106 COGNITIVE THEORY AND RESEARCH ON ANXIETY
Experimental Validation
If AS amplifies fear reactions, then high AS should lead to more intense anxiety in
response to a wider range of stimuli (Reiss & McNally, 1985; see Taylor, 2000). This
should be particularly evident in biological challenges that provoke panic attacks under
controlled laboratory conditions or other experimental manipulations that elicit the
physical symptoms of anxiety (McNally, 1996). In fact, there is now considerable empiri-
cal evidence that baseline AS predicts postchallenge anxiety symptoms and panic attacks
in people with or without diagnosable panic disorder (for reviews, see McNally, 2002;
Zvolensky, Schmidt, Bernstein, & Keough, 2006). High AS predicts fearful response
and panic symptoms to carbon dioxide (CO 2 ) inhalation (e.g., Rapee, Brown, Antony, &
Barlow, 1992; Rassovsky, Kushner, Schwarze, & Wangensteen, 2000; Schmidt & Mal-
lott, 2006), hyperventilation (Carter, Suchday, & Gore, 2001; Holloway & McNally,
1987; McNally & Eke, 1996; Rapee & Medoro, 1994), and caffeine ingestion (Telch,
Silverman, & Schmidt, 1996). Although ASI Physical Concerns may be the only AS
dimension that predicts fear response to a physical challenge (Brown, Smits, Powers,
& Telch, 2003; Carter et al., 2001; Zvolensky, Feldner, Eifert, & Stewart, 2001), these
experimental findings support the predictive validity of the ASI and its special relevance
to panic- spectrum psychopathology (Zvolensky et al., 2006).
Diagnostic Specificity
If AS is a specific cognitive- personality vulnerability factor for anxiety, then it should
be significantly more elevated in anxiety, especially panic disorder, than in other clini-
cal and nonclinical samples (McNally, 1994, 1996). Individuals with panic disorder or
agoraphobia score on average two standard deviations above the normative mean on
the ASI (McNally, 1994, 1996; Reiss, 1991; Taylor, 1995a, 2000) and anxiety disorder
samples (except simple phobias) score significantly higher than depression or nonclinical
comparisons (Taylor & Cox, 1998; Taylor, Koch, & McNally, 1992). Within the anxi-
ety disorders, persons with panic disorder and agoraphobia score significantly higher
than the other anxiety disorders, with PTSD, GAD, OCD, and social phobia scor-
ing significantly higher than nonclinical comparison groups (Deacon & Abramowitz,
2006a; Taylor, Koch, & McNally, 1992a). At the symptomatic level ASI has a specific
association with self- report of panic attacks in nonclinical child and adult populations
(e.g., Calamari et al., 2001; Cox, Endler, Norton, & Swinson, 1991; Longley et al.,
2006), although some studies have found AS relates to depressive symptoms as well
(Reardon & Williams, 2007).
The ASI subscales appear to have differential specificity for anxiety and panic.
ASI Physical Concerns is the only dimension specific to panic disorder whereas Social
Concerns may be more relevant to social phobia (e.g., Deacon & Abramowitz, 2006a;
Zinbarg et al., 1997) and Cognitive Dyscontrol may be related to depression (Cox et
al., 2001; Rector, Szacun- Shimizu, & Leybman, 2007). However, caution must be exer-
cised when using the ASI to screen for anxiety or panic. Hoyer and colleagues examined
the predictive accuracy of the ASI, BAI, and several other anxiety measures in a large
epidemiological sample of 1,877 young women in Dresden, Germany (Hoyer, Becker,
Neumer, Soeder, & Margraf, 2002). None of the measures alone were able to accu-