Cronbach's alpha for the CAS was .92. The correlation between shyness and loneliness was .44,
p < .001. The mean for the Shyness Scale was 32.67 (SD = 9.2 ). The mean for UCLA
Loneliness was 42.09 (SD = 10.5), and the CAS mean was 50.06 (SD = 16.2). CAS factor means
were 26.03 (SD = 8.5), 16.30 ( SD = 6.7), and 7.73 (SD = 3.2), respectively. CAS scores were
inversely related to age, r = -.20, p < .01. Scores on the CAS also correlated significantly with
self-reported number of movies watched per week, r = .22, p < .01, and with self-reports of the
amount of general interest in celebrities, r = .40, p < .001. All of these measures are either
consistent with previous research or reasonable expectations. Taken together they suggest that:
our participants were similar to participants in earlier studies; our participants were reasonably
careful in filling out the materials; we scored them accurately; and, there is additional reason to
be confident about the validity of the CAS.
Shyness scores correlated .13 with total CAS scores, and .11, .13, and .15 with the three factor
scores, respectively. UCLA Loneliness scores correlated .05 with total CAS scores, and .05, -.02,
and .13 with the three factor scores, respectively. The largest of these correlation coefficients
narrowly failed to make the .05 alpha level.
For the participants who made "highly visible" favorite celebrity choices ( n = 75), shyness was
correlated .17 with total CAS, and .12, .16, and .21 with the three factor scores, respectively.
Loneliness scores correlated .17 with total CAS, and .12, .13, and .24 with the three factor
scores, respectively. All eight of these correlation coefficients are larger than their counterparts
based on the entire sample and the largest was significant at .05. However, these coefficients are
still weak, and even the one significant finding should be interpreted cautiously in light of the
large number of coefficients presented here.
For 36 participants whose favorite celebrities were chosen from "lower visibility" categories
(author, model, politics, religion, sports) shyness was correlated .11 with total CAS, and .02, .16,
and .21 with the three factor scores, respectively. Loneliness scores correlated -.03 with total
CAS, and -.13, -.09, and .26 with the three factor scores, respectively. In comparison to their
"high visibility" counterparts the trend was in the predicted direction (scores for "lower
visibility" participants were generally closer to .00), but so weak that we cannot rule out chance
as an explanation for our results. With the wisdom of hindsight it could be argued that some
sports heroes, some models, and some politicians are highly visible. Indeed it may be difficult to
become a celebrity without becoming highly visible.
Perse and Rubin (1990) found that lonely people made frequent use of highly visible media,
specifically movies and television. However, their data suggested that lonely people did this
primarily to pass the time rather than form or deepen any parasocial attachments.
[129]
---------------
[130]
Our results, in conjunction with those of Perse and Rubin (1990) and Rubin, Perse, and Powell
(1985), suggest that the link between two measures of social anxiety and the strength of a
parasocial interaction to a celebrity is either very weak or non-existent. There may be a few
persons who pursue parasocial relationships with celebrities in an attempt to alleviate shyness or