Self-presentational features in childhood social anxiety

Self-presentational features in childhood social anxiety

Social anxiety has long been associated with self-presentational concerns about how one is being perceived and evaluated by others. Schlenker and Leary (1982) formulated the following key proposition: ''Social anxiety arises in real or imagined social situations when people are motivated to make a particular impression on others but doubt that they will do so, because they have expectations of unsatisfactory impression-relevant reactions from others' ' (p. 645). Similarly, clinical analyses of social anxiety disorder point to anticipated and imagined negative social evaluation as a hallmark feature (e.g., Clark, 2001), and the DSM-IV (APA, 2000) explicitly focuses on the individual's fears of humiliation and embarrassment in social situations.

The self-presentational account of social anxiety is consistent with considerable research evidence regarding the patterns of sociobehavioral skills and cognitive characteristics associated with social anxiety, both in nonclinical and clinical populations. First, there is good evidence that social anxiety is associated with perceived, and potentially also genuine, difficulties in making positive impressions on others. Spence, Donovan, and Brechman-Toussaint (1999), for example, provided not only evidence of negative expectancies and self-evaluation among social phobic children, but also convergent evidence from independent observers and peers of poorer social skills. Similar evidence of impaired social performance has also been reported by Beidel, Turner, and Morris (1999), Morgan and Banerjee (2006), and Rao et al. (2007). Undoubtedly, any such difficulties in social skills have implications for the evaluations that others form of the self, and this of course feeds into the cognitive patterns of the socially anxious individual. Indeed, Rapee and Heimberg's (1997) cognitive-behavioral model of social phobia places the 'mental representation of self as seen by audience' at the heart of the social phobic cognitive profile. A recent review of the social-cognitive patterns exhibited by socially anxious youths highlights the detection, perception, and interpretation of social-evaluative threat as a highly significant part of the disorder (Banerjee, 2008). In turn, focusing of attentional resources onto the internal representation of one's public self-image, along with hypervigilance towards any potential threat in the environment, is likely to have serious consequences for actual social performance because attention to most features of the social partner and the task at hand can become severely diminished. Where such patterns are accompanied by -or even contribute towards -actual social trauma, such as severe peer rejection, victimization, and humiliation (Rapee & Spence, 2004), it is easy to see how much more severe cases of social phobia and comorbid depressive symptomatology can ensue. To summarize, social anxiety is thought to involve an interplay of cognitive features and social skills deficits that together make social interactions highly problematic, and that clearly relate to self-presentational processes in terms of both perceived and actual impressions made on others.

It is important to stress that the connection between social anxiety and self-presentational processes is likely to emerge early in development. Although the average onset of clinical social phobia is typically reported as mid-adolescence (Rapee & Sweeney, 2001), symptoms of social anxiety can be reliably measured in middle childhood (e.g., Beidel, Turner, & Morris, 1995; Theoretical and clinical models of social anxiety highlight links with self-presentational concerns and behavior, but little is known about these features in early development. In the present investigation, a nonclinical sample of 196 children aged 8-9 years completed self-report measures of social anxiety, depressive symptoms, and usage of self-presentational tactics, as well as a self-description task measuring the capacity to differentiate between audiences known to have different preferences. After controlling for concurrent depressive symptoms, social anxiety was associated with increased usage of self-presentational tactics, but also with poorer scores on the audience differentiation task. A follow-up assessment of groups identified as highly socially anxious or non-socially anxious showed that these patterns were durable over 12 months. Directions for future research on the social developmental trajectory of children with social anxiety are suggested.

ß 2009 Elsevier Ltd. All rights reserved.

Stone, 1993), and we have steadily growing knowledge about the behavioral, cognitive, and emotional features associated with social anxiety in childhood (e.g., Beidel et al., 1999;Rao et al., 2007;Rapee & Sweeney, 2001;Spence et al., 1999). Moreover, researchers have noted that the onset of social phobia in adolescence may be connected more to the greater functional implications of longstanding social fears in this developmental period, rather than reflecting a substantial increase in actual levels of social anxiety (e.g., Rapee & Spence, 2004). Thus, there is a great need to understand more about emergence of social anxiety earlier in development, in order to clarify the patterns of cognition, motivation, and behavior that could potentially give rise to problems of increasing magnitude during adolescence. We argue that self-presentational processes are likely to play a key role in the social-evaluative fears that emerge during middle childhood. Although it is certainly the case that children's selfpresentational reasoning will continue to develop through the adolescent years as social role-taking becomes more sophisticated (e.g., Selman, 1980), there is good evidence that self-presentational processes become significant -and can be measured effectivelyin middle childhood. Parker and Gottman's (1989) careful formulation of the development of peer relations in childhood suggests that children from around 8 years onwards become increasingly concerned about gaining social acceptance into peer groups, and it seems likely that these concerns make social evaluation highly salient for children at this time. In correspondence with this, researchers have reported developmental progress between 6 and 11 years of age in children's use and understanding of various impression management tactics (Bennett & Yeeles, 1990a;Bennett & Yeeles, 1990b;Watling & Banerjee, 2007a;Watling & Banerjee, 2007b). Moreover, children from around 8 years of age become increasingly able to recognize that self-presentational behavior may be adjusted to suit the distinctive preferences of different audiences (Banerjee, 2002b). Other studies have identified corresponding developmental changes in the ability to explain others' behavior in terms of self-presentational concerns (Banerjee, 2002a;Banerjee, 2002c;Banerjee & Yuill, 1999a). Indeed, Heyman and colleagues (Heyman, Fu, & Lee, 2007;Heyman & Legare, 2005) have convincingly demonstrated that children in elementary school become increasingly skeptical about what other people say about themselves, due in large part to their recognition that self-presentational motives can distort people's self-descriptions. Thus, middle childhood may be characterized as a period of time when children increasingly appreciate the significance of their behavior for evaluation by others.

Given existing evidence on social anxiety and self-presentational awareness in middle childhood, it seems highly plausible that social anxiety may be connected with a distinctive profile of self-presentational behaviors and skills during this period of development. However, although these issues are likely to be a critical part of the psychopathology, specific evidence regarding the self-presentational tactics and skills of children with high levels of social anxiety is scarce. The present study ascertains the degree to which social anxiety is associated with the self-reported use of self-presentational tactics and with the capacity to differentiate self-presentations to different audiences.

Self-presentational tactics

Since Goffman's (1959) seminal text on self-presentation, numerous authors have fleshed out the details of how selfpresentational motives can be translated into behavioral tactics. Jones and Pittman (1982), for example, provided a taxonomy including ingratiation (making oneself seem likeable), selfpromotion (making oneself seem competent), exemplification (making oneself seem morally worthy), and supplication (making oneself seem helpless), among others. In a recent paper on the measurement of self-presentation, Lee, Quigley, Nesler, Corbett, and Tedeschi (1999) presented a scale measuring self-reported usage of twelve different self-presentation tactics, including assertive tactics designed to create a given identity (as in the four examples above) as well as defensive tactics designed to protect or restore an identity that has been threatened (e.g., excuses, disclaimers). In that study, there was excellent internal consistency across the entire scale, but the authors presented evidence for a two-dimensional structure separating assertive and defensive tactics. Moreover, they found that social anxiety was associated with increased self-reported use of defensive tactics. This is consistent with arguments that social anxiety and shyness are associated with a 'protective' self-presentational style (e.g., Arkin, 1981;Schlenker & Leary, 1982).

Studies of self-presentational tactics in children are still at a fairly early stage. However, some evidence has been found for the use or understanding by elementary school children of ingratiation and self-promotion (e.g., Aloise-Young, 1993;Bennett and Yeeles, 1990;Watling & Banerjee, 2007a), modesty (e.g., Banerjee, 2000;Watling & Banerjee, 2007b), disclaimers (e.g., Bennett, 1990), and excuses (e.g., Juvonen, 1996;Ohbuchi & Sato, 1994). We believe that individual differences in childhood social anxiety may be associated with variability in the tendency to use self-presentational tactics. As noted above, concerns about social evaluation are certainly identifiable in children, and the section on social anxiety disorder in the Child Version of the Anxiety Disorders Interview Schedule for DSM-IV (Silverman & Albano, 1996) specifically includes questions pertaining to evaluative judgements (e.g., being thought of as stupid, being laughed at). It seems highly plausible that these concerns about social evaluation will be reflected in increased usage of selfpresentational tactics when in social situations. Fears of negative evaluation in socially anxious children would seem to lead to defensive tactics, in particular, but it is not as yet clear whether children's self-presentational tactics would fall into an assertive vs. defensive dichotomy. The present study presents the first evaluation of this issue by examining links between social anxiety and scores on an adapted form of Lee et al.'s (1999) questionnaire, covering selfpromotion, ingratiation, excuses, and disclaimers.

Children completed a simplified adaptation of Lee et al.'s (1999) Self-Presentational Tactics Scale. The adapted scale measures selfreported use of twenty self-presentational tactics (e.g., ''When I do well at something, I tell others how important it was,'' ''When things go wrong, I try to explain to others why it was not my fault''). The wording of the items was amended from the original scale where necessary, in order to facilitate comprehension by children. The items used in the scale were selected from four categories in Lee et al.'s original scale, namely self-promotion, ingratiation, disclaimers, and excuses. All of these tactics have been previously studied in developmental research, and found to be relevant for school-aged children (e.g., Bennett, 1990;Bennett and Yeeles, 1990;Ohbuchi & Sato, 1994;Watling & Banerjee, 2007a). Each item was presented in written and spoken form on the computer, and children were required to click on one of five response buttons (ranging from 1 = Not at all to 5 = All of the time). Children received scores with a possible range from 20 to 100, with higher scores indicating higher self-reported usage of the tactics. Principal components analysis suggested a one-factor solution, and internal consistency was satisfactory for the scale, a = .82. 1

Our results show that elevated social fears and anxieties of a significant minority of children translate into a durable tendency Table 1 Inter-correlations between self-presentation tactics, audience differentiation, social anxiety, depressive symptoms, and gender.

Modifying self-presentation to different audiences

Self-presentation goes hand in hand with situational variability, because different audiences may evaluate the same behavior differently. Thus, it is often necessary to adjust one's selfpresentational behavior when interacting with different audiences. Importantly, the capacity to do so may be distinguished from the concerns about social evaluation discussed above. This kind of distinction is not without precedent. In an early revision of Snyder's (1974) Self-Monitoring Scale, Lennox and Wolfe (1984) extracted a specific measure of 'ability to modify self-presentation', separate from 'concern for appropriateness'. The same is true for children: Howells and Fishfader (1995) identified two factors in Graziano, Leone, Musser, and Lautenschlager's (1987) Junior Self-Monitoring Scale, namely 'concern for social appropriateness' and 'ability to modify self-presentation'. However, relatively few studies have directly targeted children's capacity to tailor their self-descriptions in front of different audiences. Aloise-Young (1993) showed that school-aged children become capable of adjusting their selfdescriptions in front of an audience in order to achieve specific social interactional goals. More recently, Banerjee (2002b) has argued that there are systematic between-and within-age group differences in the capacity to differentiate self-presentations to different audiences. In an initial study, 10-year-olds were shown to recommend different self-descriptions for a hypothetical character meeting new peers vs. meeting new adults. Two subsequent studies revealed a developmental increase, between 6 and 10 years, in the tendency to recommend differentiated self-descriptions to peers with varying preferences (e.g., hardworking vs. sporty). In addition, there was preliminary evidence that this kind of audience differentiation is linked with individual differences in social functioning: children with a greater proportion of reciprocated playmate nominations scored higher in audience differentiation.

We suggest that social anxiety is likely to be associated with impaired ability to modify self-presentation to audiences with different attributes and preferences. There is good reason to expect this kind of inverse relationship. Clark & Wells (1995) cognitive model of social phobia emphasizes increased self-focused attention, and reduced attention to external social cues. For example, Mellings and Alden (2000) have shown that socially anxious individuals who participated in a social interaction had increased negative self-focused attention and poorer recall of partner-related information. Not only will this pattern of reduced attention to partner-related attributes result in a diminished probability of encoding audience responses that could disconfirm negative expectations, but, as noted earlier, it will also hinder the capacity to respond appropriately to different audiences. In line with this argument, Lennox and Wolfe (1984) found that their 'ability to modify self-presentation' subscale was indeed negatively correlated with social anxiety. We propose that socially anxious children, too, could have similar difficulties in adjusting their self-presentation in response to audience attributes and preferences. This would be consistent with existing evidence suggesting that children with higher levels of social anxiety are poorer at recognising how self-presentational motives can give rise to effective emotion displays (Banerjee & Henderson, 2001). Difficulties in appropriately differentiating self-presentations to different audiences could be a core element of a social skills profile that contributes to the interpersonal challenges faced by children with social anxiety (e.g., Ginsburg, La Greca, & Silverman, 1998;La Greca & Stone, 1993;Rao et al., 2007;Spence et al., 1999).

The present study

The focus of our investigation was on the extent to which social anxiety in children is associated with self-reported use of selfpresentational tactics and with differentiation of self-presentations to different audiences. We used an adapted form of Lee et al.'s (1999) self-presentation tactics scale, and Banerjee's (2002b) task eliciting recommended self-descriptions for a hypothetical story character meeting new classmates with different preferences. It was hypothesized that higher levels of social anxiety would be associated with greater self-reported use of self-presentational tactics, in line with evidence that social anxiety involves close attention to social evaluation. On the other hand, we expected greater social anxiety to be associated with poorer skills in differentiating self-presentation across audience. These hypotheses were evaluated using continuous data from the whole sample. Additionally, the durability of observed differences between children relatively high vs. low in social anxiety was tested with a follow-up assessment after 12 months. To enable these comparisons, we classified children on either end of the distribution into ''high social anxiety'' or ''non-socially anxious'' groups, using the criteria recommended in the SASC-R technical manual (La Greca, 1999).

Method

Participants

Our sample consisted of 196 children (108 boys, 88 girls) aged 8-9 years (M 9.03 years, range 8.51-9.53). The children were from seven state-funded elementary schools that had elected to participate in a wider longitudinal study of social-cognitive and socio-emotional development. One classroom of children from each school took part in the study, and there were no reasons to expect differences between the participating pupils and any other classes in the school. Because the sample was not selected for any particular difficulties, we expected reading levels within the average range for the age group. The schools were situated in urban and suburban communities in and around a city in the UK, representing a wide range of socioeconomic status groups (proportion eligible for free school meals, according to national criteria, ranged from 5.3% to 45.7%, average 23.9%) and mostly of white ethnicity (all schools >85%). We obtained informed consent from the schools participating in the research, and full information about the project was sent to parents who had the opportunity to refuse to allow their children to participate. Before each data collection session, children were provided with a general introduction to the tasks and gave their assent to participation; they were advised explicitly that that they could withdraw at any time. Participation rates were above 95% across classes. Longitudinal follow-up assessments were completed again 12 months later by 106 of the 129 children initially identified on the basis of La Greca (1999) recommendations as either 'high socially anxious' or 'non-socially anxious.'

Measures

Children completed self-report measures of social anxiety, depressive symptoms, and self-presentation tactics, as well as a self-description task measuring the capacity to differentiate between audiences known to have different preferences. The measures were presented as part of larger task batteries administered through a computer interface, developed by the authors using Runtime Revolution. The self-presentation tactics and audience differentiation measures were completed in the course of one task battery in the middle of the school year, and the social anxiety and depression measures were part of a second task battery that followed approximately three months later. The measures were completed 12 months later by those children identified as 'highly socially anxious' or 'non-socially anxious' (see below), in order to evaluate the durability of any observed group differences on the self-presentation scores.

Social anxiety

Children completed La Greca and Stone's (1993) Social Anxiety Scale for Children-Revised. The 22 items (including 4 filler items) of the scale were presented in written and spoken text on the computer, and children were required to click on one of five response buttons (ranging from 1 = Not at all to 5 = All of the time). Children received a score ranging from 18 to 90, with higher scores indicating higher social anxiety. Internal consistency was satisfactory, a = .91. There is evidence of good reliability and validity for this scale, including convergent evidence relating to data on social relationships from other informants, such as sociometric measures of peer rejection (Inderbitzen, Walters, & Bukowski, 1997;La Greca, 1999;La Greca & Stone, 1993).

In addition to analysing the continuous data on social anxiety from the whole sample, we followed the recommended strategy for identifying 'non-socially anxious' and 'high socially anxious' groups as specified in the technical manual for the SASC-R (La Greca, 1999). In line with recommendations, we used cutoffs at half a standard deviation above and below the mean (specifically, !51 or 37). These cutoffs, which are virtually identical to those recommended by La Greca (1999), yielded a subsample of 70 'nonsocially anxious' and 59 'high socially anxious' children, representing 35.7% and 30.1% of the total sample, respectively. The proportion of children scoring in the ''high socially anxious'' group is similar to the corresponding proportion of an unselected sample (23%) that is recorded in the SASC-R manual.

Depressive symptoms

Children completed the short form of Kovacs's Children's Depression Inventory (1992), with the 10 test items accompanied by 10 emotionally neutral filler items. Each item consists of three statements, and children are asked to indicate the statement that is most true for them. In the computer interface, children were presented with the written and spoken text of the three statements, each represented by an on-screen button. Children selected their response by clicking on one of the three buttons. Each item is given a score of 1, 2, or 3, with 1 for the least negative and 3 for the most negative statement. This gives rise to a depression score ranging from 10 to 30, with higher scores indicating more depressive symptoms. Internal consistency was satisfactory, a = .79. Kovacs (2003) reports substantial evidence for the reliability and validity of this scale, including convergence with data from multiple informants and clinical diagnoses.

Audience differentiation

Children completed a computerized version of the audience differentiation task described by Banerjee (2002b, Studies 2 and 3). Children were presented with two illustrated stories where the protagonist is a child, matched to the participant's gender, who has just moved to a new school. The children are asked to imagine that the protagonist wants his or her new classmates to think that s/he is nice, and are further told that all the protagonist knows about the classmates is that they either like people who are good at sports (in one story) or like clever people who work hard (in the other story). The task for the participants is to recommend possible selfdescriptive statements, all of which they are told are true and all of which are positive, for the protagonist to use when meeting his or her new classmates. The self-descriptive statements are presented in pairs, where participants must choose to recommend one of the two statements. There were three types of pairs -academic skill vs. physical skill; academic skill vs. interpersonal skill; and physical skill vs. interpersonal skill -and two pairs of choices of each type.

We counted, for each story, the number of times each child selected the academic skill statement over the physical skill statement, the frequency of selecting the academic skill statement over the interpersonal skill statement, and the frequency of selecting the physical skill statements over the interpersonal skill statement (all scores out of 2). Then, following Banerjee (2002b), we calculated the difference in each of these scores across the two stories. The three resulting difference scores were then summed together to create an overall audience differentiation score ranging from À6 to +6. A score of 0 indicates no differentiation between the two audiences. Positive scores indicate the expected differentiation between audiences, such that academic skill statements were selected more often for the 'hardworking' story than for the 'sporty' story, and physical skill statements were selected more often for the 'sporty' story than for the 'hardworking' story. Previous research with elementary school children has provided some evidence of convergence of this measure with data from other informants; specifically, children scoring higher on this measure have a greater proportion of reciprocated playmate nominations (Banerjee, 2002b).

Making desired impressions on others depends not just on knowledge of effective self-presentational behaviors but also on appropriate responses to the attributes and preferences of the particular audience with whom one is interacting. This strategic shaping of one's public self-image in front of different audiences must be viewed as conceptually distinct from the tendency to use self-presentational tactics such as self-promotion and excuses. Banerjee's (2002b) research demonstrated that the capacity to recommend different verbal self-presentations for story protagonists meeting different audiences (with varying characteristics and preferences) increased significantly between 6 and 11 years of age. The present study builds on this evidence by showing that 8-to 9year-olds with relatively high levels of social anxiety, after controlling for comorbid depressive symptoms, perform more poorly on an audience differentiation task of this kind. This is consistent with work on self-monitoring with adults showing that social anxiety is negatively associated with the ability to modify self-presentation (Lennox & Wolfe, 1984).

Difficulties of socially anxious children with audience differentiation shed some light on the somewhat mixed experimental evidence regarding the precise social skills deficits of socially anxious children (e.g., Cartwright-Hatton, Hodges, & Porter, 2003;Cartwright-Hatton, Tschernitz, & Gomersall, 2005;Morgan & Banerjee, 2006;Rao et al., 2007;Spence et al., 1999). Specifically, a difficulty with audience differentiation could mean that socially anxious children will exhibit entirely adequate behavior in some interpersonal situations but display problematic responses in other situations. Such a pattern is entirely consistent with conclusions drawn by theorists and clinicans about socially anxious individuals' 'anxious self-preoccupation'. As Leary (1996, p. 192) has succinctly explained, ''When people are excessively preoccupied with their self-presentations, they pay less attention to what is going on around them.'' Ironically, the failure to attend to audience characteristics may often undermine the success of the self-presentation.

Our results imply that further research in this area should utilize designs where children have to alter self-descriptions under different situational demands (e.g., different audiences). Aloise-Young (1993) showed how elementary school children increasingly come to modify their self-descriptions in order to achieve specified social goals (e.g., get picked as partner for a game), and recent evidence has shown that such strategic selfpresentation may be impaired in special populations such as children on the autistic spectrum (Begeer et al., 2008). Present findings suggest that these self-presentation skills may also be less evident in those with relatively high levels of social anxiety. It is important to stress, however, that we cannot yet make firm predictions about how the pattern of poorer audience differentiation will change over the course of development. It seems conceivable that the socially anxious group will continue to display poor audience differentiation through the adolescent years and beyond, but a plausible alternative hypothesis is that the group has a delayed acquisition of basic audience differentiation skills during childhood. The present study found no significant effects of time, but the audience differentiation scores did tend to rise between the two timepoints for both the high and low socially anxious groups. Future research working over a wider range of ages, and using audience differentiation tasks of varying complexity, may fruitfully investigate possible developmental trends. Such work may also clarify the medium-to long-term consequences of inadequate audience differentiation for key developmental tasks such as friendship formation during adolescence, as noted above.

Procedure

For each assessment session, children were seen in small groups of typically between 3 and 6, in their schools' computer laboratories. Each child was seated at a separate computer terminal, and children were spaced apart to prevent them from seeing each other's monitors. After a brief introduction to the setup, the children donned headphones and completed the tasks. The tasks were automatically presented by the computers in randomized order, and children were not permitted to consult with each other during the administration session. Research assistants were present to answer any questions and to supervise use of the computers. Each task was preceded by a screen giving instructions, and all questionnaire measures included a practice item to ensure that the response option buttons were understood by the children. All information was presented in written and spoken form, and children could hear the response options at any time following the question by moving the mouse over the on-screen buttons. Each task battery usually lasted between 10 and 20 min, although there was never any time limit for any of the tasks. Table 1 lists the mean scores of the entire sample on each measure, and shows the inter-correlations between the measures, as well as the point-biserial correlation of each measure with gender. The analysis suggests that social anxiety was associated positively with self-reported usage of self-presentational tactics, but negatively with audience differentiation scores. However, a similar pattern was observed for depressive symptoms, and social anxiety and depressive symptom scores were highly correlated. With regard to gender, boys scored higher on self-reported usage of self-presentational tactics. However, as no other gender differences were apparent (jrjs < .07) and preliminary analysis confirmed there were no interactions of the key variables with gender, this variable was excluded from further analysis.

Results

We conducted a regression analysis for each of the two selfpresentation measures, with social anxiety and depression as predictors. As Table 2 shows, social anxiety was a significant positive predictor of self-reported usage of self-presentational tactics, whereas it was a significant negative predictor of audience differentiation scores. Depression scores were no longer significantly predictive of either self-presentation score after controlling for social anxiety.

1 Separate totals for the assertive and defensive subscales could also be calculated, in line with Lee et al. (1999), as each of these subscales had adequate internal consistency estimates. However, the pattern of results obtained when analysing each subscale separately was virtually identical to the overall pattern reported here, consistent with the unidimensional structure of our adapted scale. A copy of the adapted scale is available from the authors.

Building on the above analyses of continuous data from the whole sample, we next sought to identify the durability of the selfpresentation profiles of children presenting with relatively high vs. low levels of social anxiety. We classified children at either end of the distribution as 'non-socially anxious' (n = 70) or 'high socially anxious' (n = 59), on the basis of the recommendations set out in the technical manual for the SASC-R (La Greca, 1999), as described earlier. An initial univariate analysis of covariance (ANCOVA) on the self-reported usage of self-presentation tactics, with social anxiety group as the between-subjects variable and depression as a covariate, showed a significant effect of social anxiety group: the relatively high socially anxious group reported using selfpresentational tactics significantly more often than the nonsocially anxious group (means (SDs), 63.22 (10.67) vs. 53.70 (13.32), respectively; F(1,126) = 13.02, p < .001, partial h 2 = .09). A corresponding ANCOVA on the audience differentiation scores also showed a significant effect of social anxiety group: the relatively high socially anxious group were significantly less likely to differentiate between audiences appropriately than the nonsocially anxious group (means (SDs), .97 (1.85) vs. 2.14 (2.41), respectively; F(1,126) = 5.18, p < .05, partial h 2 = .04). It is worth noting, however, that the mean audience differentiation score was significantly greater than 0 for both groups (one-sample t-tests, ts > 4.01, ps < .001). These patterns are entirely in line with the regression analyses of the continuous data from the whole sample, described earlier.

The main purpose of examining relatively high and low social anxiety groups was to follow these groups up 12 months later in order to determine whether the self-presentational patterns would be durable over a year. First, we established that the high socially anxious and non-socially anxious groups still significantly differed on their social anxiety score 12 months later. A univariate ANCOVA on the social anxiety score at the second timepoint, with social anxiety group as the between-subjects variable and depression as a covariate, showed a significant effect of social anxiety group, with the relatively high social anxiety group continuing to report greater anxiety about social situations than the non-socially anxious group (means (SDs), 48.17 (12.87) vs. 33.15 (10.07), respectively; F(1, 103) = 19.78, p < .001, partial h 2 = .16).

Our key focus was on the extent to which the two groups would also continue to show the initially observed differences in the self-presentational features, independent of their initial depression scores. A mixed-design ANCOVA was conducted on each of the two self-presentation scores. In each of these analyses, social anxiety group was the between-subjects factor, depression was a covariate, and timepoint was a repeated-measures variable. For both self-presentation tactics and audience differentiation scores, the only significant effect in each analysis was the main effect of social anxiety group (F(1, 103) = 19.14, p < .001, partial h 2 = .16; and F(1, 103) = 5.53, p < .05, partial h 2 = .05, respectively; all other Fs < 2.1, ps > .15). Follow-up ANCOVAs on each score at the second timepoint, with social anxiety group as the between-subjects factor and depression as covariate, confirmed the main effect of social anxiety group on each measure. Specifically, the relatively high socially anxious group continued to report self-presentation tactics significantly more often than the non-socially anxious group at the second timepoint (means (SDs), 59.96 (12.98) vs. 50.24 (11.42), respectively; F(1, 103) = 15.64, p < .001, partial h 2 = .13). In addition, the high socially anxious group still tended to score lower than the non-socially anxious group on audience differentiation (means (SDs), 2.04 (2.22) vs. 2.68 (2.25), respectively; F(1, 103) = 3.03, p < .09, partial h 2 = .03).

Discussion

The present study provides evidence that childhood social anxiety is associated with distinctive self-presentational features. As hypothesized, social anxiety was uniquely related to greater self-reported usage of self-presentational tactics, as well as to poorer performance on the audience differentiation task. Moreover, differences on these measures between those scoring relatively high vs. low on social anxiety were present 12 months later. Although our data do not speak to the long-term development of socially anxious children through to adolescence, they do suggest that self-presentational features deserve further attention in models of how social skills and cognitive-motivational patterns develop in children with high levels of social anxiety.

Mean (SD)

Self-presentational tactics Audience differentiation Social anxiety Depression Gender a Self-presentation tactics 57.47 (12.55) -À.10 .29 *** .16 * À.16 * Audience differentiation 1.65 (2.10) -À.19 ** À.14 + À.05 Social anxiety 44.03 (14.57) -.62 *** .06 Depression 13.28 (3.54) -À.01 Gender a a 0 = boy, 1 = girl. + p < .10. * p < .05. ** p < .01. *** p < .001.

Table 2

Regression analyses of self-presentation tactics and audience differentiation scores, with social anxiety and depression as predictors.

Predictors

Self-presentation tactics Audience differentiation to endorse self-presentational tactics that are designed to shape their public self-image. Our adapted version of Lee et al.'s (1999) self-presentation tactics scale was internally consistent, and children's overall usage scores were associated as expected with their levels of social anxiety after controlling for depressive symptoms. Thus, children with relatively high levels of social anxiety reported using self-presentational tactics more frequently than non-socially anxious children, and this difference was still present in those same children one year later. The tendency to use self-presentational tactics more frequently is compatible with the argument that ''concerns about one's public impressions lie at the heart of social anxiety'' (Leary, 2001, p. 222). Specifically, the socially anxious individual's strong motivation to make desired impressions on others appears to be translated into greater use of self-presentational strategies. These results therefore support the self-presentational model of social anxiety (e.g., Schlenker & Leary, 1982), and provide empirical confirmation that self-presentational processes are likely to be pertinent to social anxiety in childhood, as we know they are in adulthood. Our analysis treated the usage of self-presentation tactics as a unidimensional construct, since there was no evidence for differentiation between the assertive tactics of self-promotion and ingratiation and the defensive tactics of excuses and disclaimers (see footnote 1). However, as noted earlier, Lee et al.'s (1999) study with college students found that social anxiety was significantly associated only with defensive tactics. Such differentiation is consistent with the idea that social anxiety in adults is associated with a 'protective' rather than 'acquisitive' self-presentational style (Arkin, 1981), as well as with longstanding clinical observations of 'safety-seeking behaviors' designed to minimize feared negative evaluation (e.g., Clark, 2001). In contrast, our study provides tentative evidence that social anxiety early in development may be associated with a range of tactics designed to gain favorable responses from others as well as to avoid making undesirable impressions.

It is plausible that socially anxious youths will come to exhibit a specifically defensive or protective self-presentational style only later in development, perhaps after accumulating a history of unsuccessful peer interactions. Research suggests that withdrawn behavior is likely to become increasingly perceived as deviant during the latter years of primary school (French, 1988;Rubin et al., 1998;Younger & Piccinin, 1989), so that socially anxious children, who typically show greater social avoidance, may be rebuffed when they do attempt to use assertive self-presentational tactics. Such problems may be exacerbated by ineffective use of the self-presentational tactics due to inadequate responsiveness to audience preferences, as also observed in the present study (see Section 4.2). Receiving negative reactions from peers may in turn steadily encourage socially anxious individuals to favor defensive self-presentational behaviors.

Clearly, it is too early to make firm conclusions about the specific self-presentational tactics engaged by children with social anxiety. Future research needs to measure the usage of selfpresentational tactics by socially anxious individuals from a wide range of age groups. In addition, the development of more detailed age-appropriate measures of specific assertive and defensive tactics (e.g., Lee et al., 1999, measured twelve different tactics), along with observational measures of self-presentational behavior, would help to clarify the self-presentational profile of socially anxious youths through childhood and adolescence. An important limitation of the present study lies in the reliance on self-report measures of both self-presentation and social anxiety. Although self-report methods are widely used in work on children's social anxiety from around the age of 8 onwards (Rapee & Sweeney, 2001), they do raise questions about the extent to which scores translate into actual social performance. Children with functional impairments in learning and behavior have been found to have systematic biases in self-perception (e.g., positive illusory biases in children with ADHD; Hoza et al., 2004), and children's responses to general questions about motivation and behavior may not always map accurately onto their actual social interactions in specific situations. Therefore, a key challenge for future research will be to determine precisely how socially anxious children's reported selfpresentational tactics map onto actual social behaviors.

Such work could help us better understand the role played by self-presentational tactics in other aspects of socially anxious youths' development. For example, La Greca and Lopez (1998, p. 90) have provided valuable data showing that adolescent girls with higher levels of social anxiety, in particular, ''reported having fewer best friends, feeling less competent in their friendships, and perceiving their friendships as less supportive, less intimate, and lower in companionship.'' On the other hand, Erath, Flanagan, & Bierman (2007) have more recently shown that the link between peer victimization and social anxiety in early adolescence is stronger for boys than for girls. Moreover, recent work has suggested that social skills deficits may mediate the relationship between social anxiety and negative peer outcomes (Greco & Morris, 2005). A major task for future research, then, is to understand how the self-presentational behaviors of socially anxious children may hinder their progress in key social developmental tasks, such as the establishment of intimacy in adolescent female friendships and the maintenance of a secure position within adolescent male peer groups. This requires work across a wider age range that utilizes data from multiple informants (i.e., not only self-report, but also developmentallyappropriate measurements from peers, teachers, and parents, as well as data from direct observation). Research along these lines would help to clarify the developmental sequelae of early socially anxious symptomatology through adolescence and beyond, as youths experience changes in cognition (e.g., complex perspectivetaking), emotion (e.g., stress responses), and social relationships (e.g., intimate friendships).

Clinical implications

Present data reflect the scores of a nonclinical sample of children, but our analyses have important implications for understanding -and intervening in -clinical patterns. Most importantly, the concept of 'effective' social behavior needs to incorporate self-presentational knowledge and skills, explicitly addressing differentiation between audiences and responsiveness to audience characteristics. However, before treatment approaches relating to this issue can be formulated, further research is needed to illuminate the clinical manifestation of the self-presentational characteristics. The scores of our 'high social anxiety' group provide some indication of the self-presentational correlates of relatively high levels of social anxiety, but this group is not equivalent to clinically diagnosed social phobic groups, and further research is needed with such samples in order to evaluate more precisely the clinical significance of self-presentation. Such work would benefit from a detailed assessment of self-presentational behavior across different social interactional contexts: How do high self-reported use of self-presentational tactics and poor audience differentiation scores translate into behavior during specific interpersonal scenarios? Numerous researchers have conducted behavioral assessments of socially anxious children through role-play tasks (e.g., Morgan & Banerjee, 2006;Rao et al., 2007;Spence et al., 1999) but the assessments have typically comprised broad ratings of observed anxiety and social effectiveness and/or measures of eye contact, response latency, and response length. Our research suggests that valuable clinical insights may be obtained by careful profiling of the self-presentational tactics used by socially anxious children, as well as of the extent to which their tactics are responsive to audience attributes and preferences.

We believe that considerations of the self-presentational profile of youths presenting with social anxiety disorder will neatly complement and extend existing treatment programs. Clearly, interventions that address children's social skills deficits (e.g., Beidel, Turner, & Morris, 2000;Spence, Donovan, & Brechman-Toussaint, 2000) can be extended to include self-presentational features. For example, Beidel et al. (2000) describe the Social Skills Training component of their behavioral treatment program as covering skills such as greetings and introductions, skills for joining groups, starting conversations, etc. Research findings regarding specific patterns of tactic use and audience differentiation (or lack thereof) may play a crucial role in shaping the way in which these social skills are targeted in treatment. Moreover, because self-presentation by its nature involves cognitions about others' representations of the self (see Banerjee & Yuill, 1999b), research on this topic has implications for cognitive approaches to treating social anxiety disorder. Clark (2001) describes a cognitive approach to treatment focusing on performance-related assumptions and self-beliefs, as well as information-processing patterns prior to, during, and after social encounters. For example, Clark (2001) emphasizes the need to shift attention away from the self and towards the interaction and the social environment. In the context of such therapeutic goals, modifying beliefs about selfpresentational efficacy and diverting attention to audience attributes (that often vary from interaction to interaction) will surely occupy a central role. To summarize, present study's findings of a durable pattern of self-presentational characteristics -marked by higher self-reported use of tactics coupled with poorer differentiation between audiences -provide an important foundation for intervention work that targets socially anxious youths' behavior and cognition across diverse social settings and interactions.