Shortly after research showed that psychological stress was associated with illness (Dohrenwend & Dohrenwend, 1974), researchers began to search for psychosocial characteristics that might moderate this stress-illness relation. It was in this search that the construct of hardiness (Kobasa, 1979a, 1979b) was introduced. According to theory, hardiness is a general quality that emerges from rich, varied, and rewarding childhood experiences (Maddi & Kobasa, 1984). This general quality manifests itself in feelings and behaviors that are characterized as commitment, control, and challenge. For example, Maddi and Kobasa (1984) wrote that hardiness is “a general sense that the environment is satisfying,” (p. 50) which leads a person to approach situations with curiosity and enthusiasm or commitment. According to Maddi and Kobasa, a hardy person views potentially stressful situations as meaningful and interesting (commitment), sees stressors as changeable (control), and sees change as a normal aspect of life rather than as a threat and views change as an opportunity for growth (challenge). By possessing these characteristics, a hardy person is able to remain healthy under stress.
Following its introduction, the construct of hardiness has generated considerable interest within the psychological and health sciences literature. Hardiness has been studied in an expanding range of contexts during the past decade. Early studies on hardiness focused on its relations with illness and with other known stress moderators, including Type A characteristics (Kobasa, Maddi, & Zola, 1983; Nowack, 1986; Rhodewalt & Agustsdottir, 1984; Schmied & Lawler, 1986), cynical hostility (Smith & Frohm, 1985), and social support (Ganellen & Blaney, 1984; Kobasa & Puccetti, 1983). The areas of study have expanded to include relations of hardiness with health concern (Hannah, 1988), cardiovascular reactivity (Contrada, 1989; Wiebe, 1991), optimism (M. F. Scheier & Carver, 1987), depressive symptoms (Funk & Houston, 1987), burnout (McCranie, Lambert, & Lambert, 1987; Topf, 1989), noise-induced stress (Topf, 1989), and immunoglobulin A levels in the breast milk of nursing mothers (Dillon & Totten, 1989). A variety of subject groups has also been studied, including attorneys, bankers, dentists, human services workers, law enforcement officers, nurses, teachers, single parents, nursing mothers, and the elderly.
Although a sizable body of research on hardiness exists, several fundamental issues remain unresolved. For example, several hardiness scales are currently in use. The question “Which scale is the most reliable and valid indicator of hardiness?” remains unanswered. There are conflicting opinions regarding whether hardiness is one or several characteristics. A recent discussion of multifaceted constructs (Carver, 1989) cast doubt on the adequacy of the most common way to categorize subjects as high or low in hardiness. Research studying the pathways through which hardiness exerts its effects has not been evaluated. Although critics have argued that hardiness does not buffer the effects of stress, others have speculated that hardiness buffers in specific circumstances: for working adults, for males, and in prospective analyses. Last, some authors have questioned the validity of the hardiness scales. Allred and Smith (1989), Funk and Houston (1987), and Rhodewalt and Zone (1989) suggested that hardiness scales inadvertently measure the negative personality characteristic of neuroticism (Costa & McCrae, 1985, 1987).
The purpose of this review is to address these fundamental and persistent issues in hardiness research. First, a comparison and an evaluation of available hardiness scales are presented. Second, information addressing the dimensionality of hardiness is examined. Third, the most common way of categorizing subjects as high or low in hardiness is evaluated, and alternatives are presented. Fourth, research studying causal pathways between hardiness and illness is reviewed and evaluated. Fifth, studies examining the neuroticism confound are examined, along with ways to approach this confound. Recommendations for future research are provided.
A COMPARISON OF HARDINESS SCALES
During the early years of hardiness research, 19 different scales were used to measure hardiness. Many of these scales did not discriminate between high-stress/high-illness and high-stress/low-illness subjects (Kobasa, 1979a, 1979b) and were subsequently dropped. Six scales remained, including the Alienation From Self (AFS) and the Alienation From Work (AFW) scales (Maddi, Kobasa, & Hoover, 1979) to measure commitment, the Powerlessness Scale (PS; Maddi et al., 1979) and the External Locus of Control Scale (ELOCS; Rotter, Seeman, & Liverant, 1962) to measure control, and the Security Scale (SS; Hahn, 1966) and the Cognitive Structure Scale (CSS; Jackson, 1974) to measure challenge. The CSS did not correlate strongly with the five remaining scales and was eventually dropped (Kobasa, Maddi, & Kahn, 1982). As a result, an inventory of 71 items became the most widely used measure of hardiness-the Unabridged Hardiness Scale (UHS). Scores from each of the five UHS subscales may be analyzed separately or combined into a Hardiness composite index. To form composite scores, raw subscale scores are standardized with respect to local norms (converted into z scores) and summed. Because challenge is measured by only one subscale, some researchers have doubled the Challenge score before adding it to those of the remaining subscales. Each subscale is used as a negative indicator of hardiness.
The development of the 71-item UHS was followed in 1982 by the creation of two short forms: the 20-item Abridged Hardiness Scale (AHS) and the 36-item Revised Hardiness Scale (RHS). Both short forms use UHS items, and both provide composite Hardiness measures. The RHS also provides separate Commitment, Control, and Challenge measures.
More recently, two “third-generation” hardiness scales have been developed: the 50-item Personal Views Survey (PVS; Hardiness Institute, 1985) and the 45-item Dispositional Resilience Scale (DRS; Bartone, Ursano, Wright, & Ingraham, 1989). These scales were intended to supersede previous scales and to rectify the problems of previous scales noted by critics (Bartone, 1989). They share the same format and have similar item content. Each DRS item appears in the PVS in a reworded or identical form. Both scales provide separate estimates for commitment, control, and challenge. Composite Hardiness scores are produced by adding raw scores for the three dimensions.
Both standard and nonstandard hardiness scales continue to proliferate. Some authors have created modified versions of the UHS, such as the Health-Related Hardiness Scale (Pollock, 1986). Others have used alternative scales to measure dimensions of hardiness (Nowack, 1986). In many instances, it is difficult or impossible to determine which hardiness scale a researcher has used from published information.
The proliferation of scales makes the body of hardiness research difficult to interpret. For example, it is difficult to determine whether differences in health outcomes across studies are real or reflect differences in the hardiness scales used. At first glance, the hardiness scales appear to be interchangeable. For example, PVS composite scores correlate highly with UHS Hardiness scores (r = .89 in Campbell, Amerikaner, Swank, & Vincent, 1989). Similarly, UHS Hardiness scores correlate highly with RHS Hardiness scores (r = .76 in Hull, Van Treuren, & Virnelli, 1987; r = .89 in Kobasa & Maddi, 1982, cited in Hull et al., 1987). However, the two short forms (AHS and RHS) are only moderately correlated (r = .50 in Allred & Smith, 1989). Allred and Smith (1989) found that, when they used these two scales to categorize subjects as high or low in hardiness, the categorizations derived from each scale were in agreement for only 69% of the subjects. Estimates of commitment, control, and challenge also differ significantly when measured by different scales. For example, Hull et al. (1987) found that the correlation between Challenge scores on the UHS and the RHS was moderate (r = .37). That the third-generation scales include many new items not found in the previous scales further limits comparability between scale versions.
A comparison of hardiness scale characteristics is shown in Table 1.
An examination of the table reveals that the third-generation scales have several improvements vis-à-vis the previous hardiness scales. Whereas the previous scales relied exclusively on negative indicators, the third-generation scales include positively keyed items. As shown in the table, the third-generation scales have internal consistency levels exceeding those of the AHS and the RHS. Because the new scales are not composed of preexisting scales like those used in the UHS, the practices of standardizing subscale scores and doubling the Challenge subscale score to create a Hardiness index have been eliminated. The use of raw scores also allows for the comparison of hardiness levels between subject samples.
Challenge as measured by the third-generation scales is different from challenge as measured by the previous scales. Although modest, internal consistency for Challenge on the third-generation scales has improved in comparison to that on the older RHS. Correlations of Challenge with Commitment or Control on the PVS are higher than those on the previous hardiness scales.
Although the third-generation scales are an improvement over the older hardiness scales, they have several shortcomings. Both types of scales include a majority of negatively keyed items. Although internal consistency is high for the Hardiness composite measures, internal consistency for Commitment, Control, and Challenge is less impressive (see Table 1). Test-retest reliability for the PVS is modest (rs > .60 over a 2-week period; S. Kahn, personal communication, September 1, 1986). Because hardiness theoretically develops during childhood and persists into adulthood (Maddi & Kobasa, 1984), hardiness measures would be expected to show higher stability over time. Information on DRS test-retest reliability is not available.
The DRS has several advantages over the other hardiness scales. It provides all the advantages of the third-generation scales already described. In addition, it has several advantages vis-à-vis the PVS: (a) It includes more positively keyed items, (b) it uses equal numbers of items to measure commitment, control, and challenge, and (c) items and scoring for this scale are readily available.
THE DIMENSIONALITY OF HARDINESS
As already noted, hardiness consists of three related dimensions: commitment, control, and challenge (Kobasa, 1979a, 1979b). According to Carver (1989), initial studies on hardiness generally analyzed these dimensions separately. Later studies typically combined Commitment, Control, and Challenge into a composite Hardiness index. The trend toward using a composite measure has been fostered by the fact that using a single index simplifies data analysis and the evaluation and interpretation of findings (Carver, 1989). This trend has also been fostered by the convenience of short scales that may not provide Commitment, Control, and Challenge measures. Although most authors have increasingly treated hardiness as unitary, some authors have questioned this tendency (Hull et al., 1987).
Whether hardiness has one or more dimensions is ultimately an empirical question. Factor analyses provide important information on the dimensionality of hardiness. A summary of factor-analytic studies is shown in Table 2.
An examination of Table 2 reveals that factor analyses of Hardiness items have consistently produced three factors. Studies that found three factors involved a variety of investigators, hardiness scales, and subject types.
In factor analyses of the AHS and the RHS, factor loadings have been consistent with hardiness theory. The majority of Commitment, Challenge, and Control items for the AHS (McNeil, Kozma, Stones, & Hannah, 1986) and the RHS (Hull et al., 1987) loaded together on their respective factors. UHS results (Hull et al., 1987) are somewhat inconsistent with hardiness theory. Items from two commitment (AFS and AFW) scales and one challenge scale (PS) loaded on a first factor. Items from one challenge scale (SS) and one control scale (ELOCS) loaded on a second factor. Only one challenge scale (CSS) loaded on a third factor. Unfortunately, one study that reported a factor analysis for the new scale, DRS (Bartone et al., 1989), did not include factor loadings.
In contrast to factor analyses of items, analyses of hardiness subscales (Funk & Houston, 1987; Kobasa, Maddi, & Kahn, 1982; Manning, Williams, & Wolfe, 1988) have not found three factors. Whereas two studies (Kobasa, Maddi, & Kahn, 1982; Manning et al., 1988) found a single factor, a third study (Funk & Houston, 1987) found two factors. A closer examination of studies that reported factor loadings reveals some agreement between these discrepant results. Funk and Houston (1987) found that the SS (challenge) and the ELOCS loaded on a second factor. Kobasa, Maddi, and Kahn (1982) found that the SS and the ELOCS loaded less highly on a single “General Hardiness” factor than did the remaining subscales. This pattern of subscale loadings is also consistent with the pattern of item loadings reported by Hull et al. (1987).
Commitment, Control, and Challenge correlations are a second source of information on the dimensionality of hardiness. As shown in Table 1, Commitment and Control have consistently been the most strongly related dimensions. Relations between Challenge and the other dimensions have been weak in comparison. Challenge, as measured by the RHS, has been unrelated or negatively related to Commitment or Control. When measured by the UHS, relations between Challenge and the other dimensions have been low in strength. Correlations between Challenge and the other dimensions have been moderate for the PVS.
In sum, hardiness scale items have generally formed three factors that are consistent with the conceptualization of hardiness. Hardiness dimensions have typically shown weak to moderate interrelations. Of the three dimensions, Commitment and Control have been the most strongly related. Although relations between Challenge and the remaining dimensions have historically been weak, they are higher for the new scale, the PVS.
SHOULD THE DIMENSIONS BE COMBINED?
That hardiness scale items form three factors raises the question of whether Commitment, Control, and Challenge should be combined into a single Hardiness index or studied separately. Although the use of a composite index simplifies the research process, it does so at the cost of conceptual clarity. If a researcher finds that high-hardy subjects are less ill than low-hardy subjects, this difference could be attributed to differences in commitment, control, or challenge or to some combination of these characteristics (Carver, 1989). Hardiness scores should be combined if the benefits of using composite scores outweigh the increase in conceptual ambiguity that occurs when dimensions are combined. For example, the use of a single composite score may be justified if hardiness is theoretically more important or interesting than its components (Carver, 1989). However, the theoretical importance of hardiness versus that of its components is difficult to determine and is largely a matter of opinion. Fortunately, the merits of using a composite index may be evaluated empirically.
If Hardiness dimensions are to be combined, they should be logically and empirically related (see Carver, 1989, for discussion). Because Challenge shares less than 10% of its variance with Commitment or Control when measured by the early hardiness scales, one could argue that it should not be combined with them. Because Challenge shares roughly 25% of its variance with Commitment and Control on the PVS, this argument may no longer be valid.
Another way to evaluate the Hardiness composite is to compare the strength of its relations with dependent measures (e.g., illness) against those of Commitment, Control, and Challenge (Carver, 1989). Hardiness should be more consistently and strongly related to various outcomes than should be any one dimension. If not, it would be more advantageous to study the dimensions separately. Results from one study (Contrada, 1989) indicate the relative importance of Hardiness and its dimensions. Contrada (1989) presented subjects with a mirror tracing task and found that only Challenge was related to diastolic blood pressure reactivity. He then conducted a series of analyses that compared results for Challenge to those of the Hardiness composite and to various combinations of subscales and dimensions. He concluded that none of these combinations added significantly to the ability of Challenge to predict differences in reactivity. If results similar to this were found for other outcomes and across several studies, it would suggest that the Hardiness composite offers no advantages over Challenge alone. Unfortunately, no other comparisons of this type have appeared in the literature to date.
The issue of whether Commitment, Control, and Challenge should be combined is paralleled by the issue of how they should be combined. Hardiness scores are most often formed by adding scores for the three dimensions. The resulting distribution of composite scores is then divided into high-hardy and low-hardy subject groups using a median split. Unfortunately, this way of operationalizing hardiness is not consistent with hardiness theory. As already described, hardy individuals have consistently been described as high in all three characteristics-commitment, control, and challenge (Kobasa, 1979a, 1979b; Maddi & Kobasa, 1984). If a researcher uses the procedures described here to form subject groups, his or her high-hardy subjects may not be high in all these qualities. By summing across dimensions and forming groups using a median split, it is possible for a high-hardy subject to be above the median for only one dimension but below the median for others. This way of categorizing subjects as high or low in hardiness creates empirical as well as theoretical problems. It results in groups of subjects that are heterogenous, thus making it more difficult to find significant relations between hardiness and health.
If the current way of categorizing subjects as high or low in hardiness ess is inadequate what alternatives could be used? A more more hardiness is inadequate, what alternatives could be used? A more appropriate way of defining high-hardy and low-hardy subjects involves the use of separate frequency distributions for Commitment, Control, and Challenge. Subjects whose scores are above appropriate cutoffs (e.g., medians) for each of the three distributions would be considered high in hardiness, and subjects who are below all three cutoffs would be considered low in hardiness. Another alternative involves viewing hardiness as a synergistic construct-one in which components build on one another (Carver, 1989). Using this approach, Hardiness composite scores would be formed by multiplying the terms representing Commitment, Control, and Challenge (Carver, 1989). Empirical tests are needed to determine which of these or other methods of forming the Hardiness measure are the most useful and appropriate.
CAUSAL PATHWAYS
According to Maddi and Kobasa (1984), stressful events lead to a strain reaction or increased sympathetic arousal. Chronic strain may eventually lead to exhaustion and accompanying illness or psychological distress. Hardiness modifies this strain-exhaustion process through several causal pathways. Hardiness (a) alters perceptions of events to make them less stressful, (b) leads to active or “transformational” coping, (c) influences coping indirectly through its influence on social support, and (d) leads to changes in health practices that in turn reduce illness. Each of these causal pathways is discussed and evaluated here.
One of the ways in which hardiness has been hypothesized to reduce strain and accompanying illness is through its effects on cognitive appraisals (Kobasa, 1979a). Rhodewalt and Agustsdottir (1984) and Rhodewalt and Zone (1989) found that, compared to low-hardy subjects, high-hardy subjects were more likely to view previous life events as positive and controllable. Pagana (1990) found that, compared to low-hardy medical students, high-hardy medical students evaluated their experiences as challenging and were less likely to view them as threatening. Although these three studies found relations between hardiness and appraisals, their retrospective time frame creates a directionality problem. Because subjects recalled events after they occurred, it is impossible to determine whether hardiness alters perceptions of events or the stress produced by events themselves led subjects to appear less hardy (Wiebe, 1991). Two recent studies used experimentally induced stressors to eliminate this problem. Both Allied and Smith (1989) and Wiebe (1991) found that, compared to low-hardy subjects, high-hardy subjects had more positive appraisals when faced with an evaluative threat.
If an event is appraised as stressful, personality characteristics may still buffer its effects by facilitating adaptive coping or by inhibiting maladaptive coping (Cohen & Edwards, 1989). High-hardy people theoretically use transformational coping-or change potentially stressful events into opportunities for growth; as a result, they cope in an optimistic and active way (Kobasa, 1982a; Maddi & Kobasa, 1984). In contrast, low-hardy people use regressive coping-avoiding or shrinking away from potentially stressful situations. Although a person who uses regressive coping temporarily avoids potential stressors, he or she dwells on them in a pessimistic way. A person who uses transformational coping and avoids regressive coping will cope effectively, reduce strain, and avoid illness.
The hypothesis that Hardiness dimensions are associated with coping is supported by Kobasa (1982a), who found that the use of regressive coping was correlated with Commitment. Bartone (1989) found that both hardiness and the use of avoidant coping discriminated between high-stress/low-illness and high-stress/high-illness bus drivers. In a study on psychological outcome, Pierce and Molloy (1990) found that the use of regressive coping was associated with increased “burnout” among secondary school teachers. These teachers also reported low levels of hardiness.
According to Maddi and Kobasa (1984), hardiness affects coping (a) directly and (b) indirectly through social support. High-hardy people tend to have relationships that support transformational coping in times of stress. Low-hardy people do not seek these relationships, and this increases their tendency to use regressive coping when stressed.
Curiously, no study has tested for mediated effects of hardiness and social support on coping, as outlined by hardiness theory. Two studies have tested whether social support moderates the effects of hardiness. Kobasa and Puccetti (1983) found a significant Hardiness × Family Support interaction on illness, but the Stress × Hardiness × Family Support interaction was not significant. Ganellen and Blaney (1984) found that Commitment and Challenge were significantly correlated with social support. However, a test for the Hardiness × Social Support interaction on depressive symptoms was not significant. (Ganellen & Blaney did not report results for a Stress × Hardiness × Social Support interaction.)
Theoretically, the positive appraisals and successful coping produced by hardiness reduce organismic strain and thus reduce illness. Several studies have examined relations between hardiness and strain under stress. Contrada (1989) found that, compared to low-hardy men, high-hardy men had smaller elevations in diastolic blood pressure during a mirror tracing task. Wiebe (1991) found that, compared to low-hardy men, high-hardy men had smaller elevations in heart rate during an evaluative threat task. Similarly, Lawler and Schmied (1987) found that, compared to women high in powerlessness (an indicator of control), women low in powerlessness were less reactive to a stressor. In contrast, Allred and Smith (1989) found that, compared to low-hardy men, high-hardy men had higher systolic blood pressure reactivity during an evaluative threat condition.
One final way that hardiness theoretically affects illness is through health practices (Maddi & Kobasa, 1984). According to this view, hardiness affects health practices such as exercise or sleep habits, which in turn affect illness. Wiebe and McCallum (1986) tested this hypothesis using path analysis and found that hardiness had (a) direct effects on illness and (b) indirect effects through health practices. Nagy and Nix (1989) found hardiness to be correlated with general health practices. In contrast to studies on general health practices, tests of the hypothesis that exercise mediates the effects of hardiness have produced disappointing results. Kobasa, Maddi, and Puccetti (1982) found that hardiness was not associated with exercise habits and that hardiness and exercise had additive but independent effects on illness. Results of other studies (Nagy & Nix, 1989; Roth, Wiebe, Fillingim, & Shay, 1989) have confirmed the independence of hardiness from exercise or fitness.
It is difficult to evaluate the merits of the various causal pathways between hardiness and illness. Few studies have examined each causal pathway. Of these studies, many used statistical techniques that did not test hardiness theory as outlined by its original proponents. For example, the hypothesis that hardiness affects illness indirectly through social support is appropriately tested using path analysis (Wiebe & McCallum, 1986). Because Nagy and Nix (1989) only correlated hardiness with general health practices, they did not actually test this causal pathway. The hypothesis that hardiness affects illness through the reduced use of regressive coping is also appropriately tested using path analysis. Because Bartone (1989) used discriminant analysis to study relations between hardiness and coping, he did not test this causal pathway. Studies that included appropriate tests have typically examined only parts of the causal chain between hardiness and symptoms. For example, Ganellen and Blaney (1984) studied relations between hardiness and social support but did not include coping-the critical link between social support and symptoms (Maddi & Kobasa, 1984)-in their analyses.
To summarize, results from a limited number of studies support the following tentative conclusions regarding causal pathways. Compared to low-hardy subjects, high-hardy subjects make less threatening appraisals of stressors; because only one study has linked hardiness, appraisals, and sympathetic activity, it would be premature to conclude that a hardy person’s decreased arousal under stress is due to more positive appraisals; hardiness is associated with a reduced use of regressive coping; because an absence of regressive coping does not imply the use of transformational coping, no study has tested the hypothesis that hardy individuals use transformational coping; no study has tested the hypothesis that hardiness affects coping indirectly through social support; the majority of studies show that, compared to low-hardy subjects, high-hardy subjects have smaller increases in sympathetic arousal (i.e., strain) during evaluative threat conditions; although hardiness may affect illness through general health practices, there is not enough information available to evaluate this potential pathway; hardiness does not affect illness indirectly through exercise or fitness.
DOES HARDINESS BUFFER STRESS?
Many authors have tested the overall hypothesis that hardiness buffers the effects of stress on illness or psychological distress rather than the specific causal pathways. A summary of published studies that tested for buffering effects is shown in Table 3.
The top portion of Table 3 summarizes studies on illness, and the bottom portion summarizes studies on psychological distress (e.g., burnout, depressive symptoms).
An examination of Table 3 shows that four studies have found buffering effects for hardiness on illness. One of these studies (Rhodewalt & Zone, 1989) differs significantly from the other tests in Table 3
in that the authors used (a) path mediation analysis and (b) negative life change to indicate stress. The other three studies that found buffering effects were conducted by Kobasa and associates. This observation suggests that these significant findings may be due to characteristics of the subjects studied by Kobasa et al. (e.g., working aged, male subjects) or to procedural factors that are unique to these studies.
Roth et al. (1989) suggested that hardiness may buffer stress among the working adults studied by Kobasa and associates but not among students. It is possible that, because students are relatively young and in good health, a floor effect for illness is produced. This floor effect may reduce variability in illness within the sample and make it more difficult to find significant buffering effects. An examination of subject types from Table 3 reveals that all but one of the studies that found buffering effects involved nonstudents. In addition, three of the four studies that involved students did not find buffering effects. Although these observations appear to support Roth et al.’s suggestion, closer examination of Table 3 reveals that the majority of studies with nonstudents did not find buffering effects. When all studies are considered, the hypothesis that hardiness buffers for working, aged adults is not supported.
Schmied and Lawler (1986) speculated that hardiness may buffer for men but not for women. Three of the four studies that found buffering effects involved men. However, in the majority of studies that involved only men, significant buffering effects were not found. Studies that involved only women are evenly split: Significant buffering effects were found in one study (Rhodewalt & Zone, 1989) but not in the other (Schmied & Lawler, 1986). When all studies are examined, the hypothesis that hardiness buffers for males is not supported.
Some authors have emphasized that buffering effects for hardiness have been found in prospective analyses (Maddi, Bartone, & Puccetti, 1987) and that these results provide strong support for the buffering hypothesis. However, of the four significant buffering effects that have been found, only one involved a prospective analysis. In addition, most of the retrospective and prospective studies have failed to find significant buffering effects for hardiness on illness. Consequently, when all studies are considered, there is no support for the view that buffering effects are more likely to be found in prospective analyses.
One procedural factor that might affect the results of retrospective studies is the length of time for which subjects are asked to recall physical symptoms. In the studies by Kobasa et al., subjects often reported symptoms that occurred during the previous 2 years (Kobasa et al., 1983) or 3 years (Kobasa, Maddi, & Puccetti, 1982). This contrasts with the reporting period for many of the studies that did not find buffering effects. In these studies, subjects reported prior illnesses during the previous 12 months (Schmied & Lawler, 1986), 6 months (Funk & Houston, 1987), and 1 month (Roth et al., 1989). Having subjects recall illnesses during a brief time period may reduce the average number of illnesses reported-resulting in a floor effect and making it more difficult to find significant buffering effects. However, it is also possible that recalling illnesses over an extended time period reduces the accuracy of reports and increases the influence of reporting biases such as those produced by neuroticism.
The bottom portion of Table 3 includes studies that tested for buffering effects of hardiness on psychological distress. Of the three studies that found significant buffering effects, all included female subjects. Two other studies that involved female subjects failed to find significant buffering effects. Curiously, in both these studies, the subjects were nurses. One could speculate that hardiness buffers the effects of stress on psychological symptoms only for females but that these effects do not occur for certain female subgroups such as nurses. It could be argued that the level of stress experienced by nurses is too high for buffering effects to have an impact. Because these suggestions are based on a handful of studies, they are speculative at best.
It could be argued that hardiness does not buffer stress and that the apparent buffering effects found in a few studies are an artifact. Experimentwise Type I statistical error is one possible explanation for the few buffering effects that have been found. As the number of statistical tests conducted within a study increases, the probability of finding a statistically significant relation by chance also increases. Many studies have included several tests for buffering effects, and those studies found one significant or marginally significant interaction among those multiple tests (Banks & Gannon, 1988; Ganellen & Blaney, 1984; Kobasa, Maddi, & Kahn, 1982; Kobasa, Maddi, & Zola, 1983).
The failure to consistently find moderating effects has led some authors to conclude that hardiness does not buffer stress (Funk & Houston, 1987; Hull et al., 1987). In contrast, these authors have argued that the effects of hardiness are predominantly main effects. According to this view, relations between hardiness and outcomes are independent of stress level. All but three of the studies in Table 3 that examined the effects of hardiness on illness found main effects, and all but one of the studies on psychological distress found main effects. Main effects have been found in studies that involved a variety of investigators, hardiness scales, and subject types. These studies also involved males and females, retrospective and prospective analyses.
In sum, there is little evidence that hardiness buffers the effects of stress on illness. Speculation that hardiness buffers under specific circumstances-for working adults, for males, and in prospective analyses-is not supported by the literature. The few significant buffering effects that have been found may be explained by Type I statistical error. For retrospective studies, it is possible that buffering effects are found when illnesses are recalled for multiple years. In contrast, main effects for hardiness on health and psychological distress have frequently been found.
HARDINESS AND THE NEUROTICISM CONFOUND
In recent years, the growing interest in hardiness has been paralleled by a growing concern that hardiness scales inadvertently measure neuroticism. Neuroticism is a broadly based aspect of normal personality that involves a tendency to experience negative affect including anxiety, dysphoria, hostility, and other characteristics (Costa & McCrae, 1985; McCrae, 1990). Within nonclinical populations, measures of anxiety, depression, and low self-esteem have been described as measuring aspects of neuroticism (McCrae, 1990). For this reason, they have been used as indicators of neuroticism (Allred & Smith, 1989). Individuals who are high in neuroticism tend to exaggerate somatic concerns and have frequent somatic complaints, although their actual health does not differ from that of others (Allred & Smith, 1989; Costa & McCrae, 1985). The notions associated with neuroticism have also been referred to as negative affectivity (Watson & Clark, 1984), general dysphoria (Gotlib, 1984), and maladjustment (Funk & Houston, 1987). An absence of neuroticism could be referred to as adjustment. I use the term neuroticism, as suggested by Allred and Smith (1989), due to its established relation with somatic complaints.
There are several reasons to argue that the hardiness scales measure neuroticism. These reasons include the contents of hardiness scales, the degree to which negative indicators are used to measure hardiness, and the fact that most studies have found only main effects for hardiness on illness.
In a critical analysis of hardiness research, Funk and Houston (1987) argued that the UHS resembled neuroticism scales. They pointed out that the subscales used to measure hardiness were originally designed for other purposes (e.g., to measure alienation) and that they resembled maladjustment scales. They also argued that hardiness scale items were similar to maladjustment scale items. Although the third-generation hardiness scales are not composed of preexisting scales, the criticism that hardiness scale items tap neuroticism still applies. Several problematic items from the new DRS are shown in Table 4. Items from this 45-item scale (Bartone et al., 1989) rather than from the 50-item PVS are presented because the PVS is copyrighted. All the items presented in the table have counterparts in the PVS. An examination of the table reveals several hardiness items similar to those used to measure neuroticism. Although some hardiness items presented in the table do not have direct counterparts in neuroticism scales, they tap depressive or hostile aspects of neuroticism or bear little conceptual resemblance to Commitment, Control, or Challenge.
Those who argue that hardiness scales measure neuroticism cite the heavy reliance on negative indicators to measure hardiness. Measuring hardiness through negative indicators may be inappropriate because “it entails an attempt to index high levels of one characteristic (e.g., commitment) through low scores of another (e.g., alienation)” (Funk & Houston, 1987, p. 573). It has been argued that, because the third-generation scales include positively keyed items, the previous reliance on negative indicators has been corrected (Bartone, 1989). However, the new scales include a majority of negatively keyed items as shown in Table 1. As a result, the criticism concerning the reliance on negative indicators to measure hardiness still applies.
That most studies have found only main effects for hardiness on illness supports the argument that hardiness scales measure neuroticism. Hardiness research has relied heavily on self-report measures of illness that are prone to the reporting biases produced by neuroticism (McCrae, 1990). If hardiness scales tap neuroticism, they could be correlated with symptom reports only because neuroticism and somatic complaints are correlated. In other words, because hardiness research has relied on self-report symptom measures, it is not possible to determine whether low-hardy subjects are more ill or merely have more somatic complaints.
The ultimate means of determining whether hardiness research is confounded by neuroticism is through empirical research. Studies that included results that are relevant to the neuroticism confound appear in Table 5.
As shown in the table, several studies have reported significant correlations between hardiness and neuroticism. These studies include a variety of investigators, hardiness scales, and neuroticism indicators. The magnitude of correlations between hardiness and neuroticism across the various studies is moderate. This indicates that, although hardiness and neuroticism overlap, they are not identical. However, correlations between hardiness dimensions have typically been low to moderate in strength, as already described. In fact, the correlations between hardiness and trait anxiety found by Allred and Smith (1989) were of the same magnitude as the correlation between the two hardiness scales used in that study.
In a study using the UHS, Funk and Houston (1987) reported that, whereas commitment and control were significantly correlated with neuroticism, challenge was not. Although the third-generation measure of challenge has improved in comparison to earlier measures, it has changed in a way that makes it more susceptible to the confounding influence of neuroticism. Challenge as measured by the new PVS is significantly correlated with neuroticism (Parkes & Rendall, 1988).
Although hardiness is correlated with neuroticism indicators, it is possible that relations between hardiness and illness are independent of neuroticism. One way to test for possible independence is to introduce neuroticism scores as a statistical control into analyses of hardiness and illness. Funk and Houston (1987) found that relations between hardiness and illness could not be replicated when neuroticism was statistically controlled. This result was obtained in two separate analyses using scores from two maladjustment scales. Rhodewalt and Zone (1989) also eliminated relations between hardiness and illness by introducing controls for depression.
Others have extended tests of the neuroticism confound to studies that involved outcomes other than illness. Allred and Smith (1989) found that relations of hardiness with state anxiety and negative thoughts were eliminated when trait anxiety scores were entered as a statistical control. In some analyses, however, controls for neuroticism did not eliminate effects for hardiness on outcomes other than illness. In prospective analyses, Funk and Houston (1987) found that main effects of hardiness on depressive symptoms remained after controlling for maladjustment. Allred and Smith (1989) found that relations between hardiness and positive self-statements remained after neuroticism was statistically controlled and that relations between hardiness and blood pressure reactivity remained marginally significant when neuroticism was controlled.
In sum, research has shown that both old and new hardiness measures tap neuroticism. Hardiness scale items resemble neuroticism scale items. Several studies have found that scores from hardiness and neuroticism scales are significantly correlated. Results from two studies show that the effects of hardiness on illness are eliminated when neuroticism scores are entered as a statistical control. (It is important to note that these results are for old scale versions and that they have not been replicated using the third-generation scales.) In contrast, effects of hardiness on outcomes other than illness have been eliminated by the introduction of controls for neuroticism in some analyses but not in others.
The neuroticism confound presents a serious obstacle to hardiness research. It is clear that the hardiness scales measure neuroticism. The important issue for future hardiness and health psychology research is how to approach this confound. The typical approach for dealing with neuroticism has been to treat it as a nuisance variable and to statistically control for its effects (Allred & Smith, 1989; Funk & Houston, 1987; Rhodewalt & Zone, 1989). Other possible “nuisance variable” approaches include removing neuroticism-loaded items from self-report measures of stress and illness, using measures that are not strongly affected by neuroticism (e.g., peer reports or laboratory results), using longitudinal designs, or eliminating from statistical analyses subjects high in neuroticism (McCrae, 1990).
Although controlling for the influence of neuroticism allows for independent assessment of the effects of hardiness, this approach is not without its problems. In general, treating neuroticism as a nuisance variable prevents a better understanding of the relations between stress and illness. If neuroticism is systematically removed from causal models, several potentially important questions will remain unanswered. For example, what aspects of neuroticism overlap with hardiness? Does neuroticism mediate the effects of hardiness or vice versa? Are people high in neuroticism “disease prone” (Friedman & Booth-Kewley, 1987) or vulnerable to stress (McCrae, 1990)? Do individuals high in neuroticism create personal circumstances that engender stress (Rhodewalt & Zone, 1989)? Are hardiness and neuroticism manifestations of a “superordinate buffer”-a higher order characteristic that buffers stress (Cohen & Edwards, 1989)? The examination of neuroticism may prove more fruitful in the long run than controlling for its influence.
CONCLUSIONS AND RECOMMENDATIONS
Hardiness research has been hampered by several fundamental shortcomings: measurement problems, the failure to adequately test hardiness theory, the failure to find predicted results, and the neuroticism confound. Many of these shortcomings have a common origin: the fact that hardiness theory has been poorly operationalized. Hardiness researchers have generally favored convenient, simple, and familiar research strategies at the expense of scientific rigor. These biases are exemplified in the original adoption of preexisting scales to measure hardiness (as opposed to constructing scales to fit hardiness theory). They are reflected in the use of simple but questionable methods of categorizing subjects with respect to hardiness. They are also reflected in the use of familiar statistical techniques that do not adequately test hardiness theory. If hardiness is to remain a credible scientific construct, its proponents face several difficult challenges.
First, the existence of hardiness must be justified. A theoretical rationale detailing why hardiness is more meaningful and important than its component parts is needed. In addition, this rationale should explain why hardiness cannot be understood by studying commitment, control, and challenge separately. After a theoretical rationale for the existence of hardiness is provided, the usefulness of hardiness should be demonstrated empirically. For example, studies should test whether hardiness explains more variance in outcomes than its theoretical components. Until the utility of hardiness is established, researchers should report separate results for commitment, control, and challenge.
In their efforts to test hardiness theory, researchers should adopt a standard hardiness measure. The use of multiple scales and nonstandard scales has created a body of research that is difficult to interpret. For the immediate future, researchers may wish to use the 45-item Dispositional Resilience Scale because it has several advantages over alternative scales. In the long run, hardiness proponents should further modify the hardiness scales. The number of positively keyed items should be increased. Items that tap neuroticism should be reworded or eliminated. Efforts should also be made to improve the reliability of the Commitment, Control, and Challenge subscales.
In addition to changes in scale contents, changes are needed in the way subjects are categorized as high or low in hardiness. The practice of adding Commitment, Control, and Challenge scores and dividing the resulting distribution with a median split results in groups of subjects that are not high or low in hardiness as defined by theory. As a result, alternative ways of categorizing subjects should be tested.
If hardiness research is to continue, researchers must ensure that their studies actually test hardiness theory. Many of the causal pathways outlined by hardiness theorists involve moderating or mediating effects. As a result, researchers should avoid using simple correlations and discriminant analyses in their attempts to test theory. Instead, they should use (a) interactive tests to examine possible buffering effects or (b) techniques such as path analysis to examine mediated or indirect effects of hardiness. In addition, researchers who include mediating variables in their theoretical models should include measures of those variables in their research designs. For example, if hardiness is hypothesized to reduce illness through cognitive appraisals, a measure of appraisals should be included (e.g., Wiebe, 1991).
Tests of the primary hypothesis that hardiness buffers the effects of stress on illness have produced lackluster results. Studies that have found significant stress-buffering effects for hardiness are the exception rather than the rule. Among studies that have found buffering effects, effects were found in one of several tests. Potential explanations for the failure to find buffering effects-that hardiness buffers for working adults, for males, or in prospective analyses-are not supported by the literature. Statistical error or procedural factors are more plausible explanations for the few significant buffering effects that have been found. There is no compelling scientific reason for maintaining the hypothesis the hardiness buffers stress. Instead, the results of research to date strongly suggest that the buffering hypothesis should be discarded. As a result, researchers who wish to study hardiness should consider alternative causal models.
Research on hardiness has accelerated in spite of criticism and of findings that fail to confirm the buffering hypothesis. As a result, it is reasonable to ask why the study of hardiness is so hardy. Hardiness reflects a fundamental perspective within psychology-one that emphasizes mental health over psychopathology. This emphasis on health is evident in the writings of humanistic theorists (e.g., Maslow, 1970). In fact, the descriptions of hardy and self-actualizing persons are very similar (Campbell et al., 1989). Self-actualizing people have a mission or focus outside themselves (i.e., commitment). They feel a sense of control over events and anticipate change as exciting or challenging.
In this light, it is ironic that hardiness scales tend to measure negative characteristics such as alienation, insecurity, powerlessness, and lack of personal control. It is not surprising, then, that critics of hardiness research have deemphasized healthy personality characteristics and pointed out the relation between hardiness and the more pathological characteristic of neuroticism. In short, the controversy concerning what hardiness is can be more broadly viewed as a debate over whether positive or negative characteristics have a greater influence on health. At this time, hardiness research does not provide information that could end this great debate.
Acknowledgements
Corresponding Author
I thank Virginia Blankenship, Cheri Brown, Erin Driver, Alice Mariahazy, Andrea Peterson, and Kristin Syverson for their assistance in conducting the literature review and in preparing the manuscript.Correspondence concerning this article should be addressed to Requests for reprints should be sent to Steven C. Funk, Department of Psychology, P.O. Box 15106, Northern Arizona University, Flagstaff, AZ 86011-5106.
Allied, K. D., & Smith, T. W. (1989). The hardy personality: Cognitive and physiological responses to evaluative threat. Journal of Personality and Social Psychology, 56, 257–266.
Banks J. K., & Gannon, L. R. (1988). The influence of hardiness on the relationship between stressors and psychosomatic symptomatology. American Journal of Community Psychology, 16, 25–37.
Bartone, P. T. (1989). Predictors of stress-related illness in city bus drivers. Journal of Occupational Medicine, 32, 657–663.
Bartone, P. T., Ursano, R., Wright, K., & Ingraham, L. (1989). The impact of military air disaster on the health of assistance workers. Journal of Nervous and Mental Disease, 177, 317–328.
Beck, A. T., Ward, C. H., Mendelson, M. Mock, J., & Erbaugh, J. (1961). An inventory for measuring depression. Archives of General Psychiatry, 4, 561–571.
Bell, H. M. (1938). The Adjustment Inventory: Adult Form. Stanford, CA: Stanford University Press.
Campbell, J. M., Amerikaner, M., Swank, P., & Vincent, K. (1989). The relationship between the hardiness test and the Personal Orientation Inventory. Journal of Research in Personality, 23, 373–380.
Carver, C. S. (1989). How should multifaceted personality constructs be tested? Issues illustrated by self-monitoring, attributional style, and hardiness. Journal of Personality and Social Psychology, 56, 577–585.
Cohen, S., & Edwards, J. R. (1989). Personality characteristics as moderators of the relationship between stress and disorder. In R. Neufeld (Ed.), Advances in the investigation of psychological stress (pp. 235–283). New York: Wiley.
Contrada, R. (1989). Type A behavior, personality hardiness, and cardiovascular responses to stress. Journal of Personality and Social Psychology, 57, 895–903.
Cook, W. W., & Medley, D. M. (1954). Proposed Hostility and Pharisaic-Virtue scales for the MMPI. Journal of Applied Psychology, 38, 414–418.
Costa, P. T., Jr., & McCrae, R. R. (1985). Hypochondriasis, neuroticism, and aging: When are somatic complaints unfounded?American Psychologist, 40, 19–28.
Costa, P. T., Jr., & McCrae, R. R. (1987). Neuroticism, somatic complaints, and disease: Is the bark worse than the bite?Journal of Personality, 55, 299–316.
Derogatis, L. R., Lipman, R. S., Rickels, K., Uhlenhuth, E. H., &Covi, L. (1974). The Hopkins Symptom Checklist: A self-report symptoms inventory. Behavioral Science, 19, 1–14.
Dillon, K. M., & Totten, M. C. (1989). Psychological factors, immunocompetence, and health of breast-feeding mothers and their infants. Journal of Genetic Psychology, 150, 155–162.
Dohrenwend, B. S., & Dohrenwend, B. P. (Eds.). (1974). Stressful life events: Their nature and effects. New York: Wiley.
Eysenck, H. J., & Eysenck, S. (1968). Manual for the Eysenck Personality Inventory. San Diego: Educational and Industrial Testing Service.
Fitts, W. H. (1964). Tennessee Self Concept Scale. Nashville: Counselor Recordings and Tests.
Friedman, H. S., & Booth-Kewley, S. (1987). The “disease-prone personality”: A meta-analytic view of the construct. American Psychologist, 42539–555,
Funk, S. C., & Houston, B. K. (1987). A critical analysis of the Hardiness Scale’s validity and utility. Journal of Personality and Social Psychology, 53, 572–578.
Ganellen, R. J., & Blaney, P. H. (1984). Hardiness and social support as moderators of the effects of life stress. Journal of Personality and Social Psychology, 47, 156–163.
Gotlib, I. H. (1984). Depression and general psychopathology in university students. Journal of Abnormal Psychology, 93, 19–30.
Hahn, M. E. (1966). California Life Goals Evaluation Schedule. Palo Alto, CA: Western Psychological Services.
Hannah, E. T. (1988). Hardiness and health behavior: The role of health concern as a moderator variable. Behavioral Medicine, 14, 59–63.
Hardiness Institute. (1985). Personal Views Survey. Arlington Heights, IL: Author.
Hull, J. G., Van Treuren, R. R., & Virnelli S. (1987). Hardiness and health: A critique and alternative approach. Journal of Personality and Social Psychology, 53, 518–530.
Jackson, D. N. (1974). Personality Research Form manual. Goshen, NY: Research Psychologists Press.
Kleinmuntz, B. (1960). Identification of maladjusted college students. Journal of Counseling Psychology, 7, 209–211.
Knapp, R. R. (1976). Handbook for the Personal Orientation Inventory. San Diego: Education and Industrial Testing Service.
Kobasa, S. C. (1979a). Personality and resistance to illness. American Journal of Community Psychology, 7, 413–423.
Kobasa, S. C. (1979b). Stressful life events, personality, and health: An inquiry into hardiness. Journal of Personality and Social Psychology, 37, 1–11.
Kobasa, S. C. (1982a). Commitment and coping in stress resistance among lawyers. Journal of Personality and Social Psychology, 42, 707–717.
Kobasa, S. C. (1982b). The hardy personality: Toward a social psychology of stress and health. In G. S. Sanders & J. Sals (Eds.), Social psychology of health and illness (pp. 3–32). Hillsdale, NJ: Lawrence Erlbaum Associates, Inc.
Kobasa, S. C., Maddi, S. R., & Courington, S. (1981). Personality and constitution as mediators in the stress-illness relationship. Journal of Health and Social Behavior, 22, 368–378.
Kobasa, S. C. Maddi, S. R., & Kahn, S. (1982). Hardiness and health: A prospective study. Journal of Personality and Social Psychology, 42, 168–177.
Kobasa, S. C., Maddi, S. R., & Puccetti, M. C. (1982). Personality and exercise as buffers in the stress-illness relationship. Journal of Behavioral Medicine, 5, 391–404.
Kobasa, S. C. Maddi, S. R., & Zola, M. A. (1983). Type A and hardiness. Journal of Behavioral Medicine, 6, 41–51.
Kobasa, S. C., & Puccetti, M. C. (1983). Personality and social resources in stress resistance. Journal of Personality and Social Psychology, 45, 839–850.
Lanyon, R. I. (1970). Development and validation of a psychological screening inventory. Journal of Consulting and Clinical Psychology, 35, 1–24.
Lawler, K. A., & Schmied, L. A. (1987). The relationship of stress, Type A behavior and powerlessness to physiological responses in female clerical workers. Journal of Psychosomatic Research, 31, 555–563.
Levenson, H. (1974). Activism and powerful others: Distinctions within the concept of internal-external control. Journal of Personality Assessment, 38, 377–383.
Maddi, S. R., Bartone, P. T., & Puccetti, M. C. (1987). Stressful events are indeed a factor in physical illness: Reply to Schroeder and Costa (1984). Journal of Personality and Social Psychology, 52, 833–43.
Maddi, S. R., & Kobasa S. C. (1984). The hardy executive: Health under stress. Homewood, IL: Dow Jones-Irwin.
Maddi, S. R., Kobasa, S. C, & Hoover, M. (1979). An alienation test. Journal of Humanistic Psychology, 19, 73–75.
Manning, M. R., Williams, R. F., & Wolfe, D. M. (1988). Hardiness and the relationship between stressors and outcomes. Work and Stress, 2,205–216.
Maslow, A. (1970). Motivation and personality. New York: Harper & Row.
McCrae, R. (1990). Controlling neuroticism in the measurement of stress. Stress Medicine, 6, 237–241.
McCranie, E. W.,Lambert, V. A., & Lambert, C. E. (1987). Work stress, hardiness, and burnout among hospital staff nurses. Nursing Research, 36, 374–378.
McNeil, K., Kozma, A., Stones, M. J., & Hannah, E. (1986). Measurement of psychological hardiness in older adults. Canadian Journal on Aging, 5, 43–48.
Nagy, S., & Nix, C. L. (1989). Relations between preventive health behavior and hardiness. Psychological Reports, 65, 339–345.
Nowack, K. M. (1986). Type A, hardiness, and psychological distress. Journal of Behavioral Medicine, 9, 537–548.
Pagana, K. D. (1990). The relationship of hardiness and social support to student appraisal of stress in an initial clinical nursing situation. Nursing Education, 29, 255–261.
Parkes, K. R., & Rendall, D. (1988). The hardy personality and its relationship to extraversion and neuroticism. Personality and Individual Differences, 9, 785–790.
Pierce, C. M., & Molloy, G. N. (1990). Psychological and biographical differences between secondary school teachers experiencing high and low levels of burnout. British Journal of Educational Psychology, 60, 37–51.
Pollock, S. E. (1986). Human responses to chronic illness: Physiologic and psychosocial adaptation. Nursing Research, 35, 90–95.
Rhodewalt, F., & Agustsdottir S. (1984). On the relationship of hardiness to the Type A behavior pattern: Perception of life events versus coping with life events. Journal of Research in Personality, 18, 212–223.
Rhodewalt, F., & Zone, J. B. (1989). Appraisal of life change, depression, and illness in hardy and nonhardy women. Journal of Personality and Social Psychology, 56, 81–38.
Roth, D. L., Wiebe, D. J., Fillingim, R. B., & Shay, K. A. (1989). Life events, fitness, hardiness, and health: A simultaneous analysis of proposed stress-resistance effects. Journal of Personality and Social Psychology, 57, 136–142.
Rotter, J. B., Seeman, M., & Liverant, S. (1962). Internal vs. external locus of control of reinforcement: A major variable in behavior theory. In N. F. Washburne (Ed.), Decisions, values, and groups (pp. 473–516). London: Pergamon.
Scheier, I. H., & Cattell, R. B. (1961). Manual for the Neuroticism Scale Questionnaire. Champaign, IL: Institute for Personality and Ability Testing, Inc.
Scheier, M. F., & Carver, C. S. (1987). Dispositional optimism and physical well-being: The influence of generalized outcome expectancies on health. Journal of Personality, 55, 169–210.
Schmied, L. A., & Lawler, K. A. (1986). Hardiness, Type A behavior, and the stress—illness relationship in working women. Journal of Personality and Social Psychology, 51, 1218–1223.
Sipprelle, R. C. Gilbert, F. S., & Ascough, J. A. (1976). The Affect Rating Scale: Unipolar, semantic space components of affective state. Unpublished manuscript, Purdue University, West Lafayette, IN.
Smith, T. W. & Frohm, K. D. (1985). What’s so unhealthy about hostility? Construct validity and psychosocial correlates of the Cook and Medley Ho scale. Health Psychology, 4, 503–520.
Spielberger, C. D., Gorsuch, R. L., & Lushene, R. E. (1970). Manual for the State-Trait Anxiety Inventory. Palo Alto, CA: Consulting Psychologists Press.
Topf, M. (1989). Personality hardiness, occupational stress, and burnout in critical care nurses. Research in Nursing and Health, 12, 179–186.
Watson, D., & Clark, L. A. (1984). Negative affectivity: The disposition to experience aversive emotional states. Psychological Bulletin, 96, 465–490.
Wiebe, D. J. (1991). Hardiness and stress moderation: A test of proposed mechanisms. Journal of Personality an d Social Psychology, 60, 89–99.
Wiebe, D. J., & McCallum D. M. (1986). Health practices and hardiness as mediators in the stress-illness relationship. Health Psychology, 5, 425–438.
Zuckerman, M. (1964). Development of a Sensation Seeking Scale. Journal of Consulting and Clinical Psychology, 28, 477–482.
© 1992. Notwithstanding the ProQuest Terms and Conditions, you may use this content in accordance with the associated terms available at https://publishing.aip.org/publications/journals/covid-19/.
