Content area
Full Text
A newly developed measure of family harmony, the Scale of Family Atmosphere (SOFA), is described and evaluated. The brief 10-item questionnaire is time-efficient and uses language that is readily understood by respondents ranging in age from early to late adolescence. The SOFA was administered, with a battery of other scales, to a sample of 224 adolescents (106 males, 118 females), ranging in age from 12yrs to 18yrs (M=15yrs, SD=2.2yrs). Analyses of key psychometric properties of the SOFA indicate that the scale has good internal consistency (Cronbach alpha=.87), with all items loading on a single underlying factor. The validity of the scale is supported by its expected associations with measures of psychological wellbeing (anxiety, depression, self esteem), and the personality dispositions of neuroticism and psychoticism. Scores were found to be relatively independent of verbal ability and socially desirable response bias.
Numerous studies have reported statistically and socially significant links between adolescents' self-reports of family atmosphere and psychological adjustment. One of the most widely used measures of domestic atmosphere is the family environment scale (FES; Moos & Moos, 1986). The FES is one of ten social climate scales developed by Rudolf Moos and his colleagues and comprises 90 true-false items purporting to assess three dimensions of family environment described as relationships, personal growth and system maintenance. Ten subscales make up the three dimensions.
Despite its popularity, several surveys evaluating the FES have questioned its reliability and validity and thus the scale's usefulness as a research instrument (Boyd, Gullone, Needleman & Burt, 1997; Roosa & Beals, 1990a, b). Roosa and Beals summarised responses of 385 families and reported that five of the ten FES subscales yielded Cronbach alpha internal consistency values of less than .70. More recently Boyd and her colleagues (1997) administered the FES to 1,289 Australian adolescents. Alpha reliability estimates for this sample on the ten subscales ranged from .39 for Independence to .72 for Conflict. Only two of the ten subscales yielded alpha values in excess of .70. In the light of these data Boyd et al. (1997) argued that, regardless of the type of scale, reliability estimates of such magnitude are simply inadequate for research purposes (see also; Molloy, 2001; Nunnally, 1978; Pallant, 2001; Tabachnick & Fidell, 1996) and called for a revision...