Psychometric properties of the Swedish version of the satisfaction with life scale in a sample of individuals with mental illness

Background: Health assessment among individuals with mental health problems often involves measures of ill-being (e.g., anxiety, depression). Health is, however, defined as a state of physical, mental and social well-being and not merely the absence of disease (WHO, 1948, 2001). Hence, in order to address mental illness during the 21st century, we need to develop methods for the prevention, identification and treatment of mental illness; but also, for the promotion, identification, and maintenance of well-being. In this context, over three decades of subjective well-being research have resulted in the development of measures of positive aspects of human life, such as the Satisfaction with Life Scale (Diener et al., 1985). Our aim was to investigate the psychometric properties of the Satisfaction with Life Scale in a Swedish population of individuals with mental illness using both Classical Test Theory (CTT) and Item Response Theory (IRT). Method: A total of 264 participants (age mean = 43.46, SD = 13.31) diagnosed with different types of mental illness answered to the Swedish version of the Satisfaction with Life Scale (five items, 7-point scale: 1 = strongly disagree, 7 = strongly agree). Results: We found positive and significant relationships between the five items of the scale (r ranging from 0.37 to 0.75), good reliability (Cronbach's alpha = 0.86), and that the one-factor solution had best goodness of fit (loadings between 0.52-0.88, p < 0.001). Additionally, there were no significant differences in comparative fit indexes regarding gender and occupation status. All items had high discrimination values (between 1.95-3.81), but item 5 ("If I could live my life over, I would change almost nothing"); which had a moderate discrimination value (1.17) and the highest estimated difficulty on response 7 (3.06). Moreover, item 2 ("The conditions of my life are excellent") had less discrimination and redundant difficulty with both item 1 ("In most ways my life is close to my ideal"; 2.03) on response 7 and with item 3 ("I am satisfied with my life"; -1.21) on response 1. The five items together provided good information, with especial good reliability and small standard error within -1.00 up to about 2.00 and the highest amount of test information at 0.00 of the level of life satisfaction within this population. Conclusions: Consistent with previous research, the scale had good reliability and provided good information across most of the latent trait range. In addition, within this population, sociodemographic factors such as gender and occupation status do not influence how individuals respond to the items in the scale. However, the items couldn't measure extreme levels of low/high life satisfaction. We suggest replication of these findings, the test of additional items, and the modification of items 2 and 5 in order to use the scale among individuals with mental illness.