health Behavior Constructs: Theory, Measurement, & Research  You can Quit smoking Now!