The relationship between social anxiety and social support in adolescents: a test of competing causal models

Adolescence, Spring, 2005 by Robert J. Calsyn, Joel P. Winter, Gary K. Burger

Social anxiety, particularly in children and adolescents, has received considerable attention over the years (see Chorpita & Barlow, 1998 and Patterson & Ritts, 1997 for reviews). Social anxiety is correlated with a number of other negative behaviors and cognitive responses including avoidance of others, decreased verbal and nonverbal interaction in the presence of others, negative thoughts, negative self-evaluations, irrational beliefs, and evaluation apprehension (Patterson & Ritts, 1997). In addition, previous studies have indicated that anxiety was often a precursor to depression in adolescents (Cole, Peeke, Martin, Truglio, & Seroczynski, 1998; Reinherz, Giaconia, Pakiz, Silverman, Frost, & Lefkowitz, 1993).

Prior cross-sectional research also revealed that more anxious individuals report receiving less social support than do less anxious individuals (Caldwell & Reinhart, 1988; Haemmerlie, Montgomery, & Melchers, 1988). In discussing the relationship between social anxiety and social support, most researchers have employed a social causation model, i.e., they have assumed that lack of social support causes social anxiety (Johnson, 1991). The social selection model, on the other hand, argues that social anxiety causes social support (Johnson, 1991). According to this hypothesis, individuals who are more anxious are less able to attract and maintain supportive relationships than are less anxious individuals. A combined causal model, known as the reciprocal effects model, argues that the causal relationship between social support and social anxiety is largely reciprocal. Thus, although social support impacts subsequent social anxiety, social anxiety also affects subsequent social support (Stice & Barrera, 1995). This model is similar to the transactional model of Sameroff and Fiese (1990) proposed for younger children.

Longitudinal data on both social support and social anxiety is needed to adequately test these competing causal models. We are not aware of any longitudinal panel studies that have examined the causal relationship between social anxiety and social support. However, several longitudinal studies have compared the social causation versus the social selection model in examining the relationship between social support and other measures of psychological adjustment, particularly depression. One study with older adults supported the social causation model (Krause, Liang, & Yatomi, 1989). Two studies supported the social selection model, one study with middle age adults (Johnson, 1991) and the other study with older adults (Cutrona, Russell, & Rose, 1986). The combined reciprocal effects model was supported in three studies with adolescents (Compas, Wagner, Slavin, & Vannatta, 1986; Lakey, 1989; Stice & Barrera, 1995), and four studies with adults (Turner, 1981).

One contribution of this study is the distinction between two indices of social support: perceived support and enacted support. Perceived support is the perception that support would be available if needed, whereas enacted support is measured by asking respondents if certain supportive acts (e.g., loaned money or gave advice) were actually provided the respondent over a specified period of time. (See Barrera, 1986, for more detailed descriptions of enacted and perceived support.) With the exception of the Johnson study (1991) which assessed social networks contacts, all previous studies used a measure of perceived support in examining the causal relationship between social support and psychological functioning. None of the longitudinal studies included a measure of enacted support.

This study also seeks to further the research in this area by proposing a mediated reciprocal effects model. A mediated model tests whether a hypothesized cause-effect relationship can be better explained by specifying a construct that is more closely related to the outcome. We believe that perceived support mediates the effects in the reciprocal effects model for several reasons. First, in many prior studies, the correlation between perceived support and enacted support was quite low, indicating that these two variables were tapping somewhat different constructs (Lakey & Drew, 1997). In addition, prior research suggests that perceived support operates more as a cognitive schema which interprets the meaning of various acts (Lakey & Drew, 1997). For example, individuals with high perceived available support are more likely to interpret an ambiguous behavior of another as supportive, whereas persons with low perceived available support are more likely to interpret the same ambiguous behavior negatively (Lakey & Cassady, 1990). We hypothesize that perceived support functions as an intervening cognitive filter that interprets the meaning of both specific acts (e.g., enacted support) as well as emotional states (e.g., social anxiety).

The current study collected longitudinal data on social anxiety and social support from college students to test the reciprocal effects model, and to compare this model to a mediated reciprocal effects model. Thus, the current study provided an opportunity to see if the reciprocal effects model also provided the best explanation of the causal relationship between social anxiety and social support in adolescents and young adults, replicating the previous research on the causal relationship between depression and social support. Additionally, this study examined whether perceived support functions as a mediator between social anxiety and enacted support, and assesses the strength of the mediated (i.e., indirect) effects.