The Structure of Mental Health: Higher-order Confirmatory Factor Analyses of Psychological Distress and Well-being Measures

This paper addresses the question of whether psychological distress and subjective well-being are the opposite poles of the same axis of mental health or independent constructs that should be measured on two independent axes. The measures used in this study originate from a preliminary ethnosemantic study and the content analysis of narratives of psychological distress and well-being episodes experienced by a random sample of francophone Quebecers (Canada). Two scales were produced: a Psychological Distress Manifestation Scale (PDMS) based on 23 items and four factors (Anxiety/Depression, Irritability, Self-Depreciation, and Social Disengagement), and a Psychological WellBeing Manifestation Scale with 25 items and six factors (Self-Esteem, Social Involvement, Mental Balance, Control of Self and Events, Sociability, and Happiness). Structural equation modeling analyses confirm that these 10 factors can be viewed as components of two correlated dimensions (psychological distress and well-being) (r = −0.65) of a two-dimensional latent construct which reflects a higher-order concept of mental health. We conclude that assessment of mental health in general populations should use concomitant measures of psychological distress and well-being. It has been proposed that to increase the sensitivity and the precision of the mental health measures we should include, in mental health measurement, signs of psychological well-being that will distinguish among individuals who report almost perfect scores on measures of psychological distress (Veit and Ware, 1983). In fact, a substantial proportion of people in the general population report only a small percentage of symptoms included in psychological distress scales. Also, if measures of the frequency or intensity of distress symptoms, usually derived from clinical populations and diagnostic criteria, are better predictors of psychiatric disorders Social Indicators Research 45: 475–504, 1998. © 1998 Kluwer Academic Publishers. Printed in the Netherlands. 476 RAYMOND MASSÉ ET AL. in general, positive affects could be more useful in differential diagnosis. For example, it has been proposed that low scores on positive affect scales is “a critical factor in distinguishing depression from anxiety and other disorders” (Watson and Kendall, 1989: 21). Such a combination of psychological distress and well-being measures could also be important for national epidemiological studies interested in monitoring national health status and predicting demands in mental health care services. For now, it is not yet clearly known whether it is a deterioration in psychological distress or a decrease in the psychological well-being that is the best predictor of health-seeking behaviour. Few researches have addressed these issues notably because the time frame to report distress symptoms (usually one or two weeks) is shorter than that for well-being (from one to twelve months). Here lies a first methodological prerequisite to further researches. But a second more fundamental prerequisite is to question if psychological distress and well-being are opposite poles of a same continuum of mental health or if they do refer to two independent constructs and realities that should be measured on two independent axes? It has been proposed (Santé nationale and Bien-être Canada, 1988) that mental disorders and mental health are two interrelated but globally independent constructs that should be measured on two independent axis. A first axis based on the high or low prevalence of mental disorders expresses a gradation in level of incapacity and distress related to a given mental disorder. A second one oppose an optimal to a minimal mental health state and refers to an harmony or equilibrium between social, economic, professional or living conditions that permit optimization of mental capabilities. So even an individual presenting severe mental health disorders (i.e. a schizophrenic) with an adequate medication can show a good “mental health” if his (or her) life conditions and social life (access to a professional life, to a social support network, to adequate living conditions, etc.) are adequate. So high levels of distress symptomatology are not mechanically incompatible with high scores on life satisfaction, happiness, hedonic affect, cognitive appraisal of living conditions and other components of well-being. The goal of this paper will then be to test the hypotheses that psychological distress and well-being are two independent latent constructs based THE STRUCTURE OF MENTAL HEALTH 477 on specific first-order constructs of negative and positive manifestations of ill and well-being and that they are both related dimensions of a higher-order mental health construct. We will then answer the question of the relevance of measurement scales combining both positive and negative dimensions of mental health. Psychological Distress and Well-Being: Unidimensional and Multidimensional Hypotheses Dohrenwend and his colleagues (1980) have launched a debate on the patterns of inter-relationship between the dimensions of both psychological distress and well-being arguing for a singlefactor interpretation. The observed strong inter-correlations between measures of self-esteem, hopelessness, sadness, depression, anxiety and general well-being and their correlations with clinically assessed disorders support the unidimensional hypothesis. Referring to Frank (1973) they propose that all the scales used to assess mental health would refer to a global concept of demoralization or to a nonspecific psychological distress (Dohrenwend et al., 1980). Contrary to that position, others argue for the independence between distress and well-being (Goldberg et al., 1982; Veit and Ware, 1983). Bradburn (1969) has already interpreted high interitems correlations within groups of positive and negative items and the instability of the correlation of the positive and negative affect subscales with emotions such as anxiety and depression as proof that the Affect Balance Scale defines two distinct independent and uncorrelated dimensions. The reservations based on the fact that the independence between positive and negative scales is true for some specific factors (i.e stressful life events, physical activity, smoking) (McDowell and Praught, 1982) and that it could result from an inappropriate use of factor analysis (Van Schuur and Kruijtbosch, 1995) partially invalidates Bradburn’s independency hypothesis. Cherlin and Reeder (1975) suggested from an analysis of the Affect Balance Scale that this negative-positive axis hide other components of well-being that should be measured on a pleasantness-unpleasantness and an energy-tiredness axis. Finally, the transcultural validity of the independence between distress and positive affect is not demonstrated. For example, while it has been 478 RAYMOND MASSÉ ET AL. shown (Clark et al., 1981) that positive affect along with other firstorder constructs (such as depressed affect, somatic and retarded activities and interpersonal relations) refer to the same higher-order construct i.e., depressive symptomatology in U.S. subjects, this positive affect construct presents low correlations with these distress constructs and did not load on a second-order factor in a Japanese population (Iwata and Roberts, 1996). Veit and Ware (1983) offer an explanation for this dependenceindependence dilemma and concluded that a large third-order mental health factor underlies the positive and negative symptoms included in the highly correlated second-order construct of psychological distress and well-being. The goal of this paper is to test this hypothesis but from an original set of distress and well-being manifestations reported by a general population of french-speaking Quebecers (Canada) through lived episodes of psychological distress and well-being. In fact, we believe that the validity of this hypothesis testing rests on the content and construct validity of the dimensions of psychological distress and well-being that defined the virtual mental health construct. The Structure of Psychological Distress Psychological distress is defined as a nonspecific syndrome that covers constructs such as anxiety, depression, cognitive problems, irritability, anger or obsession-compulsion (Ilfeld, 1976; Préville et al., 1991; see Gotlib and Cane, 1989; Weissman et al., 1988 for a review of distress scales and their dimensions). Depression and anxiety are usually recognized as core distress syndromes that each have psychological and somatic components (Mirowsky and Ross, 1989; Ilfeld, 1976; Watson and Kendall, 1989). Sleep disorders, eating disorders, loss of energy, physical manifestations of stress are somatic symptoms also associated with clinical cases of depression and anxiety (APA, 1994; Kirmayer, 1984) and then included in distress scales. It has not been clearly established whether they are true components of psychological distress or simply physical consequences of it (Katon et al., 1991). Some argue that somatic complaints are instead concomitant manifestations but not components of a distress construct, and that they must not be included in distress measurements because of their correlation with both THE STRUCTURE OF MENTAL HEALTH 479 physiological and emotional problems (Wells, Golding and Burnam, 1988). This debate is still left open. Relying on exploratory and confirmatory factor analyses, it has been demonstrated (Préville et al., 1995; Martin et al., 1989; Veit and Ware, 1983; Zautra et al., 1988) that a second-order hierarchical model expressing a more general syndrome of “nonspecific psychological distress” represents the symptoms more adequately than a first-order model based on orthogonal or oblique solutions. In the same way, Tanaka and Huba (1984) have shown that although the primary factor structure of items from the Beck Depression Inventory and the Psychiatric Epidemiology Research Interview varied slightly across different samples, a second-order factor was always found that related strongly to physiological, cognitive and motivational manifestations of depressive and psychological distress symptomatology. So there seems to exist a virtual construct of psychological distress that structure various affective, cognitive and possibly somatic dimensions of ill-being. The Structure of Well-Being Even though psychological distress symptoms can be defined as negative reactions to recent life difficulties, they give an incomplete picture of people’s long-term mental health. A recent review of the progress and opportunities in assessing subjective wellbeing (Diener, 1994), proposed that people’s well-being must be conceived as, and assessed through measurement of, multiple cognitive (life satisfaction, morale) and affective (positive emotions, negative affects) components and that such a long-term well-being is a meaningful construct in terms of cross-situational consistencies as well as of temporal stabilities (Diener, 1994). Psychological wellbeing is generally considered as a component of quality of life scales (Parmentier, 1994) General well-being has also been defined as a balanced nourishment of the mind, body and spirit (Vella-Brodrick and Allen, 1995). According to Veit and Ware (1983) a general positive affect and emotional ties are dimensions of subjective well-being (and not objective well-being usually associated with quality of life concept, (Parmenter, 1994)). Okun and Stock (1987) define well-being as “an umbrella construct referring to the affective reactions of individuals 480 RAYMOND MASSÉ ET AL. to their life experiences along a positive-negative continuum” (1987: 481). It should have three subordinate components that involve specific time frame cognitive content: 1) life satisfaction as an evaluation of goal attainment, past oriented with strong cognitive content; 2) morale or morale condition toward discipline and confidence, future oriented with moderate cognitive content and 3) happiness as affective reaction toward daily life founded on positive and negative emotions, present oriented with low cognitive content. Happiness is then a major component of well-being. It refers to both an affective component expressing a “hedonic level of affect” (pleasant affective experiences outbalance unpleasant ones) and a cognitive component regarding the contentment or the perception that wants have been met (Veenhove, 1994). Independently of the debate on happiness as a fixed trait of individuals, either rooted in temperamental disposition, (Tellegen et al., 1988), in cognitive inclination or in acquired disposition (Costa et al., 1987), or in a fixed national character (Inkeles, 1990/91, cited by Veehoven, 1994), it has been convincingly proposed that happiness is no immutable trait. Inasmuch as it is defined as a positive evaluation of the quality of life-as-a-whole, not as an elated mood, happiness is dictated by living conditions and then, in spite of inner disposition, better society make people happier (Veehoven, 1994, 1998). Others stress the cognitive component of subjective well-being and define it as “a thoughtful appraisal of quality of life as a whole, a judgment of satisfaction with life” (Argyle, 1987: 5). On the whole, researches confirm (Diener, 1984) that subjective well-being is based both on an affective (hedonic) component and a cognitive component namely life satisfaction defined as “a global judgment that people make when they consider their life as a whole” (Diener, 1994).