Self‐focused attention and social anxiety

Abstract

Cognitive models propose that self‐focused attention (SFA) interacts with fear of negative evaluation to maintain social anxiety. Thus, the effect of SFA on anxiety would be expected to be specific to those with existing social concerns. However, much research suggests that the effect of SFA on anxiety occurs across anxiety levels. Manipulations of attention focus have been criticised for (1) lack of ecological validity and (2) eliciting fear of negative evaluation directly. The present study examined the role of SFA in social anxiety using an ecologically valid procedure that did not elicit fear of negative evaluation directly. Self‐reported anxiety was assessed among high and low socially anxious individuals under conditions of SFA or external‐focused attention. The manipulation successfully altered focus of attention but did not directly affect fear of negative evaluation or self‐reported anxiety. Taken together with the findings of previous studies, the results suggest that focusing on internal physiological states per se does not increase self‐reported social anxiety, and that self‐focus that does not have an explicitly evaluative dimension does not elicit social anxiety. The findings have implications for approaches to reducing social anxiety through reducing SFA.

Key words:

• external‐focused attention
• self‐focused attention
• social anxiety

SELF‐FOCUSED ATTENTION AND SOCIAL ANXIETY

Social phobia is a common and disabling condition, with lifetime prevalence estimates around 12% (Kessler et al., 2005) and significant impact on quality of life and service utilisation (Acarturk, de Graaf, van Straten, ten Have & Cuijpers, 2008). Evaluation from others is a central concern for those with social phobia (Weeks, Heimberg, Rodebaugh, & Norton, 2008). Self‐focused attention (SFA) refers to the tendency to direct attention to closely monitoring the self rather than to features of the environment. Cognitive models of social anxiety have suggested that focusing attention on one's own thoughts, behaviour, and signs of physiological arousal interacts with fear of negative evaluation to maintain social anxiety (e.g. Clark & Wells, 1995; Rapee & Heimberg, 1997). Reducing SFA has been proposed as a therapeutic strategy in social phobia (McManus, Sacadura, & Clark, 2008).

Cognitive models of social phobia imply that the effect of SFA on the experience of social anxiety should be specific to those with elevated social concerns. It is understood that the interaction of SFA and fear of negative evaluation gives rise to anxiety (Vassilopoulos, 2008; Zou, Hudson, & Rapee, 2007). This is consistent with the conclusions of Mor and Winquist (2002) who found a stronger association between SFA and negative affect among those with anxiety and depression conditions in their meta‐analysis. However, many studies have found that the effect of SFA on social anxiety is not specific to those with elevated social concerns, but that increasing SFA increases anxiety regardless of baseline social concerns (e.g. Bögels & Lamers, 2002; George & Stopa, 2008; McManus et al., 2008; Woody, 1996; Woody & Rodriguez, 2000). In addition, some studies have found that SFA has no effect on anxiety (Bögels, Rijsemus, & de Jong, 2002), and others have found that SFA is associated with reductions in anxiety (Vassilopoulos, 2008).

Explaining the apparently contradictory findings concerning SFA and anxiety requires a closer analysis of the experimental manipulations used to elicit SFA. Some manipulations of focus of attention bear little resemblance to social interactions. For example, Bögels and Lamers (2002) manipulated focus of attention using a task in which participants read a series of hypothetical scripts. Similarly, Vassilopoulos (2008) asked students selected for high and low scores on the Fear of Negative Evaluation Questionnaire (Watson & Friend, 1969) to focus on either the sensory experience (experiential SFA) or the meaning (analytical SFA) of a series of written statements. The ecological validity of procedures such as these, and therefore the generalisability of findings to more realistic social situations, has been questioned.

The extent to which more ecologically valid manipulations of SFA are confounded by directly arousing concerns about negative evaluation, rather than interacting with fear of negative evaluation, has been raised (Bögels & Mansell, 2004). Bögels and Mansell argued that any manipulation that explicitly or implicitly included a suggestion that performance would be evaluated, such as video recording performance, was open to this criticism. These authors suggested that using a mirror positioned so that participants viewed their own reflections during social interactions was both ecologically valid and unlikely to arouse evaluation concerns directly. However, Bögels et al. (2002) used a mirror to induce SFA during social interactions and found that while the mirror increased self‐awareness, it was not associated with anxiety. Of the studies they reviewed in 2004, Bögels and Mansell concluded that only Bögels and Lamers (2002) could be regarded as manipulating SFA in a way that (1) did not directly elicit evaluation concerns and (2) demonstrated a relationship between SFA and social anxiety. However, as noted above, the ecological validity of the SFA manipulation used in that study is questionable.

More recently, Zou et al. (2007) used a manipulation of SFA in which high and low blushing‐anxious participants engaged in a brief conversation with a female conversational partner. Those in the SFA condition were instructed to focus their attention on their own breathing, heart rate, voice, and signs of blushing. High blushing‐anxious participants reported more anxiety following the SFA manipulation. Zou et al. recognised that the SFA instructions may have elicited fear of negative evaluation directly, as those with anxiety about blushing may have thought that their conversation partner was aware of their blushing and was evaluating them in light of this. Thus, this finding is open to the same alternative explanations as earlier work. George and Stopa (2008), however, reported that a mirror manipulation increased both self awareness and anxiety. It is not clear why these findings differ from those of the methodologically similar work of Bögels et al. (2002).

Clark and Wells (1995) argued that the effectiveness of psychological treatment for social phobia would be enhanced by basing therapy on the cognitive processes involved in the maintenance of the condition. Thus, from their model, it would be expected that interventions aimed at reducing both SFA and fear of negative evaluation would reduce the experience of social anxiety (Wells & Papergeorgiou, 2001). Voncken, Dijk, de Jong, and Roelofs (2010) experimentally manipulated state anxiety among people high and low in fear of blushing during a 5‐minute video recorded conversation with two confederates by giving false feedback to half the group that they were blushing continuously during the interaction. For both high and low blushing‐fearful groups, false feedback about blushing elicited higher ratings of SFA and negative beliefs and lower likeability and social performance ratings. For the high blushing‐fearful group, a relationship between social anxiety and social performance was found, and this was mediated by negative beliefs but not by SFA. Voncken et al. argued that SFA may be a by‐product of social anxiety, and that clinical interventions to reduce social anxiety should focus on negative beliefs and poor social performance. Voncken et al.'s study further illustrates the methodological challenges for researchers investigating anxiety, SFA, and evaluative concerns. Their manipulation was designed to elicit anxiety; but as it directed attention to physiological signs of blushing, it is likely to have also elicited SFA directly.

Despite these methodological challenges, establishing clearly whether SFA is an independent factor mediating social anxiety has considerable theoretical and practical significance. The present study aimed to investigate the role of SFA in social anxiety using an ecologically valid task designed to manipulate direction of attention without directly eliciting fear of negative evaluation on self‐reported anxiety among analogue samples high and low in social anxiety. Previous literature suggests three possible models of the relationship between SFA and social anxiety:

1	If SFA elicits social anxiety regardless of level of ongoing social concerns (e.g., Bögels & Lamers, 2002; George & Stopa, 2008; McManus et al., 2008; Woody, 1996; Woody & Rodriguez, 2000), then a main effect of attention is expected, where participants in the SFA condition report higher anxiety during the interaction compared with those in the external‐focused attention (EFA) condition regardless of level of social anxiety.
2	If the interaction of SFA and concerns about negative evaluation gives rise to social anxiety (Zou et al., 2007), then an interaction between direction of attention and level of social anxiety would be expected here. For those with high social anxiety, significantly higher self‐reported anxiety would be expected in the SFA condition compared with the EFA condition. For those with low social anxiety, no differences in self‐reported anxiety would be expected between attention focus conditions.
3	If findings that SFA elicits social anxiety reported in the literature are a result of a confound whereby the SFA manipulation also elicits concerns about negative evaluation (Bögels & Mansell, 2004), then no impact of SFA on social anxiety would be expected in the present study, where the SFA manipulation is designed to specifically target attention focus and not to directly arouse concerns about negative evaluation.

METHOD

Design

A 2 (anxiety group: high social anxiety; low social anxiety) × 2 (attention condition: SFA; EFA) between‐subjects design was used. The dependent variables were self‐reported anxiety measured on a visual analogue scale (VAS) immediately before and after the attention manipulation. Pilot testing with a sample of undergraduate students (n = 6) who were not included in the present study confirmed that the SFA and EFA manipulations were successful in altering direction of attention.

Several measures were included to control for alternative explanations of findings. These were completed after the attention manipulation task, and participants were asked to rate their experience during the task. First, VASs assessing attention focus were used to verify that the attentional focus manipulations were effective.

Second, as differences in familiarity with conversation topics may be related to anxiety, participants were asked to rate ‘how familiar you were with the topic of conversation’.

Third, a modified Brief Fear of Negative Evaluation Scale (Modified BFNE; Leary, 1983) was used to assess fear of negative evaluation during the conversation.

Participants

First‐year psychology students (n = 203) were screened using Mattick and Clarke's (1998) Social Interaction and Anxiety Scale (SIAS). Students in the upper (≥32) and lower (≤19) tertiles on the SIAS were invited to participate. Based on the tertile split, the minimum SIAS score for the high social anxiety group (32) and the maximum SIAS score for the low social anxiety group (19) were close to the cut‐off scores recommended by Heimberg, Mueller, Holt, Hope, and Liebowitz (1992) from their validation study (34 and 20 for high and low social anxiety groups, respectively).

Seventy high and low socially anxious students were randomly allocated to either the SFA or EFA condition, and the researcher conducting the attention focus manipulation was blind to anxiety group membership. Social anxiety group membership was confirmed on the day of testing using the SIAS, and nine participants were eliminated because their scores were no longer in the upper or lower tertile range. The number and gender distribution of those eliminated from the high anxiety group (n = 4; one male) and the low social anxiety group (n = 5; two male) was approximately equal.

The final sample comprised 61 participants: 19 males (31%) and 42 females (69%), ranging in age from 17 to 40 years (mean (M) = 19.87, standard deviation (SD) = 4.20). There were 29 people in the high social anxiety group (SFA: n = 14; EFA: n = 15) and 32 in the low social anxiety group (SFA: n = 17; EFA: n = 15). The high socially anxious group had SIAS scores ranging from 32 to 66 (M = 46.5, SD = 7.6), and the low socially anxious group had SIAS scores ranging from 5 to 19 (M = 12.6, SD = 3.7). The mean SIAS scores for the high and low social anxiety groups were comparable with those of the student sample reported by Zou et al. (2007; M = 44.9 and M = 15.5 for their high and low blushing anxious samples, respectively). With regard to age, there were no significant differences between the high and low anxiety groups (F(1, 56) = 0.60, p > .05) or the attention conditions (F(1, 56) = 1.65, p > .05), and the interaction between anxiety group and attention condition was not significant for age (F(1, 56) = 0.08, p > .05; see Table-1). The distribution of males and females across social anxiety groups and attention conditions was not significantly different (χ²(1, N = 61) = 0.29, p > .05 and χ²(1, N = 61) = 0.04, p > .05 for anxiety group and attention group, respectively; see Table-1).

Table 1 Means and standard deviations for participant characteristics by anxiety group and attention condition

Characteristic	High social anxiety		Low social anxiety
	SFA (n = 14)	EFA (n = 15)	SFA (n = 17)	EFA (n = 15)
Number of females (%)	9 (64.3)	10 (66.7)	12 (70.6)	11 (73.3)
Age (years)	18.5 (1.1)	20.3 (3.3)	19.7 (4.0)	20.8 (6.4)
Interaction length (mins)	4.3 (0.4)	4.4 (0.4)	4.4 (0.6)	4.6 (1.2)
SIAS	45.1 (7.3)	47.7 (7.9)	13.2 (4.4)	11.8 (3.0)
SPS	32.1 (13.7)	35.4 (15.6)	8.8 (5.5)	9.3 (4.7)
Self‐focus	54.0 (15.9)	44.6 (23.4)	52.7 (16.8)	33.9 (25.9)
External‐focus	25.3 (24.0)	64.5 (18.3)	22.1 (16.7)	65.5 (20.6)
Modified BFNE	33.3 (11.2)	40.5 (10.2)	22.8 (7.0)	24.0 (6.6)
Familiarity with topic	78.6 (16.0)	43.1 (27.4)	72.9 (28.8)	76.1 (20.6)

Note. EFA = external‐focused attention; Modified BFNE = Modified Brief Fear of Negative Evaluation Scale; SFA = self‐focused attention; SIAS = Social Interaction and Anxiety Scale; SPS = Social Phobia Scale.

Materials

SIAS and Social Phobia Scale (SPS; Mattick & Clarke, 1998)

The SIAS assesses social interaction fears, and the SPS assesses fears of being scrutinised. Both have been found to have strong internal consistency and test‐retest reliability and good discriminant validity (Mattick & Clarke, 1998).

BFNE (Leary, 1983)

The BFNE is a short form of the Fear of Negative Evaluation Scale (FNE; Watson & Friend, 1969). The BFNE was modified so that participants indicated their fear of negative evaluation by the experimenter during the conversation. The items from the BFNE were framed in the past tense to refer to the interaction that the person had just participated in, and references to other people generally were replaced with references to the conversational partner specifically. For example, the item ‘I worry about what other people think of me even when I know it doesn't make a big difference’ was changed to ‘I worried about what the other person thought of me even when I knew it didn't make a big difference’. The instructions were changed so that instead of being asked to rate the extent to which each item was ‘characteristic of you’, participants were asked to indicate how true each statement was ‘of you during the conversation task with the experimenter’ on a scale from ‘not at all true of you’ to ‘extremely true of you’. The original BFNE has high internal consistency and correlates significantly with the original FNE scale (Leary, 1983). The internal consistency of the modified BFNE in the present sample was high, with Cronbach Alpha = 0.93.

VAS

VASs have frequently been used to assess level of anxiety and self‐awareness during social tasks (e.g., Bögels & Lamers, 2002; Zou et al., 2007). The VAS was used before and after each conversation to measure anxiety using a rating scale from 0 (not at all) to 100 (extremely).

SFA, EFA, and Familiarity with the conversation topic

VASs were included to verify that the conversation topic was successful in affecting attention focus and familiarity with the conversation topic. For all items, participants responded on a scale from 0 (not at all) to 100 (extremely). Average responses to VAS ratings of the extent to which participants focused on ‘internal bodily reactions’ and ‘yourself and your feelings’ during the conversation were used to assess extent of SFA, and average responses to VAS ratings of the extent to which participants focused on ‘experiences of other people’ and ‘general environmental conditions’ during the conversation were used to assess the extent of EFA. Participants also rated their familiarity with the conversation topic from 0 (not at all familiar) to 100 (extremely familiar).

Procedure

Those with SIAS scores from the initial screening of ≥32 or ≤19 were invited to participate in a study to investigate the relationship between social behaviour and emotions where they would complete further questionnaires and take part in a brief conversation with a female researcher. Upon arrival, written informed consent was obtained, and participants completed the SIAS, SPS, and the anxiety VAS. Participants were then engaged in conversation about exercise with a female researcher who responded to participants in a friendly, neutral manner. While the broad topic was the same for all participants, the SFA condition was designed to focus participants' attention on their own physiological state by asking them about their personal experiences with exercise, particularly the immediate physiological effects of exercise on their body (e.g., temperature, heart rate, breathing, and sweating). In the EFA condition, the conversation was about environmental factors that could affect athletes' performance at the Beijing Olympic Games (e.g., climate, humidity, and air pollution) so that attention was focused on the experiences of other people in response to external, environmental factors. The average conversation was 4-min and 41-s (SD = 56.3-s) long, and there were no significant differences between anxiety groups (F(1, 57) = 1.20, p > .05) or attention conditions (F(1, 57) = 0.97, p > .05) in the length of the conversation and no interaction between anxiety group and attention condition (F(1, 57) = 0.22, p > .05; see Table-1).

Immediately following the conversation, participants completed the VASs measuring focus of attention, anxiety, and familiarity with the conversation topic followed by the modified BFNE.

RESULTS

SIAS and SPS

Separate 2 (anxiety group) × 2 (attention condition) analyses of variance (ANOVAs) were conducted to confirm the allocation to high and low social anxiety groups (see Table-1). As expected, the high socially anxious group scored significantly higher than the low socially anxious group on the SIAS (F(1, 57) = 500.61, p < .05) and SPS (F(1, 57) = 79.66, p < .05). However, prior to the manipulation, there were no significant main effects for attention condition on SIAS or SPS (F(1, 57) = 0.15, p > .05; F(1, 57) = 0.45, p > .05 for SIAS and SPS, respectively), and no significant interactions between anxiety group and attention condition on either of these variables (F(1, 57) = 1.76, p > .05; F(1, 57) = 0.26, p > .05 for SIAS and SPS, respectively).

Manipulation check

Separate 2 (anxiety group) × 2 (attention condition) ANOVAs were carried out to determine whether the attention focus manipulation had been successful (see Table-1). The dependent variables were the average of responses to the two VASs concerning SFA and EFA. For SFA, there was a significant main effect of attention condition indicating that participants were more self‐focused in the SFA condition (M = 53.3) relative to the EFA condition (M = 39.2; F(1, 57) = 7.05, p < .05), no significant main effect of anxiety group (F(1, 57) = 1.27, p > .05), and no significant interaction between anxiety group and attention condition (F(1, 57) = 0.80, p > .05). For EFA, there was a significant main effect of attention condition with higher external focus in the EFA condition (M = 65.0) compared with the SFA condition (M = 23.7; F(1, 57) = 65.36, p < .05), no significant main effect of anxiety group (F(1, 57) = 0.04, p > .05), and no significant interaction between anxiety group and attention condition (F(1, 57) = 0.17, p > .05).

Fear of negative evaluation during the conversation

A 2 (anxiety group) × 2 (attention condition) ANOVA was conducted to examine modified BFNE scores during the conversation (see Table-1). There was a significant main effect for anxiety group, indicating that high socially anxious participants reported greater fear of negative evaluation during the conversation than low socially anxious participants (M = 36.9 and M = 23.4 for high and low social anxiety groups, respectively; F(1, 57) = 33.39, p < .05). There was no significant main effect of attention condition (F(1, 57) = 2.97, p > .05) and no significant interaction between anxiety group and attention condition (F(1, 57) = 2.00, p > .05).

Familiarity with conversation topic

A 2 (anxiety group) × 2 (attention condition) ANOVA was conducted to examine differences in familiarity with the conversation topic (see Table-1). There was a significant main effect of anxiety group (F(1, 57) = 27.11, p < .05) and of attention condition (F(1, 57) = 6.02, p < .05). Low socially anxious participants reported greater familiarity with the conversation topics than high socially anxious participants (M = 74.5 and M = 60.9, respectively), and participants reported being more familiar with the conversation topic in the SFA condition compared with the EFA condition (M = 75.8 and M = 59.6, respectively). There was also a significant interaction between anxiety group and attention condition (F(1, 57) = 9.84, p < .05). Follow‐up tests used to examine familiarity ratings in attention conditions for anxiety groups separately confirmed that there was a significant difference in ratings between the SFA and EFA conditions in the high social anxiety group (F(1, 57) = 14.10, p < .05) but no difference between SFA and EFA conditions in the low social anxiety group (F(1, 57) = 0.24, p > .05).

Self‐reported anxiety

A 2 (anxiety group) × 2 (attention condition) × (2; time: (before or after conversation)) ANOVA was conducted to determine whether there was a change in self‐reported anxiety following the manipulation (see Fig. 1). There was a significant main effect for anxiety group (F(1, 57) = 22.90, p < .05) where, averaged over time, high socially anxious participants reported greater anxiety than low socially anxious participants (M = 40.7 and M = 17.4 for high and low social anxiety groups, respectively). There was also a significant main effect for attention condition (F(1, 57) = 4.63, p < .05) where, averaged over time, participants in the EFA condition reported greater anxiety than participants in the SFA condition (M = 34.5 and M = 23.6 for the EFA and SFA conditions, respectively). There was no significant interaction between anxiety group and attention condition (F(1, 57) = 0.04, p > .05). The main effect of time was not significant (F(1, 57) = 0.18, p > .05), and there was no significant interaction between anxiety group and time (F(1, 57) = 0.68, p > .05), attention condition and time (F(1, 57) = 0.18, p > .05) or anxiety group, and attention condition and time (F(1, 57) = 0.77, p > .05; see Fig. 1). The pattern of results was the same when familiarity rating was included as a covariate.

Figure 1 Self‐rated anxiety by anxiety group and attention condition: Total sample (n = 61; upper panel); Matched sample (n = 44; lower panel).

Random allocation to conditions resulted in variations in self‐rated anxiety before the conversation where anxiety was higher in the EFA condition for both high and low social anxiety groups, and it is possible that this contributed to the findings reported here. To examine this possibility, the analysis was repeated with participants matched for self‐rated anxiety before the conversation (n = 11 per group). Only the main effect for anxiety group was significant in this analysis (F(1, 40) = 13.14, p < .05; see Fig. 1).

DISCUSSION

This study investigated the effects of SFA compared with EFA on self‐rated anxiety in high and low socially anxious individuals. The study was designed to manipulate attention focus in the context of an ecologically valid conversation task similar to that used by Zou et al. (2007). The attention focus task used in the present study, however, did not require participants in the SFA condition to directly focus attention during the conversation ‘inwards by concentrating on your breathing, heart‐rate, your voice, or any signs of blushing’ (p. 2329). This instruction is likely to elicit evaluation concerns particularly as participants were selected for anxiety about blushing, and anxiety about blushing is understood to be centrally related to fear of negative evaluation. Here, the SFA condition indirectly focused attention on the individual's internal physiological state, asking participants to think about their experiences with the physiological signs of exercise (body temperature, heart rate, breathing, and sweating). Using measures of SFA and EFA similar to those used in previous work (e.g., Bögels & Lamers, 2002), it was demonstrated that the conversation task successfully altered focus of attention. Importantly, scores on the modified BFNE did not differ across attention conditions, although expected effects of anxiety condition on the modified BFNE were found: Those in the high social anxiety group reported higher fear of negative evaluation during the conversation than those in the low social anxiety group.

No effect of SFA on self‐reported anxiety, either as a main effect or as an interaction with social anxiety group, was found in the present study. Therefore, the results do not support the view that SFA will lead to social anxiety irrespective of anxiety status (e.g. Bögels & Lamers, 2002; George & Stopa, 2008; McManus et al., 2008; Woody, 1996; Woody & Rodriguez, 2000), nor do they support the prediction that those with high social anxiety would have higher self‐reported anxiety in the SFA condition compared with the EFA condition, while attention focus would have no effect on anxiety for those with low social anxiety (Zou et al., 2007). Instead, using a manipulation of attention that was ecologically valid task and designed to influence focus of attention without directly influencing fear of negative evaluation, no association between SFA and anxiety was found (cf. Bogels et al., 2002). Inconsistencies between the findings of the present study and many earlier results may be explained by these important methodological differences.

Cognitive models of social anxiety argue that it is the interaction of SFA and fear of negative evaluation that gives rise to social anxiety, and interventions to reduce social anxiety have been designed to target both of these processes. The findings of the present study have implications for interventions to reduce social anxiety, suggesting that SFA alone may not be a suitable target for treatment (see also Voncken et al., 2010). The findings are consistent with suggestions that manipulations of SFA reported to affect anxiety in the literature may have acted by directly affecting fear of negative evaluation (Bögels & Mansell, 2004).

Several limitations of the present study must be acknowledged. First, the number of participants was relatively small (n = 14 to n = 17 in each group). However, these numbers are similar to those reported in other published studies (e.g., Vassilopoulos, 2008; Zou et al., 2007). Second, participants in the present study were asked to rate their level of anxiety immediately before and immediately after the conversation, rather than being asked to rate their anxiety during the conversation. Thus, the findings of the present study may reflect apprehension about the task before the conversation, and relief after the conversation was over. Other work in this area has also required participants to rate their mood immediately before and immediately after an experimental manipulation (e.g., Vassilopoulos, 2008), and the accuracy of post‐task ratings of the experience during the task may be questioned. Third, unlike previous work examining the relationship between SFA and anxiety, participants here were asked to rate their fear of negative evaluation during the conversation to directly assess the extent to which the attention manipulation was associated with concerns about negative evaluation. The BFNE was designed as a trait measure and was modified here to detect state fear of negative evaluation. It is therefore possible that the measure was not sensitive enough to detect situational differences in fear of negative evaluation.

In summary, this study addressed several concerns associated with previous research concerning the role of SFA in eliciting social anxiety. The attention manipulation occurred in the context of a commonly encountered social interaction between two people. The attention manipulation was effective in influencing direction of attention and did not appear to be confounded by directly causing fear of negative evaluation. Under these conditions, SFA was not associated with self‐reported anxiety. The inconsistency between the conclusions of the present study and those of earlier work where SFA was associated with increased anxiety may be explained with reference to Watkins and Teasdale's (2004) distinction between adaptive, experiential SFA and maladaptive, analytical SFA. For example, the processes engaged during the SFA manipulation used by Zou et al. (2007) appear to be analytic and evaluative, while the processes engaged by the SFA manipulation in the present study reflect a more experiential focus on the self. Further research comparing analytical and experiential SFA using an ecologically valid conversation task is needed to clarify this interpretation.

There is evidence to suggest that undergraduate populations with high levels of social anxiety constitute an appropriate analogue for clinical samples (Turner, Beidel, & Larkin, 1986), and the average SIAS scores of the high social anxiety group in the present study (M = 46.5) were higher than that of a sample with social phobia reported by Mattick and Clarke (1998; M = 34). However, extension of this study to a clinical sample would strengthen the findings. Further research in this area is important given that social anxiety is both common and disabling, and that therapeutic interventions targeting attention are recommended to treat social anxiety. Increasing our understanding of the factors that maintain social anxiety will enable more targeted treatment approaches.