Anxiety sensitivity, conscious awareness and selective attentional biases in children

ARTICLE IN PRESS

Behaviour Research and Therapy 45 (2007) 497–509 www.elsevier.com/locate/brat

Anxiety sensitivity, conscious awareness and selective attentional biases in children Caroline Hunta, Edmund Keoghb,, Christopher C. Frencha a

Department of Psychology, Goldsmiths College, University of London, New Cross, London, SE14 6NW, UK b Department of Psychology, University of Bath, Claverton Down, BA2 7AY, UK Received 18 May 2005; received in revised form 13 April 2006; accepted 13 April 2006

Abstract The current study investigated whether physical anxiety sensitivity (AS) is associated with selective attentional biases to affective stimuli in children. The dot-probe paradigm was used to examine the deployment of attention towards words pertaining to anxiety symptomatology, socially threatening words, and positive words, in samples of 8–10-year-old children. Word pairs were presented under both masked and unmasked conditions. Irrespective of masking, children high in physical AS displayed an attentional vigilance for emotional words relative to neutral words, whereas those low in physical AS displayed a relative avoidance of such material. The results of this study are interesting as they not only suggest the presence of automatic AS-related biases in childhood, but that this is a general emotionality bias rather than one related to specific anxiety-related stimuli. r 2006 Elsevier Ltd. All rights reserved. Keywords: Anxiety sensitivity; Selective attention; Dot-probe paradigm; Children

Introduction Anxiety sensitivity (AS) is the fear of anxiety-related sensations such as palpitations, trembling, and dizziness, which arises from the belief that such sensations have aversive consequences. According to Reiss’ expectancy theory, AS also plays a role in the pathogenesis of fear, anxiety, and panic (Reiss, 1991; Reiss & McNally, 1985), with support coming from psychometric studies, panic provocation tasks, and prospective investigations (e.g., Asmundson, Norton, Wilson, & Sandler, 1994; Schmidt, Lerew, & Jackson, 1997; Schmidt, Lerew, & Joiner, 1998, 2000; Sturges, Goetsch, Ridley, & Whittal, 1998). Although the subject of considerable attention in adults, there is emerging research to suggest that AS may also be important in both children and adolescents. Not only have individual differences in adult AS been found to be partly attributable to differences in childhood experiences, but also levels of AS are higher in children meeting diagnostic criteria for panic disorder (PD) than those with other anxiety disorders (e.g., Kearney, Albano, Eisen, Allan, & Barlow, 1997; Lau, Calamari, & Waraczynski, 1996; Scher & Stein, 2003). Corresponding author. Tel.: +44 (0)1225 383671; fax: +44 (0)1225 386752.

E-mail address: [email protected] (E. Keogh). 0005-7967/$ - see front matter r 2006 Elsevier Ltd. All rights reserved. doi:10.1016/j.brat.2006.04.001

ARTICLE IN PRESS 498

C. Hunt et al. / Behaviour Research and Therapy 45 (2007) 497–509

Furthermore, childhood AS predicts anxiety experienced in response to behavioural challenge tasks as well as the onset of panic attacks (Hayward, Killen, Kraemer, & Taylor, 2000; Rabian, Embry, & MacIntyre, 1999; Weems, Hayward, Killen, & Taylor, 2002; Wilson & Hayward, 2005). Although there are inconsistencies in findings with clinical anxiety, in non-clinical groups, parental AS may also be related to children’s AS levels (Mannuzza et al., 2002; Tsao et al., 2005; van Beek, Perna, Schruers, Muris, & Griez, 2005). Finally, recent factor-analytic investigations of the Anxiety Sensitivity Index (ASI; Reiss, Peterson, Gursky, & McNally, 1986) and the Childhood Anxiety Sensitivity Index (CASI; Silverman, Fleisig, Rabian, & Peterson, 1991) are similar in that both have found AS to be best represented by a hierarchical factor structure with two, three or four lower-order factors (pertaining to physical, social, and mental incapacitation concerns) loading on to a single higher-order general AS factor (Cox, Parker, & Swinson, 1996; Dehon, Weems, Stickle, Costa, & Berman, 2005; Muris, 2002; Silverman, Ginsburg, & Goedhart, 1999; Silverman, Goedhart, Barrett, & Turner, 2003; Stewart, Taylor, & Baker, 1997; Walsh, Stewart, McLaughlin, & Comeau, 2004; Zinbarg, Brown, & Barlow, 1997). Furthermore, adult studies suggest that the strongest and most consistent lower-order factor relates to physical concerns (Keogh, 2004). Given the similarity between AS in adults and children, it is possible that similar mechanisms underpin both. Within adults, research has focused on isolating the cognitive mechanisms that may be important in AS, especially those relating to awareness or hypervigilance for bodily sensations. Objective measures of attentional bias, such as the emotional Stroop and the dot-probe paradigms have been generally used to examine emotion-related attentional biases (Williams, Watts, MacLeod, & Mathews, 1997). These tasks reveal that both non-clinical high trait-anxious individuals and clinically anxious patients exhibit automatic (hyper)vigilance for stimuli pertaining to their current concerns (e.g., McNally, Kaspi, Riemann, & Zeitlin, 1990; Mogg, Bradley, & Williams, 1995; Mogg & Marden, 1990; Mogg, Mathews, Bird, & Macgregor-Morris, 1990). Research investigations that have applied such tasks to adults high in AS suggest that similar selective attentional biases exist (Hunt, Keogh, & French, in press; Keogh, Dillon, Georgiou, & Hunt, 2001; Koven, Heller, Banich, & Miller, 2003; Stewart, Conrod, Gignac, & Pihl, 1998). In addition, such biases may be dependent on the component of AS under investigation. For example, Keogh et al. (2001) found that individuals high in physical AS showed a specific vigilance for physical threat-related words, whereas individuals with low physical AS showed avoidance of these words. More recently, Hunt et al. (in press) not only partly replicated this effect, but also found that such specific physical AS-related vigilance may occur below the threshold of conscious identification; this suggests that biases may reflect automatic rather than strategic processes. Unfortunately, while such AS-related attentional biases have been found in adults, there has been relatively little investigation of such effects in children. Instead, most of the research conducted on attentional biases in children has either used subjective measures of vigilance or focused on anxiety-related biases without consideration of the AS construct (Daleiden & Vasey, 1997). For example, subjective studies into heart rate awareness suggest that children high in AS may be particularly vigilant for such cues (Eley, Stirling, Ehlers, Gregory, & Clark, 2004). Of the studies that use more objective attentional tasks it seems that anxious children may indeed be characterised by attentional biases similar to those found in adults (e.g., Dalgleish, Moradi, Taghavi, Neshat-Doost, & Yule, 2001; Martin, Horder, & Jones, 1992; Moradi, Taghavi, Neshat-Doost, Yule, & Dalgleish, 1999; Schippell, Vasey, Cravens-Brown, & Bretveld, 2003; Taghavi, Neshat-Doost, Moradi, Yule, & Dalgleish, 1999; Vasey, Daleiden, Williams, & Brown, 1995; Vasey, El-Hag, & Daleiden, 1996). There are, however, inconsistencies in the pattern of results found in such child studies. For example, two studies with children aged 8–12 have failed to find a fear-specific Stroop interference effect, in that threat biases were found in both high- and low-fear groups (Kindt, Bierman, & Brosschot, 1997; Kindt, Brosschot, & Everaerd, 1997). Also when Waters, Lipp, and Spence (2004) presented threat-related, positive and neutral pictures in a dot-probe study, they found that both clinically anxious and non-anxious control children showed a threat bias. There was also a suggestion that the clinically anxious children show a bias towards positive stimuli as well. Some of these inconsistencies may be due to methodological differences between studies. For example, stimulus type may be important in that differences in results are found when using words compared to pictures (Kindt & Brosschot, 1999; Kindt, van den Hout, de Jong, & Hoekzema, 2000). There has also been a failure to included positive stimuli, which allow for the examination of a general emotional bias as opposed to a specific

ARTICLE IN PRESS C. Hunt et al. / Behaviour Research and Therapy 45 (2007) 497–509

499

threat-related bias (Schippell et al., 2003; Vasey et al., 1995, 1996). Furthermore, the majority of studies with children have used the modified Stroop task (e.g., Kindt & Brosschot, 1999; Martin et al., 1992; Moradi et al., 1999), and the use of alternative paradigms have not been examined. To complicate matters even further, at least one study reports age-related effects. Kindt et al. (2000) compared children aged 8–11 and found that within the 8-year-olds, the non-anxious controls exhibited greater interference effects for spider-related words. Furthermore, it seemed that for spider-fearful children there was an age-related increase in interference effects, whereas for the low-anxious controls there was a relative decrease. Although numbers were small in the younger age group, this suggests that age may be an important factor. Whereas such biases are a characteristic of all young children, as they develop they learn to suppress (or control) such effects. Those who do not do so may be more susceptible to the development of an anxiety disorder (Kindt & van den Hout, 2001). This is important when one considers that the age range within studies varies; some are fairly restrictive (e.g., children aged 8–12; Kindt & Brosschot, 1999), whereas in others, usually clinical studies (where presumably sampling opportunities are restricted), ages tend to span a much wider range (e.g., children aged 9–18; Taghavi et al., 1999). In sum, it seems that whereas there is evidence emerging to suggest that AS, especially the physical concerns component, may be associated with a selective attentional bias in adults, such an effect has yet to be demonstrated within children. In light of this, the experiment to be reported here utilised the dot-probe task to examine the selective attentional biases of children who vary in their levels of AS. The dot-probe task was chosen over the emotional Stroop task, mainly due to the success that we have had using this paradigm with adults. The primary objective of this study was to extend the findings obtained by Keogh et al. (2001) and Hunt et al. (in press) with adults, to investigate such effects in children. We therefore sought to keep as many of the design characteristics of our adult AS studies to allow for future comparisons. In order to examine the specificity of any attentional bias, neutral words were paired with three different types of emotional words: anxiety symptomatology, social threat, and positive. Since there is good evidence to suggest that the strongest dimension of AS is related to physical concerns, and given that the findings from adult AS dot-probe studies suggest that the physical concerns components of AS is related to attentional biases, we selected children on the basis of this component of AS. The secondary objective was to examine whether such biases occur above and below the level of conscious awareness, by running the study under both masked and unmasked conditions. Similar procedures were used to those reported in a recent AS study with adults (Hunt et al., in press). Given the results obtained in adult samples, it was hypothesised that children with high physical AS would exhibit a specific attentional vigilance to words pertaining to anxiety symptomatology relative to neutral words, in both unmasked and masked conditions. Methods Design A mixed-groups design was utilised. The between-groups factors were physical AS group (high vs. low) and presentation condition (masked vs. unmasked), and the within-groups factors were word-pair type (anxiety symptomatology/neutral vs. social threat/neutral vs. positive/neutral), emotional word position (upper vs. lower), and probe position (upper vs. lower). The dependent variable was probe detection latency measured in milliseconds (ms). Participants A total of 166 children (84 girls and 82 boys) from the year 4 classes of two schools in the United Kingdom initially completed the CASI. Physical AS was determined by totalling the CASI items (3, 4, 6, 8–11, 14, 16, and 18) that represented AS-physical concerns in a previous study we conducted that examined the factor structure of the CASI (Hunt, Keogh, & French, in prep.). Physical AS groups were determined on the basis of the upper and lower quartile scores obtained on the AS physical concerns sub-scale in the aforementioned study. Of these half from each AS group were randomly assigned to the unmasked or masked word presentation condition. Of those in the unmasked condition, the data of five children were subsequently

ARTICLE IN PRESS C. Hunt et al. / Behaviour Research and Therapy 45 (2007) 497–509

500

excluded from analyses due to having an excessive number of outliers or incorrect responses on the dot-probe task. Thus there were 23 children in the high physical AS group (14 girls, 9 boys), and 16 in the low physical AS group (6 girls, 10 boys). For those in the masked condition, the data of five participants were subsequently excluded from analyses: four had an excessive number of outliers or incorrect responses on the dot-probe task, and one had above-chance performance on the awareness checks. Thus, for the masked trials, the high physical AS group consisted of 20 children (11 girls, 9 boys), and the low physical AS group consisted of 19 children (11 girls, 8 boys). The age range of the final sample was between 98 and 130 months (i.e., 8–10 years). Stimuli The stimulus words (see Table 1) consist of 108 pairs of words; each pair comprised of an emotional word matched with a neutral word. Emotional words were grouped into three categories (anxiety symptomatology, social threat, positive), and were used to determine if physical AS would be associated with a specific bias to

Table 1 Word pairs used in the dot-probe task Anxiety Symptomatology–neutral

Social threat–neutral

Positive–neutral

Afraid Anxious Beating Blackout Burning Collapse Confuse Cramps Crazy Dizzy Faint Fear Forget Fright Gasping Giddy Insane Mad Mental Nervous Numb Panic Pounding Scared Shaking Shiver Sick Smother Sting Strangled Sweat Terrified Throbbing Thumping Tingle Tremble

Alone Annoyed Ashamed Avoid Awkward Blamed Blushed Coward Critical Defeat Despise Detested Disgust Dislike Embarrass Exclude Fail Fault Foolish Hate Idiotic Idle Lazy Lonesome Loser Pity Reject Revolted Selfish Shy Stared Stupid Teased Unkind Useless Wrong

Admire Adored Approving Brave Brilliant Calm Charm Clever Darling Delight Enjoy Excellent Fond Freedom Fun Gentle Gifted Happiness Helpful Jolly Joyful Laugh Lovely Lucky Nice Paradise Peaceful Pleasure Praise Rejoice Relax Smart Smile Super Win Wonderful

Window Bathing Address Doorknob Painted Bookcase Flannel Wiring Broom Whisk Guest Bell Dining Sponge Hallway Grill Toilet Tap Scales Bedroom Pane Tiles Cupboard Shower Washing Frying Bowl Freezer Sweep Corkscrew Spoon Sideboard Wallpaper Banister Beaker Roofing

Glass Laundry Furnish Frame Storage Aerial Cutlery Board Bathroom Repair Ironing Teaspoon Dresser Cleaner Armchair Ornament Rail Plate Candles Stir Shampoo Wipe Lock Wardrobe Tapes Tray Boiler Resident Drawers Rug Mirror Pillow Blinds Socket Curtain Radio

Neutral–neutral

Poster Lounge Hairbrush Fence Bedspread Fork Stool Garage Chimney Cabinet Cloth Fireplace Oven Kitchen Gas Sheets Drapes Insurance Blanket Attic Teapot Clock Dishes Stove Desk Radiator Upstairs Interior Cellar Toaster Drain Shelf Cover Photo Cup Building

Bath Bleach Bricks Brushing Carpet Clean Comb Contain Cook Decorated Dusted Floor Furniture Grater Heating Housework Jug Keys Lighting Loft Mugs Neighbour Newspaper Occupy Pan Pegs Picture Pipe Polished Rack Saucepan Sink Stair Towels Vase Water

Soap Cooker Kettle Decorate Mopped Chair Rent Staircase Dust Household Plants Steps Magazine Vacuum Landing Lightbulb Mat Nail Doorbell Sofa Lamp Microwave Telephone Rented Tin Sill Surface Room Basement Plug Lavatory Hook Table Heater Tidy House

ARTICLE IN PRESS C. Hunt et al. / Behaviour Research and Therapy 45 (2007) 497–509

501

stimuli pertaining to their concerns (symptoms of anxiety), or whether this bias would generalise to any threatening or emotional stimuli. The three sets of words were comparable in terms of word length, word frequency, and emotionality of words. An additional 36 neutral word pairs, with each pair matched for length and frequency, were generated to act as filler items. In order to ensure that stimuli were appropriate for the age group taking part in the current study, the words used here were drawn from an initial pool of words that had been shown to children of the same age group at a different school. Following Vasey et al. (1995), words were selected for use in the current study if their meaning was known by 75% of the sample. Dot-probe task Attentional bias was assessed using the dot-probe task and based on the protocol described by Hunt et al. (in press). The task was custom-made and written by technicians at Goldsmiths College using Visual Basic. For the unmasked trials, each trial started with a white fixation point that was displayed in the middle of a computer monitor for 500 ms. A word pair was then presented for 1000 ms, with one word above and one word below the fixation point. This presentation time is in line with previous research that has examined attentional biases to word stimuli in adult and child studies, which can range between 500 and 1500 ms (e.g., Hunt et al., in press; Keogh et al., 2001; Taghavi et al., 1999; Vasey et al., 1996). Immediately after its offset, the probe (a small asterisk) appeared in the position of one of the words and remained on the screen until the participant indicated whether its location was in the upper or lower half of the screen. Timing of response latencies commenced at the onset of probe presentation and terminated when either response key was pressed or after 3000 ms. Following the participant’s response, there was an interval of 500 ms before the commencement of the next trial. Both the emotional word and the probe could appear in either the upper or lower position with equal probability. Thus, the combination of these two independent variables yielded four possible conditions, and within each of the three word-pair types these four conditions occurred an equal number of times. For the masked condition a similar procedure was followed. The main difference was the stimulus onset asynchrony (SOA) between presentation of the word pair and probe. In the unmasked condition the SOA was 1000 ms; whereas to restrict awareness of the word stimuli in the masked condition, word pairs were initially presented for 14 ms and then replaced with a pattern mask pair (consisting of random character strings e.g., &#?£%!$) for 986 ms. Awareness check To determine whether participants were aware of masked stimuli, an awareness check was administered before and after the masked task. Each trial started with a white fixation cross displayed in the middle of the screen for 500 ms. A word pair or non-word pair was presented for 14 ms and then replaced with the mask pair. Participants responded following the offset of the mask and the next trial began 500 ms after either response key was pressed. The stimuli used in the awareness check were the same as those we used in our examination of AS and preconscious processing in adults (Hunt et al., in press). These consisted of 24 neutral–neutral word pairs and 24 non-word pairs consisting of random letter strings (e.g., trjnglsa). Each word pair and non-word pair was matched for length. Word and non-word pairs were replaced by masks consisting of a pair of random character strings (e.g., &#?£%!$) that were matched in length to the preceding words. Questionnaire measures The questionnaires that were completed by participants were the CASI (Silverman et al., 1991), the Revised Children’s Manifest Anxiety Scale (RCMAS; Reynolds & Richmond, 1978), the children’s Depression SelfRating Scale (DSRS; Birleson, 1981), and the basic reading subtest of the Wechsler Reading Test (WRT; Wechsler, 1991). The CASI is an 18-item self-report scale that assesses fear of anxiety symptoms in children. They rate the extent to which each item applies to them on a 3-point Likert scale: ‘never’ (1), ‘sometimes’ (2), and

ARTICLE IN PRESS 502

C. Hunt et al. / Behaviour Research and Therapy 45 (2007) 497–509

‘most of the time’ (3). Factor-analytic studies suggest that the CASI has sub-scales pertaining to physical, social, and mental incapacitation concerns. Scores for the three sub-scales were calculated by summation of the relevant items as obtained in a previous factor-analytic investigation of the CASI (Hunt et al., in prep.). The RCMAS was included to ensure that any effects found were not attributable to general trait anxiety. The RCMAS is a 37-item self-report scale. Twenty-eight of these items assess the presence of cognitive, behavioural, or affective symptoms of anxiety and the other nine are intended to measure social desirability. This measure has a dichotomous scale (yes/no) and is scored by summation of ‘yes’ responses. The social desirability scale is potentially useful in studies utilising self-report measures of anxiety since high scores on this scale are indicative of a tendency to under-report symptoms of anxiety. The DSRS was included to ensure that any effects found were not attributable to depression. The DSRS is an 18-item self-report scale assessing the presence of mood disturbance. Children are asked to rate the extent to which each item has applied to him or her over the previous week. Responses are made on a 3-point scale: ‘never’, ‘sometimes’, and ‘most of the time’. At the request of the College Ethics Committee, one item (I think life isn’t worth living) was excluded. Finally, the reading sub-scale of the WRT was also administered to ensure that any effects found could not be accounted for by differences in reading ability.

Procedure Prior to testing, letters were circulated informing parents of the aim of the study, outlining that participation was voluntary, and describing procedures to ensure confidentiality of participants. Parental consent was indicated by return of a signed form to the school. Child assent was obtained by explaining prior to both initial screening and testing that participation was voluntary and that they could withdraw at any time. Screening took place 1 week before the administration of the dot-probe task and other questionnaires. This took place in a classroom setting with the experimenter (CH) reading the CASI items out aloud as the children read along and completed the questionnaire. Where necessary, assistance in completing the questionnaires was given by either the teacher or classroom assistant. It was stressed that there were no right or wrong responses and children were encouraged to ask questions if they were unsure of anything. Participants completed the dot-probe task individually in a quiet room seated at a distance of about 60 cm from the computer screen. For those in the unmasked condition, the experimenter told each participant that they should silently read the words that appeared on the screen and then indicate whether the small star that subsequently appeared was in the upper or lower half of the screen by respectively pressing the ‘T’ or ‘B’ key on a standard keyboard using fingers from each hand. We used this approach as we have had success using it in the past. Although adult studies do not direct participant’s attention to the fact that there are words on the screen, we felt that such instructions were required to combat potential problems associated with slower (automatic) reading speed in children. This important methodological point is returned to within the discussion. Participants completed a practice consisting of four trials. The experimental trials consisted of the presentation of 144 word pairs that were presented in a new fully randomised order for each participant. Halfway through the experimental trials participants received a short break. Upon completion of the dot-probe task the experimenter administered the WRT, and then participants completed the remaining questionnaires (RCMAS, DSRS) with the experimenter assisting if necessary. For those in the masked condition, a similar protocol was followed. Participants were told that they would probably not be able to read the content of the words that they would be shown, as they would be presented very quickly. Immediately prior to and after the dot-probe task, participants also completed the awareness check. For this they were instructed to guess whether each pair of masked letter strings consisted of real words or non-words by respectively pressing the ‘Y’ or ‘N’ key on a standard keyboard using fingers from each hand. Participants then completed a practice consisting of four trials. The awareness check trials consisted of the presentation of 48 masked pairs that were presented in a new fully randomised order for each participant.

ARTICLE IN PRESS C. Hunt et al. / Behaviour Research and Therapy 45 (2007) 497–509

503

Results Participant characteristics A summary of participant characteristics is shown in Table 2. Chi-square analysis revealed that the groups did not differ significantly in gender ratio. A series of ANOVAs were conducted with physical AS group (high vs. low) and presentation condition (masked vs. unmasked) as the independent variables and the CASI subscales (social AS and mental AS sub-scale scores), RCMAS sub-scales, DSRS, WRT-Reading, and age as the dependent variables. Physical AS groups were found to differ with regards to scores on the CASI: social subscale (F (1, 75) ¼ 5.70, po.05, partial Z2 ¼ .071) and CASI mental sub-scale (F (1, 75) ¼ 16.53, po.001, partial Z2 ¼ .181). Of particular importance to the current investigation was the finding that the high physical AS group also had significantly higher RCMAS trait anxiety scores than the low physical AS group (F (1, 75) ¼ 16.78, po.001, partial Z2 ¼ .183). With respect to presentation condition, main effects were found for mental AS (F (1, 75) ¼ 4.47, po.05, partial Z2 ¼ .056) and reading ability (F (1, 75) ¼ 39.99, po.001, partial Z2 ¼ .348). No other main or interaction effects were found. The fact that there are such differences between the two presentation condition groups is a concern, due to an unintended sampling bias. This issue will be considered further within the regression analysis below. Awareness check In both awareness checks, the percentage of trials with correct responses was calculated for each participant. Binomial tests revealed that one participant’s ability to detect the lexical content of words was significantly above chance level in both awareness checks (70.8% and 75.0%, respectively) and thus the data of this individual were excluded from any further analyses. For the remaining participants, the performance of each fell within the 95% confidence limits of chance level of performance (awareness 1: m ¼ 49.3% correct, awareness 2: m ¼ 51.2% correct). There were no significant differences between the physical AS groups with regards to percentage correct scores in either awareness task. Taken together this suggests that the efficacy of the masking procedure was validated and that performance on this task was not moderated by level of physical AS. Selective attentional bias The data for each participant were initially screened for outliers (i.e., response times of 200 ms or less, or 1500 ms or more) and incorrect responses. Where this made up over 20% of that individual’s possible Table 2 Means of participants’ characteristics by physical AS (high vs. low) group and presentation condition (unmasked vs. masked) Unmasked

CASI: Physical CASI: Social CASI: Mental CASI: Total DSRS RCMAS: Anxiety RCMAS: Lie Scale Reading ability Age (in months)

Masked

Low AS (N ¼ 16)

High AS (N ¼ 23)

Low AS (N ¼ 19)

High AS (N ¼ 20)

12.75 7.44 5.44 25.63 7.50 10.69 3.13 44.69 112.81

22.48 8.57 6.96 38.00 9.13 15.22 3.96 42.13 113.30

11.89 7.74 4.42 24.05 8.37 8.58 4.11 53.32 114.05

21.05 8.65 6.20 35.90 9.80 13.45 4.00 53.50 114.25

(1.06) (2.03) (1.31) (3.30) (3.85) (4.13) (2.16) (9.03) (6.34)

(2.33) (1.65) (2.20) (3.53) (3.55) (4.66) (2.20) (4.50) (5.88)

(1.49) (1.66) (.84) (2.37) (3.83) (6.02) (2.77) (8.22) (8.50)

(2.04) (1.98) (1.88) (4.48) (4.27) (5.50) (2.99) (6.19) (8.93)

Standard deviations in parentheses. Note: CASI ¼ Children’s Anxiety Sensitivity Index; DSRS ¼ children’s Depression Self-Rating Scale; RCMAS ¼ Revised Children’s Manifest Anxiety Scale; reading ability ¼ basic reading subtest of the Wechsler Reading Test.

ARTICLE IN PRESS C. Hunt et al. / Behaviour Research and Therapy 45 (2007) 497–509

504

responses, they were excluded from all further analyses. For the remaining participants, outliers and incorrect scores were simply removed. As mentioned above five participants in the unmasked condition had an excessive number of incorrect responses or outliers, and thus were excluded from analyses. Following these exclusions, the amount of data lost due to errors and outliers among the remaining participants was 5% of the data. In the masked condition four participants were excluded due to having an excessive number of incorrect responses or outliers, and one excluded due to having above-chance performance on the awareness task. Following these exclusions, the amount of data lost due to errors and outliers among the remaining participants was 4% of the data. Due to an input error one of the mean RTs for one participant was missing, and so was replaced with the whole sample mean for that variable. There were no major differences in results with this participant’s data excluded or included, and so they were left in. Mean reaction times were calculated for each participant and attentional bias indices were calculated for each type of emotional word (see Table 3) using a procedure commonly used in research utilising the dotprobe paradigm (e.g., MacLeod & Mathews, 1988). Thus, the following equation was used to create attentional bias indices: Index ¼ 12½ðeupl  elplÞ þ ðelpu  eupuÞ, where e is the emotional word, p the probe, u the upper position, and l the lower position. In essence, this subtracts the mean reaction time when the probe is in the same position as the emotional word from the mean reaction time when the probe is in a different position to the emotional word. A positive score on these indices indicates a selective attentional bias to that category of emotional words (vigilance), and a negative score indicates a selective attentional bias away from that category of emotional words (avoidance). A repeated-measures ANOVA was conducted with physical AS group (high vs. low) and presentation condition (unmasked vs. masked) as the between-groups factors and bias index (anxiety symptomatology vs. social threat vs. positive) as the within-groups factors. There was a significant main effect of physical AS group (F (1, 74) ¼ 8.17, po.01, partial Z2 ¼ .099), with the high physical AS group exhibiting vigilance for all emotional words (mean ¼ 8.27) relative to the low physical AS group (mean ¼ 11.24). Also, a main effect was found for presentation condition (F ¼ (1, 74) ¼ 6.53, po.05, partial Z2 ¼ .081), with greater emotional vigilance being exhibited in the unmasked condition (mean ¼ 7.23) than in the masked condition (mean ¼ 10.21). No other significant effects were found. A near significant interaction was observed between physical AS group and valence (F(2, 148) ¼ 3.03, po.06, partial Z2 ¼ .039). Although this does not reach the conventional level of acceptable significance, we examined the means in Table 3, which suggested that within the low AS group there is greater avoidance of social threat words, whereas the high AS group display greater vigilance to positive words. However, due to lack of overall significance little can be reliably concluded from this result, and so is not considered further. We also examine whether the attention bias index scores were significantly different from zero using one sample t-tests, but none were, either generally, or when conducted separately by AS and presentation groups (at a corrected .01 level). Table 3 Mean emotional bias index (ms) by physical AS group and presentation condition Unmasked

Anxiety index Social threat index Positive index

Masked

Low AS (N ¼ 16)

High AS (N ¼ 23)

Low AS (N ¼ 19)

High AS (N ¼ 20)

.61 (28.81) 22.41 (39.62) 2.06 (67.57)

11.71 (42.43) 15.16 (40.13) 36.24 (65.81)

11.31 (46.04) 30.43 (66.44) 28.59 (50.91)

11.94 (65.15) 13.66 (79.19) 12.08 (47.67)

Standard deviations in parentheses. Note: AS ¼ Anxiety Sensitivity.

ARTICLE IN PRESS C. Hunt et al. / Behaviour Research and Therapy 45 (2007) 497–509

505

Correlation and regression analysis In order to establish that the pattern of results reported above are attributable to physical AS rather than any of the other questionnaire measures, a series of correlations were calculated. Given that the ANOVA indicated a general vigilance for all emotional words in the high physical AS group, a general emotional bias index was created by calculating an average bias index, which was then correlated with the questionnaire measures (see Table 4). In the unmasked condition, there were significant correlations between the emotional bias index and both the total, physical concerns and mental concerns components of the CASI. For the masked condition, there were no significant correlations found, although there was a near-significant association between the physical concerns scale and the emotional bias index in the predicted direction (p ¼ .067). When the bias index was collapsed across presentation groups, a similar pattern to that found in the unmasked condition was found, although with the addition of a significant negative relationship between the bias index and reading level. Since it is possible that differences in reading ability, anxiety and other AS sub-scales may have had a confounding influence on results, the final analysis took the form of a hierarchal regression, predicting variance in the general emotional bias index. At the first step, the social and mental concerns sub-scales of the CASI, anxiety and reading scores were entered. At the second step presentation condition and physical AS scores were entered, and at the third step an interaction term between physical AS and presentation condition was entered (Holmbeck, 1997). As can be seen from Table 5, the only significant predictor of the emotional bias index was the physical concerns scale of the CASI at the second step. Discussion The results of the experiment reported here found that children display anxiety sensitivity (AS)-related biases in response to emotional stimuli presented above and below the threshold of conscious identification. These results suggest the presence of attentional biases in children who are specifically fearful of anxietyrelated sensations, as well as highlighting the utility of employing objective measures of information processing such as the dot-probe task in examining the attentional processing of children. With respect to how our results compare with adult studies (Hunt et al., in press; Keogh et al., 2001; Koven et al., 2003; Stewart et al., 1998), the biases reported in adults tend to be specific to words pertaining to anxiety symptomatology rather than extending to include all emotional words. Previous reports of anxiety-related (i.e., non-AS) attentional biases in childhood samples also report specific threat-related biases. For instance, Vasey et al. (1995) found that clinically anxious children aged between 9 and 14 were characterised by a vigilance for threat words, whereas the pattern of response latencies evinced by their non-anxious counterparts was not a function of word type. Similar results have been obtained on the dot-probe paradigm with testanxious and trait-anxious children (e.g., Schippell et al., 2003; Vasey et al., 1996), and on the Stroop task with spider-fearful children (Kindt & Brosschot, 1999; Martin et al., 1992). However, these studies tend not to include stimuli that are positive, or that pertain to threat that is not specific to the individual’s predominant concerns, which limits any direct comparisons that can be made. Specifically, the attentional biases displayed in the current study are indicative of a hyper-responsivity to all emotional stimuli, whereas it is not clear if the attentional biases in the aforementioned studies are indicative of content-specific biases, general threat biases, or general emotion biases. Table 4 Correlations between questionnaire measures and emotional bias index for the different presentation groups (unmasked vs. masked)

Combined Unmasked Masked

CASI: Total

CASI: Physical

CASI: Social

CASI: Mental

Depression

Anxiety

Lie Scale

Reading

.397** .482** .221

.408** .441** .296

.098 .234 .032

.283* .326* .078

.026 .008 .103

.202 .314 .005

.148 .265 .117

.248* .188 .025

Note: *p ¼ .05; **po.01; CASI ¼ Children’s Anxiety Sensitivity Index; combined ¼ emotional bias index collapsed across masked and unmasked conditions.

ARTICLE IN PRESS C. Hunt et al. / Behaviour Research and Therapy 45 (2007) 497–509

506

Table 5 Results of regression analysis on the emotional bias index Step

Variables

Beta

t

DR2

Total R2

F

1

Social AS Mental AS Reading ability Anxiety Physical AS Condition Physical AS  condition

.039 .213 .195 .035 .371 .189 .071

.330 1.654 1.676 .267 2.821** 1.435 .527

.120

.120

2.480

.111**

.231

3.561**

.007

.238

3.124**

2 3

Note: *p ¼ .05; **po.01. AS ¼ anxiety sensitivity; condition ¼ presentation condition (1 ¼ unmasked, 2 ¼ masked).

Prior to the present study, there have been only been a few published studies that have included different classes of emotional stimuli in the examination of attentional bias to word stimuli in anxious children (Dalgleish et al., 2001; Moradi et al., 1999; Taghavi et al., 1999). For instance, Moradi et al. (1999) compared the colour-naming interference of 9- to 17-year-olds with a diagnosis of post-traumatic stress disorder (PTSD) with that of controls in response to words that were trauma-related, threat-related, depression-related, positive, and neutral. Analyses revealed evidence of content-specific biases, in that the PTSD patients evidenced greater colour-naming interference for trauma-related words relative to neutral words, whereas the response latencies of controls were not differentially affected by word type. However, an important difference is that the results from these previous studies were obtained in clinical samples containing adolescents as well as children and so it is possible that the specificity of anxiety-related attentional biases are a function of variables not systematically varied in the current study including clinical status and age. As mentioned within the introduction, there is a suggestion that age-related effects between the ages of 8 and 12 may account for some of the variability in effects found more generally within studies of childhood anxiety. It could be argued that the generalised biases to any emotional stimuli become more specific to the source of fear through increased encounters with the feared stimuli and/or with age. The current study represents one of the first known investigations into the preconscious processing of threat in anxiety-sensitive children. Since no differences were found between the masked and unmasked conditions with respect to the AS-related bias, this suggests that the effects are likely to occur at an automatic level. Although interesting, presentation condition was a between-groups factor and so we did not directly compare attentional performance within the same individuals. It is therefore possible that had a within-groups design been employed, a different pattern of effects may have been found. It is also possible that the presentation interval of stimuli (14 ms) that has been used to demonstrate the existence of automatic processing biases in adults (e.g., Mogg et al., 1995) is inappropriate for use with children. Given that there is great variation in reading ability within and across children of different age groups, then it is possible that presentation intervals might have to be individually tailored for each child (Vasey, Dalgleish, & Silverman, 2003). One feature of the current study was to determine whether any observed attentional biases could be attributable to the other Childhood Anxiety Sensitivity Index (CASI) sub-scales or to other related constructs such as anxiety and depression. Although the two physical AS groups differed with regard to trait anxiety, there was no evidence to suggest that the vigilance for emotional words in the high physical AS group was attributable to trait anxiety. Although the regression analysis showed that the physical AS-related general emotional bias occurs even when controlling for the CASI items pertaining to social and mental incapacitation concerns, this result must be treated with some caution as we did not examine total CASI scores. Unfortunately, since we selected children on the basis of their physical concerns score we were not able to examine general AS effects. Future research should consider whether the biases found here are specific to the physical concerns component of childhood AS or whether they may be related to general AS. Methodological differences between the current and previous studies should also be considered when interpreting these results. For example, in the unmasked trials, although we used a word stimulus presentation time which is in line with previous research that has examined attentional biases to word stimuli, child dotprobe studies have tended to use longer presentation times (e.g., 1500 ms, Taghavi et al., 1999). We also

ARTICLE IN PRESS C. Hunt et al. / Behaviour Research and Therapy 45 (2007) 497–509

507

instructed our participants to read the stimulus words (in the unmasked condition) and so directed attention, which may in turn seem generally inconsistent with the notion that threat stimuli selectively capture attention. Indeed, previous research tended not to focus children’s attention on the word stimuli and so it is possible that this had a confounding influence. Future research is required to determine whether these methodological issues are important or not. An additional methodological concern relates to our assessment of anxiety and depression. Indeed, we did not include a standard assessment of psychopathology (e.g., structured interview) and instead relied on selfreport measures. This could be problematic for two reasons. First, the measures we used to assess anxiety and depression could be questioned, as there may be better measures available. While this may be a concern, given that the focus of the current study was on AS rather than general anxiety or depression, we do not feel it detracts too much from the main findings of the current study. A second measurement issue relates to our assumption that CASI scores represent a non-clinical risk measure. This assumption was based on the theoretically conceived notion that AS is a non-clinical personality trait that serves as a vulnerability factor in the development of clinical conditions such as panic. However, some may view the CASI as a measure of psychopathology, and so it is possible that some of our samples comprised of clinically anxious children. In light of this we should, therefore, be careful in drawing too many definitive conclusions until this issue has been resolved. With respect to future research, it would be particularly interesting to examine whether the effects found here vary across different age groups. No study that we are aware of has investigated age-related effects in the attentional biases found within children high and low in AS. It would also be interesting to examine the potential moderating effect of sex on the effects found here. Not only have there been sex differences reported in both adults and children with respect to AS, but also important AS by sex interactions have been reported in adult studies on information processing biases and related responses to physical challenges (Stewart et al., 1998; Keogh & Birkby, 1999; Keogh, Hamid, Hamid, & Ellery, 2004). In sum, using an objective measure of selective attention, we demonstrated that beliefs about the harmfulness of bodily sensations associated with anxiety influence the deployment of attention to affective stimuli in children. The results of this study add further support to the proposition that AS is associated with attentional biases and that these may be present in children as young as 8. The general emotionality bias observed here was found for stimuli presented above and below the threshold of conscious identification, which suggests that children may be able to process briefly presented stimuli in a relatively automatic manner. Acknowledgements The authors would like to thank Ian Hannent for writing the software used in the current study, as well as the reviewers for their very helpful comments and suggestions. References Asmundson, G. J. G., Norton, G. R., Wilson, K. G., & Sandler, L. S. (1994). Subjective symptoms and cardiac reactivity to brief hyperventilation in individuals with high anxiety sensitivity. Behaviour Research and Therapy, 32, 237–241. Birleson, P. (1981). The validity of depressive disorder in childhood and the development of a self-rating scale: A research report. Journal of Child Psychology and Psychiatry, 22, 73–88. Cox, B. J., Parker, J. D. A., & Swinson, R. P. (1996). Anxiety sensitivity: Confirmatory evidence for a multidimensional construct. Behaviour Research and Therapy, 34, 591–598. Daleiden, E. L., & Vasey, M. W. (1997). An information-processing perspective on childhood anxiety. Clinical Psychology Review, 17, 407–429. Dalgleish, T., Moradi, A. R., Taghavi, M. R., Neshat-Doost, H. T., & Yule, W. (2001). An experimental investigation of hypervigilance for threat in children and adolescents with post-traumatic stress disorder. Psychological Medicine, 31, 541–547. Dehon, C., Weems, C. F., Stickle, T. R., Costa, N. A., & Berman, S. L. (2005). A cross-sectional evaluation of the factorial invariance of anxiety sensitivity in adolescents and young adults. Behaviour Research and Therapy, 43, 799–810. Eley, T. C., Stirling, L., Ehlers, A., Gregory, A. M., & Clark, D. M. (2004). Heartbeat perception, panic/somatic symptoms and anxiety sensitivity in children. Behaviour Research and Therapy, 40, 439–448. Hayward, C., Killen, J. D., Kraemer, H. C., & Taylor, C. (2000). Predictors of panic attacks in adolescents. Journal of the American Academy of Child Adolescent Psychiatry, 39, 207–214.

ARTICLE IN PRESS 508

C. Hunt et al. / Behaviour Research and Therapy 45 (2007) 497–509

Holmbeck, G. N. (1997). Toward terminological, conceptual, and statistical clarity in the study of mediators and moderators: Examples from the child-clinical and pediatric psychology literatures. Journal of Consulting and Clinical Psychology, 65, 599–610. Hunt, C., Keogh, E., & French, C. C. (in press). Anxiety sensitivity, the role of conscious awareness and selective attentional bias to physical threat. Emotion. Hunt, C., Keogh, E., & French, C. C. (in prep.). Factor structure of the Childhood Anxiety Sensitivity Index: Gender and age differences in dimensions of anxiety sensitivity. Kearney, C. A., Albano, A. M., Eisen, A. R., Allan, W. D., & Barlow, D. H. (1997). The phenomenology of panic disorders in youngsters: An empirical study of a clinical sample. Journal of Anxiety Disorders, 11, 49–62. Keogh, E. (2004). Investigating invariance in the factorial structure of the Anxiety Sensitivity Index across adult men and women. Journal of Personality Assessment, 83, 153–160. Keogh, E., & Birkby, J. (1999). The effect of anxiety sensitivity and gender on the experience of pain. Cognition and Emotion, 13, 813–829. Keogh, E., Dillon, C., Georgiou, G., & Hunt, C. (2001). Selective attentional biases for physical threat in physical anxiety sensitivity. Journal of Anxiety Disorders, 15, 299–315. Keogh, E., Hamid, R., Hamid, S., & Ellery, D. (2004). Investigating the effect of anxiety sensitivity, gender and negative interpretative bias on the perception of chest pain. Pain, 111, 209–217. Kindt, M., Bierman, D., & Brosschot, J. F. (1997). Cognitive bias in spider fear and control children: Assessment of emotional interference by a card format and a single-trial format of the Stroop task. Journal of Experimental Child Psychology, 66, 163–179. Kindt, M., & Brosschot, J. F. (1999). Cognitive bias in spider-phobic children: Comparison of a pictorial and a linguistic spider Stroop. Journal of Psychopathology and Behavioural Assessment, 21, 207–220. Kindt, M., Brosschot, J. F., & Everaerd, W. (1997). Cognitive processing bias of children in a real life stress situation and a neutral situation. Journal of Experimental Child Psychology, 64, 79–97. Kindt, M., & van den Hout, M. (2001). Selective attention and anxiety: A perspective on developmental issues and the causal status. Journal of Psychopathology and Behavioral Assessment, 23, 193–202. Kindt, M., van den Hout, M., de Jong, P., & Hoekzema, B. (2000). Cognitive bias for pictorial and linguistic threat cues in children. Journal of Psychopathology and Behavioral Assessment, 22, 201–219. Koven, N. S., Heller, W., Banich, M. T., & Miller, G. A. (2003). Relationships of distinctive affective dimensions to performance on an emotional Stroop task. Cognitive Therapy and Research, 27, 671–680. Lau, J. J., Calamari, J. E., & Waraczynski, M. (1996). Panic attack symptomatology and anxiety sensitivity in adolescents. Journal of Anxiety Disorders, 10, 355–364. MacLeod, C., & Mathews, A. (1988). Anxiety and the allocation of attention to threat. Quarterly Journal of Experimental Psychology, 40A, 653–670. Mannuzza, S., Klein, R. G., Moulton, J. L., Scarfone, N., Malloy, P., Vosburg, S. K., et al. (2002). Anxiety sensitivity among children of parents with anxiety disorders: A controlled high-risk study. Journal of Anxiety Disorders, 16, 135–148. Martin, M., Horder, P., & Jones, G. V. (1992). Integral bias in naming of phobia-related words. Cognition and Emotion, 6, 479–486. McNally, R. J., Kaspi, S. P., Riemann, B. C., & Zeitlin, S. B. (1990). Selective processing of threat cues in post-traumatic stress disorder. Journal of Abnormal Psychology, 99, 398–402. Mogg, K., Bradley, B. P., & Williams, R. (1995). Attentional bias in anxiety and depression: The role of awareness. British Journal of Clinical Psychology, 34, 17–36. Mogg, K., & Marden, B. (1990). Processing of emotional information in anxious subjects. British Journal of Clinical Psychology, 29, 227–229. Mogg, K., Mathews, A., Bird, C., & Macgregor-Morris, R. (1990). Effects of stress and anxiety on the processing of threat stimuli. Journal of Personality and Social Psychology, 59, 1230–1237. Moradi, A. R., Taghavi, M. R., Neshat-Doost, H. T., Yule, W., & Dalgleish, T. (1999). Performance of children and adolescents with PTSD on the Stroop colour-naming task. Psychological Medicine, 29, 415–419. Muris, P. (2002). An expanded childhood anxiety sensitivity index: Its factor structure, reliability, and validity in a non-clinical adolescent sample. Behaviour Research and Therapy, 40, 299–311. Rabian, B., Embry, L., & MacIntyre, D. (1999). Behavioural validation of the Childhood Anxiety Sensitivity Index in children. Journal of Clinical Child Psychology, 28, 105–112. Reiss, S. (1991). Expectancy model of fear, anxiety and panic. Clinical Psychology Review, 11, 141–153. Reiss, S., & McNally, R. J. (1985). Expectancy model of fear. In S. Reiss, & R. R. Bootzin (Eds.), Theoretical issues in behaviour therapy (pp. 107–121). New York: Academic Press. Reiss, S., Peterson, R. A., Gursky, D. M., & McNally, R. J. (1986). Anxiety sensitivity, anxiety frequency and the predictions of fearfulness. Behaviour Research and Therapy, 24, 1–8. Reynolds, C. R., & Richmond, B. O. (1978). What I think and feel. A revised measure of children’s manifest anxiety. Journal of Abnormal Child Psychology, 6, 271–280. Scher, C. D., & Stein, M. B. (2003). Developmental antecedents of anxiety sensitivity. Journal of Anxiety Disorders, 17, 253–269. Schippell, P., Vasey, M. W., Cravens-Brown, L., & Bretveld, R. (2003). Attentional bias in reactive aggressive children. Journal of Clinical Child and Adolescent Psychology, 32, 40–55. Schmidt, N. B., Lerew, D. R., & Jackson, R. J. (1997). The role of anxiety sensitivity in the pathogenesis of panic: Prospective evaluation of spontaneous panic attacks during acute stress. Journal of Abnormal Psychology, 106, 355–364. Schmidt, N. B., Lerew, D. R., & Joiner, T. E. (1998). Anxiety sensitivity and the pathogenesis of anxiety and depression: Evidence for symptom specificity. Behaviour Research and Therapy, 36, 165–177.

ARTICLE IN PRESS C. Hunt et al. / Behaviour Research and Therapy 45 (2007) 497–509

509

Schmidt, N. B., Lerew, D. R., & Joiner, T. E. (2000). Prospective evaluation of the etiology of anxiety sensitivity: Test of a scar model. Behaviour Research and Therapy, 38, 1083–1095. Silverman, W. K., Fleisig, W., Rabian, B., & Peterson, R. A. (1991). Childhood Anxiety Sensitivity Index. Journal of Clinical Child Psychology, 20, 162–168. Silverman, W. K., Ginsburg, G. S., & Goedhart, A. W. (1999). Factor structure of the Childhood Anxiety Sensitivity Index. Behaviour Research and Therapy, 37, 903–917. Silverman, W. K., Goedhart, A. W., Barrett, P., & Turner, C. (2003). The facets of anxiety sensitivity represented in the Childhood Anxiety Sensitivity Index: Confirmatory analyses of factor models from past studies. Journal of Abnormal Psychology, 112, 364–374. Stewart, S. H., Conrod, P. J., Gignac, M. L., & Pihl, R. O. (1998). Selective processing biases in anxiety sensitive men and women. Cognition and Emotion, 12, 105–133. Stewart, S. H., Taylor, S., & Baker, J. M. (1997). Gender differences in dimensions of anxiety sensitivity. Journal of Anxiety Disorders, 11, 179–200. Sturges, L. V., Goetsch, V. L., Ridley, J., & Whittal, M. (1998). Anxiety sensitivity and response to hyperventilation challenge: Physiological arousal, interoceptive acuity, and subjective distress. Journal of Anxiety Disorders, 12, 103–115. Taghavi, M. R., Neshat-Doost, H. T., Moradi, A. R., Yule, W., & Dalgleish, T. (1999). Biases in visual attention in children and adolescents with clinical anxiety and mixed anxiety-depression. Journal of Abnormal Child Psychology, 27, 215–223. Tsao, J., Myers, C. D., Craske, M. G., Bursch, B., Kim, S. C., & Zeltzer, L. K. (2005). Parent and child anxiety sensitivity: Relationship in a nonclinical sample. Journal of Psychopathology and Behavioral Assessment, 27, 259–268. van Beek, N., Perna, G., Schruers, K., Muris, P., & Griez, E. (2005). Anxiety sensitivity in children of panic disorder patients. Child Psychiatry and Human Development, 35, 315–324. Vasey, M. W., Daleiden, E. L., Williams, L. L., & Brown, L. M. (1995). Biased attention in childhood anxiety disorders: A preliminary study. Journal of Abnormal Child Psychology, 23, 267–279. Vasey, M. W., Dalgleish, T., & Silverman, W. K. (2003). Research on information-processing factors in child and adolescent psychopathology: A critical commentary. Journal of Clinical Child and Adolescent Psychology, 32, 81–93. Vasey, M. W., El-Hag, N., & Daleiden, E. L. (1996). Anxiety and the processing of emotionally threatening stimuli: Distinctive patterns of selective attention among high- and low-test-anxious children. Child Development, 67, 1173–1185. Walsh, T. M., Stewart, S. H., McLaughlin, E., & Comeau, N. (2004). Gender differences in Childhood Anxiety Sensitivity Index (CASI) dimensions. Journal of Anxiety Disorders, 18, 695–706. Waters, A. M., Lipp, O. V., & Spence, S. H. (2004). Attentional bias toward fear-related stimuli: An investigation with nonselected children and adults and children with anxiety disorders. Journal of Experimental Child Psychology, 89, 320–337. Wechsler, D. (1991). Wechsler Intelligence Scale for Children (3rd ed.). The Psychological Group. Weems, C. F., Hayward, C., Killen, J., & Taylor, C. B. (2002). A longitudinal investigation of anxiety sensitivity in adolescence. Journal of Abnormal Psychology, 111, 471–477. Williams, J. M. G., Watts, F. N., MacLeod, C., & Mathews, A. (1997). Cognitive psychology and emotional disorders (2nd ed). Chichester: Wiley. Wilson, K. A., & Hayward, C. (2005). A prospective evaluation of agoraphobia and depression symptoms following panic attacks in a community sample of adolescents. Journal of Anxiety Disorders, 19, 87–103. Zinbarg, R. E., Barlow, D. H., & Brown, T. A. (1997). Hierarchical structure and general factor saturation of the Anxiety Sensitivity Index: Evidence and implications. Psychological Assessment, 9, 277–284.