Attributional style for negative events: a proposition for a more reliable and valid measure of attributional style.
Seligman (1990) refers to attributional style as a developmentally acquired personality characteristic and proposes two types of style, the optimistic explanatory style (OES) and the pessimistic explanatory style (PES). People who are characterized by an OES usually attribute failures to external, unstable and specific causes and successes to internal, stable and global ones. On the other hand, people who have a PES often attribute bad events to internal, stable and global factors and good events to external, unstable and specific causes. Seligman (1990) also suggests that attributional style 'stems from your view of your place in the world - whether you think you are valuable and deserving, or worthless and hopeless' (p. 44), directly relating attributional style to self-esteem.
There are, however, a number of studies which, although they did not directly address the issue of whether a person who attributes negative events to internal, stable and global factors will make external, unstable and specific attributions for positive events, provided a number of findings that are not consistent with this theoretical assumption of the reformulated model. Hull & Mendolia (1991) used structural modelling techniques to test the mediational role of expectancies in the relation between attributions and depressive affect. Because the specification of attributional style was particularly problematic within Lisrel VI, they modelled it separately in an attempt to specify appropriately its structure. The best fitting model hypothesized that attributions for positive and negative events do not form a single latent variable. Moreover, Hull & Mendolia's (1991) finding that attributions for positive outcomes were only indirectly related to depression by means of expectancies, whereas attributions for negative outcomes were both directly and indirectly related to depression, can be interpreted as an indication that attributional style for positive and negative outcomes are separate variables.
Ahrens & Haaga (1993), in a study of attributional style and expectancies to the measures of affectivity, anxiety and depression, provide an intercorrelation between the scores on the six positive scenarios drawn from the Attributional Style Questionnaire (ASQ; Peterson, Semmel, von Baeyer, Abramson, Metalsky & Seligman, 1982) and the scores on the Expanded Attributional Style Questionnaire (EASQ; Peterson & Villanova, 1988) which contains only negative events. The attributional style for positive events measured by the six positive scenarios of ASQ (Peterson et al., 1982) was not significantly correlated with attributional style for negative events measured by the EASQ (Peterson & Villanova, 1988). Moreover, Corr & Gray (1994), in an one-year predictive validity study investigating the role of attributional style, socialization and cognitive ability factors in insurance sales performance, reported a non-significant correlation between the combined positive events scale and the combined negative events one of the ASQ. These findings suggest that attributional styles for positive and negative events might well be independent from each other; specifically that the people who attribute negative events to internal, stable and global causes do not necessarily attribute positive events to external, unstable and specific causes - in contrast to the reformulated model (Abramson et al., 1978) and Seligman's (1990) definition of attributional style.
We suggest that certain people who attribute unfavourable outcomes to external, unstable and specific factors do not tend to attribute positive outcomes to internal, stable and global causes. This might be the case because a high value placed on realism or modesty has an effect on the process of making attributions. Taking the complete responsibility for negative outcomes and viewing them as stable and global can be perceived as a pathetic reaction to misfortunes which prevents attributing negative events to internal, stable and global factors, while attributing good events to one's self and considering them as stable and global might be moderated by modesty and a sense of realism. The literature on the partial independence of positive and negative affect suggests that optimism is not necessarily the opposite of pessimism. For example, Argyle & Henderson (1985), in the field of marital satisfaction, showed that positive and negative affect can be independent of each other. Partners can have strong positive feelings (i.e. related to frequency of intercourse), as well as negative ones (i.e. related to the frequency of rows). In the same way, the simultaneous manifestation of optimism and pessimism by an individual is possible within a specific area of his/her life, for example, at work. Hence, it is important to find a way around the problem of using events that are representative of a specific domain in order to be able to measure the tendency of a person to be generally optimistic or pessimistic, as far as this specific domain is concerned.
Another issue concerning attributional style involves internality as a component of attributional style. Even though the reformulation of the learned helplessness model (Abramson et al., 1978) distinguishes between the concepts of personal and universal helplessness and suggests that a person can be internally or externally helpless, rather confusingly internality is still incorporated as a causal dimension of a vulnerable attributional style. The reformulated model suggests that 'attributing the cause of helplessness internally often but not always implies a grimmer future than attributing the cause externally, since external circumstances are usually but not always in greater flux than internal factors' (p. 56). However, it has been suggested that one can feel helpless in the face of the future without feeling that the future will necessarily be bad (Brewin, 1988; Weiner, 1986). Weiner (1986) argues that, even though the notion of helplessness is associated with the concept of locus of control, helplessness is not the core theme of pessimism. More specifically, he suggests that a person can feel helpless when he or she perceives lack of personal control in the situation but it is not necessary to feel hopeless as well. The person might believe that an external force will change a negative situation and consequently might be in a state of guarded optimism or, at least, not pessimism. Weiner's attributional theory of motivation and emotion proposes that the dimension of causal stability (and perhaps globality) is the one that affects one's expectations of future success or failure, mediating the effect of good or bad outcomes on motivation and emotion, and not the causal dimensions of locus or controllability.
Abramson, Metalsky & Alloy (1989) have recently introduced the hopelessness theory of depression which is a revision of the reformulated theory of helplessness and depression. In their revision they suggest there are individual differences in attributional style, which is defined as a distal factor leading to hopelessness and depression. Some individuals may exhibit a general tendency to use stable and global causes in explaining important negative events, while other individuals may not. Those who tend to explain negative events in stable, global terms should have bleak expectations of the future which, in turn, proximally lead to bad feelings. This tendency is defined as a depressogenic attributional style which provides 'specific vulnerability' to depression; but it does not necessarily cause the occurrence of the symptoms. Thus, although many cases of hopelessness depression will occur among cognitively vulnerable people when they are confronted with negative events, people who do not have a depressogenic attributional style also may develop hopelessness depression when they are confronted with events sufficient to engender hopelessness. Hopelessness theory focuses on the role of stability and globality in the formulation of expectations, while de-emphasizing the causal dimension of internality.
It should be made clear that the hopelessness theory predicts not only depression, but also how people can maintain a positive emotional state. According to this theory, the occurrence of a negative event provides a challenge to a positive emotional state. The attribution of a negative event to stable and global causes, the perceived importance of the event's consequences, the inference of negative characteristics about the self, and finally, a depressogenic attributional style contribute to the occurrence of hopelessness and, as a result, to the disturbance of a positive emotional state. In contrast, refraining from making these inferences, as well as the absence of a depressogenic attributional style, would allow hope to endure and positive state to maintain. Consequently, attributional style is not a sufficient cause leading to depression, as the theory also specifies other factors which determine whether a negative event causes hopelessness and depression.
Although Cutrona, Russell & Jones (1985) mainly questioned whether most individuals are sufficiently consistent across situations in the causal attributions that they make to justify the concept of attributional 'style', they provided evidence of a stronger cross-situational consistency of the stability and the globality dimensions compared to the dimension of internality. Cutrona et al. (1985) used a confirmatory factor analysis to test the construct validity of the ASQ (Peterson et al., 1982). The best fitting model hypothesized both cross-situational consistency and situational specificity to underlie their dataset. The results suggested little evidence of cross-situational consistency for the internality dimension. This finding is consistent with the low levels of internal reliability reported by Cutrona et al. (1985) for this subscale (.33). The results for stability and globality provided stronger evidence of cross-situational consistency.
At this point one must consider the fact that the locus of control scales (Furnham & Steele, 1993) are more reliable compared to the internality dimension subscale of the attributional style questionnaires. It can be argued that the locus of control scales generally have more items which contribute to their greater reliability. However, even in the case of the EASQ, which is an attempt to increase the internal consistency of the attributional style questionnaires by adding more items, the alphas of negative stability and globality had quite higher coefficients (.85 and .88, respectively) compared to the negative internality dimension (.66). Thus, the paradox might not only be explained by the number of items, but also by the different nature of the items contained in these scales. More specifically, in the locus of control scales the respondents are asked to rate a number of locus of control beliefs, as for example to what extent they agree that 'Making money is a matter of good fortune' (item taken from Spector's, 1988, WLC scale), while the attributional style questionnaires ask respondents to imagine vividly an event that might or might not have happened to them. In the case of the attributional style questionnaires the event items cause the retrieval of a more complex causality representation where single factors cannot be perceived as sufficient for explaining an outcome, and because respondents are asked to rate only one cause they tend not to be consistent across situations. This is the issue primarily with the internality scale because attributions to internal or external factors are not necessarily perceived by the respondents as mutually exclusive. It has been shown in the relevant literature that participants are more likely to employ combinations of both internal and external attributions for complex interpersonal events (Bradbury & Fincham, 1988). On the other hand, locus of control beliefs probably do not lead to complex causal schemata as the person is not asked to find the causes of an event but rather to state the extent to which the statement is in accordance with his/her belief system.
Finally, the superiority of attributional style for negative events over attributional style for positive events as a more reliable and valid measure of performance can be argued on the basis of findings in the relevant literature.
Burns & Seligman (1989) analysed the attributional style across the life span. Participants whose average age was 72, completed the ASQ (Peterson et al., 1982) and provided diaries and letters written in their youth, an average of 52 years earlier. The results showed that the attributional style for negative events was stable throughout adult life, but their attributional style for positive events was not stable across this interval. Furthermore, Burns & Seligman (1989) reported that the stability and the globality dimensions showed intra-subject consistency, while the internality dimension was less consistent.
Seligman & Schulman (1986), looking at the influence of attributional styles on the performance of life insurance sales agents, found that an attributional style of external, unstable and specific attributions for negative outcomes was correlated with higher productivity, whereas an attributional style of internal, stable and global attributions for positive outcomes was not. Moreover, Peterson (1991) suggested that attributional style for positive events might not yield rich data because people are less 'mindful' in thinking about good events and, therefore, offer unrevealing, stereotyped responses to them. This speculation is suggestive of Weiner's attributional paradigm (1986) in which he proposes that causal research is more likely undertaken when an event is negative, unexpected or important.
The aim of this study is to propose a more reliable and valid measure of attributional style by focusing on the dimensions of stability and globality for negative events. In order to do that the following hypotheses were tested: (1) attributions for positive and attributions for negative outcomes are not negatively correlated and do not load on three correlated factors, namely internality, stability and globality, in an oblique factor analytic investigation of their nature; (2) the causal dimension of internality does not show acceptable levels of cross-situational consistency, while stability and globality show stronger cross-situational consistency; and (3) the causal dimensions of stability and globality for negative events are valid measures of the concept of pessimism and confidence.
A total of 189 insurance sales staff of a British financial organization took part in this study. There were 34 (18 per cent) women and 155 (82 per cent) men in the group. They ranged in age from 22 to 53 years (the mean being 31 years, SD = 5 years); 3.7 per cent had a basic salary of [pounds]14 000 to [pounds]15 999 per year, 70.4 per cent [pounds]16 000 to [pounds]17 999, 21.8 per cent [pounds]18 000 to [pounds]19 999 and 3.7 per cent had a salary of more than [pounds]20 000 per year. The respondents received a profile of their own results with an interpretive guide after the study. Selling insurance is a job particularly suitable for the investigation of attributional style because sales staff repeatedly encounter failure, rejection and indifference from prospective clients. Consequently, the turnover rate among insurance agents is very high (Seligman & Schulman, 1986).
Participants completed the Occupational Attributional Style Questionnaire (OASQ; Furnham, Sadka & Brewin, 1992) and the Social Problem Solving Inventory (SPSI; D'Zurilla & Nezu, 1990). For the measurement of attributional style the OASQ (Furnham et al., 1992) is preferred to the ASQ (Peterson et al., 1982) because it is specifically related to the work setting. The Negative Problem Orientation subscale of the SPSI, which is a measure of pessimistic orientation in problem solving, was used for the concurrent validation of OASQ mainly for three reasons: first, because the subscale measures a motivational component of problem solving and learned helplessness theory suggests that attributional style effects motivation; second, because motivation is assumed to be necessary for performance; and third, because problem solving is part of the job of selling insurance which requires the sales staff to find the best fit between client financial needs and the organization's financial services. Respondents' scores on the Rapid Personality Questionnaire (RPQ; Rust, 1991) were also available. This instrument is used by the organization in the selection of insurance agents.
Occupational Attributional Style Questionnaire (Furnham et al., 1992). The OASQ, which was closely modelled on the ASQ, presents respondents with eight different situations concerning work life, half of which are positive events (e.g. getting a promotion), while the other half are negative (e.g. your boss acts aggressively towards you). Respondents are required to imagine themselves vividly in the situation described, write down the most likely cause of the situation and finally, rate this cause on separate seven-point scales for nine causal dimensions, namely internality, stability, globality, externality, chance, personal control, colleague control, foreseeability and importance. One of the advantages of this questionnaire compared to other methods of attribution research concerns the fact that respondents are asked to find their own causes of events, not to simply choose one from a provided list of causes. According to Wimer & Kelley (1982), the process of thinking about and arriving at an attribution may provide a fuller or possibly clearer understanding of the meaning of a cause than in the more passive case when one simply reads an attribution.
The OASQ was found to have alphas ranging from .52 to .84 and test-retest reliability for a four-week period yielded r = .87. The coefficients of internal consistency obtained for the current study's data are presented in the results section. As far as construct validity is concerned, attributions for positive events particularly (and to a lesser extent with positive and negative combined) were significantly and consistently correlated with salary, job satisfaction, intrinsic motivation and social class variables that previous research has demonstrated to be correlated with optimistic attributional style. In this study the dimensions of internality, stability and globality were used.
Social Problem Solving Inventory (D'Zurilla & Nezu, 1990). The SPSI was developed by D'Zurilla & Nezu (1990) and is based on the theoretical framework of a problem-solving model. According to this model, social problem solving is a complex cognitive-affective-behavioural process that consists of two major components: (a) problem orientation and (b) problem-solving skills, which compose the two main scales of the questionnaire.
Problem orientation refers to the motivational component of the problem solving process, arising primarily from past problem-solving experiences. There are two types of problem orientation, positive problem orientation (PPO) and negative problem orientation (NPO). For the purposes of this study the subscale of NPO was used. The subscale of NPO (10 items) concerns the extent to which a person is vulnerable to anxiety, self-doubt, has low self-efficacy and a tendency to see things in a pessimistic way at early signs of poor outcomes. Test-retest reliability (average of three weeks apart) for the whole questionnaire and the Problem Orientation Scale were .87 and .83, respectively (D'Zurilla & Sheedy, 1991). Alpha coefficients for the same measures are .94 and .94, respectively (D'Zurilla & Sheedy, 1991). The coefficient of internal reliability for the NPO subscale obtained in the current study is .88.
Rapid Personality Questionnaire (Rust, 1991). The RPQ contains 80 items which are designed to elicit a spectrum of the individual's personality profile in occupational settings. The items are adjectives rated on separate five-point scales. The respondents are required to respond to the items in terms of how they think each item applies to them while at work. The questionnaire contains five scales, namely confidence, extraversion, tough-mindedness, conformity and being structural. Regarding the reliability of the instrument, there are reported coefficients of split-half reliability ranging from .76 to .84, coefficients of internal consistency ranging from .76 to .86 and test-retest reliability coefficients ranging from .85 to .93 (Rust, 1991). According to Rust (1991), the factorial validity of the RPQ is demonstrated by the fact that its five-factor solution matches the five domains of the test specification, as well as conforming to the 'big five' solution by McCrae & Costa (1987). In the current study the total score on the 16-item scale of confidence was used. The coefficient of internal consistency for the present data could not be computed because the researchers had only the total score of each respondent on the scale but not the scores on each of the 16 items of the scale.
The respondents completed the questionnaires at work. As a motivational incentive towards accurate responding the respondents were promised and received an individualized bar graph profile and a guide to enable them to interpret their own scores. A total of 189 questionnaires were returned which is a response rate of 54 per cent.
Descriptive statistics, coefficients of internal reliability and the intercorrelations between the OASQ's six scales for both positive and negative events are presented in Table 1. There are moderate positive correlations among the scales of positive internality, positive stability and positive globality at a .001 level of significance. Moreover, the scales measuring the causal dimensions for negative events show moderate positive intercorrelations at a .05 and .001 level of significance. These positive correlations are in accordance with Seligman's conceptualization of attributional style in which the dimensions of internality, stability and globality are assumed to be related to one another.
There are primarily positive intercorrelations among the scales for the positive events and the scales for the negative ones (apart from a low negative correlation -.07 between the positive globality and the negative internality, which is not statistically significant). Four of these correlations are very low, ranging from .02 to .10 and have failed to be significant. The significant positive correlations are quite low, ranging from .13 to .30. The partial correlations between the OASQ scales controlling for age, sex and salary are not much different than the zero-order correlations. Finally, the combined positive events scale, which contains the dimensions of positive internality, stability and globality, has a low positive correlation (.18, p [less than] .006) with the combined negative events scale (negative internality, stability and globality). The partial correlation between the combined positive events scale and the combined negative events scale with the variables of age, sex and salary controlled is positive and not much higher than the zero-order correlation (.20, p [less than] .003).
It is interesting that, first, the correlations between the scales for positive and negative events were not negative and, second, the correlation between the combined [TABULAR DATA FOR TABLE 1 OMITTED] positive events and the combined negative events scales was low and positive. This is contrary to the supposition that an optimistic explanatory style is the habitual way of attributing successes to internal, stable, global causes and failures to external, unstable and specific ones and that a pessimistic explanatory style is the habitual way of attributing successes to external, unstable, specific factors and failures to internal, stable and global ones.
A higher-order factor analysis of the OASQ's scales was carried out to examine whether the causal dimensions of internality, stability and globality for positive and negative events would load on three correlated factors. According to the reformulated model of learned helplessness, internality, stability and globality are three distinct but related dimensions of causality and should, therefore, form three separate factors. Moreover, attributions for good and bad events of each dimension should load on the same factors, with dimensions for bad events having a negative loading, since the reformulated model suggests that individuals who attribute positive events to internal, stable and global factors would attribute negative events to external, unstable and specific causes. Consequently, the factor analysis should generate three factors, namely internality, stability and globality, each of which will include the scales of positive and negative internality, stability and globality, respectively.
A principal component analysis with OBLIQUE rotation and predefined number of factors (three factors) was carried out to see whether attributions for positive and negative events for each dimension would load on the same factors. The three extracted factors accounted for 74.7 per cent of the common factor variance. The OBLIQUE rotation indicated that the factors were not highly correlated ranging from .08 to -.17. The results, as well as the correlations between the extracted factors, are presented in Table 2.
The results of the factor analysis showed that the causal dimensions for positive events loaded on one factor, while the same dimensions for negative events loaded on two other factors. Consequently, the causal dimensions for positive and negative events form distinct constructs, which are virtually independent from each other (the correlation between Factor 1 and Factor 2 is r = .08 and between Factor 1 and Factor 3 is r = -.16). These results support the hypothesis that a specific attributional style for positive events is not associated with a specific attributional style for negative events.
As far as the testing of hypothesis 2 is concerned, the alpha coefficients of internal consistency were calculated for each dimension to see whether the items that are contained in every dimension are highly intercorrelated. The Cronbach alpha coefficients for all the dimensions are presented in Table 1. The results show that the OASQ has moderately high coefficients of internal consistency for stability and globality, ranging from .65 for positive globality to .72 for positive stability. However, the scale of internality shows quite low coefficients of internal reliability (.28 for positive internality and .40 for negative internality), which means that the respondents' attribution of positive or negative events to internal factors depends on the specific event and is not cross-situational. Therefore, the causal dimension of internality does not appear to demonstrate acceptable levels of within-subjects consistency, while stability and globality do show stronger cross-situational consistency. After a literature review on the concept of attributional style, a table was produced to present a number of other studies' coefficients of internal consistency showing the same pattern of findings with the current study (Table 3).
Table 2. Factor analysis of the OASQ's scales of internality, stability and globality for positive and negative events (loadings above [absolute value of .40] are shown in boldface) Scales Factor 1 Factor 2 Factor 3 Positive internality .72 .20 .03 Positive stability .81 -.01 -.39 Positive globality .82 -.01 -.10 Eigenvalue = 2.1 Variance explained = 35.9 Negative internality -.00 .91 -.07 Negative globality .23 .73 -.52 Eigenvalue = 1.4 Variance explained = 24.6 Negative stability .13 .22 -.94 Eigenvalue = 0.85 Variance explained = 14.2 Factor correlation matrix Factor 1 Factor 2 Factor 1 1.00 Factor 2 .08 1.00 Factor 3 -.16 -.17 Note. N = 189.
Since the attributional style for negative events was shown to be a more reliable and valid measure compared to the attributional style for positive events (see the introductory section), a stepwise multiple regression was carried out in order to test the validity of negative stability and negative globality as measures of pessimism. The SPSI's scale Negative Problem Orientation (NPO) was used as the dependent variable, while the independent variables were the negative stability and the negative globality scales. A stepwise multiple regression coefficient of R = .19, F(1,187) = 7.50, p [less than] .006 was generated between the NPO subscale and the predictor, namely, negative globality (Table 4).
Finally, a stepwise multiple regression was conducted using the RPQ's scale of confidence as the criterion and the scales of negative stability and negative globality as the predictor variables. Confidence is interchangeably used in the literature with the concept of self-esteem, which according to Seligman (1990) determines attributional style. For Seligman, attributional style is a result of how valuable and deserving one perceives himself or herself to be. Table 5 presents the stepwise multiple regression results. A multiple regression coefficient of R = .21, F(1,187) = 9.23, p [less than] .002 was generated between the dependent variable, namely the RPQ's [TABULAR DATA FOR TABLE 3 OMITTED] [TABULAR DATA FOR TABLE 4 OMITTED] [TABULAR DATA FOR TABLE 5 OMITTED] scale of confidence and the predictor which succeeded in entering the regression equation, namely negative globality.
The data indicate that the causal dimension of negative globality is a predictor of negative problem orientation suggesting that individuals with high scores on negative globality tend to have a pessimistic attitude towards the confrontation of emerging problems. Moreover, negative globality was found to be a statistically significant predictor of confidence with a negative beta weight, which means that low levels of negative globality are related to high levels of confidence.
The intercorrelations between the scales of internality, stability and globality for positive events showed moderate positive correlations at a .001 level of significance ranging from .33 to .56. The same causal dimensions for negative events showed moderate positive correlations, ranging from .16 to .44. However, the inter-correlations between the causal dimensions for positive outcomes and the causal dimensions for negative outcomes were primarily positive and statistically non-significant. The higher-order factor analysis of the six scales generated three factors; the first was made up of the scales for positive events, the second of the scales of negative internality and negative globality, and the third of the scale of negative stability.
We suggest the lack of significant intercorrelations between the three scales for positive events and the three scales for negative events makes it reasonable to support hypothesis 1; that is, attributional styles for positive and negative events should be considered as separate variables. Therefore, high scores on the scales for positive events are not related to low scores on scales for negative events. This finding is in contrast with Seligman's (1990) definition of attributional style in which he suggests that a person who attributes successes to internal, stable and global factors will also attribute failures to external, unstable and specific causes.
In addition, to further test hypothesis 1, a factor analysis of the six scales was conducted. According to the reformulated model of learned helplessness, three factors are to be extracted, namely internality, stability and globality, which correlate with each other. Moreover, the theory implies that the scales for the negative events will load on their intended factor negatively, for example the scale of negative internality should load negatively on the general factor of internality, while the scale of positive internality will load on the same factor but positively. The results of our factor analysis do not support this theoretical model. The first factor can be called 'attributional style for positive events' as it contains the scales of positive internality, positive stability and positive globality. The second factor consists of negative internality and negative globality, while the third factor holds the scale of negative stability. Consequently, the results of the higher-order factor analysis, as well as the intercorrelation coefficients, indicate that attributional style for positive events and attributional style for negative events should not be considered as opposite poles of a general attributional style, but rather as separate variables. It is possible that subjective categorization of these events takes place; possibly the events used in OASQ are not a representative sample of aspects of the occupational setting and their importance varies among the respondents. Thus, the subjective importance of the events can be a factor that influences the grouping of the subscales in the factor analysis. Future research should try to handle this problem and one way around it can be to give a weight to each answer depending on how important the event was reported to be by the respondent.
The alpha coefficients for the dimensions of stability and globality for both positive and negative events were acceptable, ranging from .65 to .78. However, the alpha reliabilities of internality for positive as well as for negative events were quite low; that is, .28 for positive internality and .40 for negative internality. The coefficients of alpha reliability for the six scales provide support for hypothesis 2. The low alphas for the scales of positive and negative internality show low cross-situational consistency. This is important because high consistency is an essential attribute of any concept of style. On the other hand, the causal dimensions of stability and globality for both positive and negative events do show high coefficients of internal consistency, especially if we take into account the fact that their calculation was based on only three or four items. Therefore, since stability and globality seem to demonstrate quite a strong cross-situational consistency, we interpret them to be a more reliable measure of the attributional style concept.
The psychometric issue of the low reliabilities for the internality dimension reported by the current study, as well as by other studies (see Table 3), needs a theoretical explanation. The small number of items contained in the attributional style questionnaires can be one factor, but why does this primarily effect the alphas for the internality scale? Even in the case of the EASQ (Peterson & Villanova, 1988), which is an attempt to improve the reliability of the ASQ (Peterson et al., 1982) by lengthening it to include 24 negative outcomes, internality once again is the least reliable dimension.
A second explanation can be offered within the framework of attributional theory. The emphasis in the early models of attributional reasoning was on people's explanations for specific events, and usually for a single cause of the effect in question (Kelley, 1967). However, social psychologists have proposed that this approach is not appropriate when people are trying to explain complex social phenomena. For these ambiguous and complicated issues it is not surprising that people think that many interrelated causes have an effect on the emergence of social phenomena. Consequently, it has been proposed that ordinary people's knowledge of a domain can be represented as a network of causal intercorrelations or 'perceived causal structure' (Antaki, 1985; Kelley, 1983; Livingstone & Lunt, 1989). Researchers have discussed this 'perceived causal structure' by examining the causal explanations given by people for complex social phenomena, as for example, unemployment (Furnham, 1982a) and poverty (Furnham, 1982b).
However, it may be that complex social phenomena are not the only stimuli that make people produce complex causal explanations. Single events may also stimulate such perceived causal schemata when people are asked to imagine vividly such events happening to them. In the case of the attributional style questionnaires, as respondents are requested to retrieve an event in their memory, they likely tend also to retrieve many associated factors that led to or caused this event. An indication of the activation of complex causal schemata lies in the fact that respondents frequently complain about having to find the single most likely cause of an event, arguing that a number of causes seem to them to be equally important. As respondents are obliged to restrict themselves to one cause of the event and at the same time perceive a complex causal reality, they have the tendency to come up with causes that will in one event represent attributions about the self (internal) and in another event attributions about others or the situation (external), which are the two main aspects of causality. On the other hand, because choosing a stable or unstable cause and a global or specific one does not determine whether an attribution is biased or self-centred as in the case of internal or external causes, the activation of the complex causal schema does not decrease the internal reliability of stability and globality.
In addition, the suggestion of Miller, Smith & Uleman (1981) that the distinction of internal/external causes can be regarded as reflecting an underlying and more fundamental causality dimension, namely, the chosen/not chosen dimension, can be utilized for explaining the low cross-situational consistency of internality. The proposed dimension of chosen/not chosen refers to the distinction between acts deliberately chosen by the actor, or outcomes more deterministically caused. Therefore, respondents might not be consistent on attributing various events to internal or external factors because they rather focus on choosing causes which explain events either as the outcome of their own actions or as deterministically caused. However, such an explanation is a mere speculation that needs future research to address it.
Finally, the dimension of negative globality emerged as a predictor of negative problem orientation in the stepwise multiple regression where the positive beta weight of .19 was found. It was also a predictor of confidence with a negative beta weight of -.21. Negative stability failed to enter any of the multiple regression equations. Consequently, the third hypothesis was partly confirmed since the causal dimension of negative globality was shown to be a predictor of negative problem orientation and confidence, while stability did not predict either of the independent variables such that its validity as a measure of pessimism and confidence does not seem to be supported.
The results suggest that attributional styles for positive and negative events are distinct variables. Our study demonstrated that workers who gave high scores to one of the dimensions of attributional style for positive events did not simultaneously give low scores to the same dimension for negative outcomes. For example, high scores on positive internality were not related to low scores on negative internality. In addition, the results of our study advocate the omission of internality for both positive and negative outcomes, as we found low levels of consistency across situations. Our findings parallel the low Cronbach alphas for the internality dimension reported by several other studies of attributional style (Burns & Seligman, 1989; Cutrona et al., 1985; Furnham, Brewin & O'Kelly, 1994; Furnham et al., 1992; Heaven, 1994; Peterson & Villanova, 1988). Finally, the scale of negative globality appeared to be a valid measure of pessimism and confidence, while the dimension of negative stability did not appear to be related. Consequently, high scores on the causal dimension of negative globality might well be used as an indicator of a generalized tendency to hold low expectations for future success, while scores on negative stability should be interpreted with caution.
At this point one should take into account that the results of the current study are based on a specific population; that is, insurance sales staff. Therefore, there is a need for future research to replicate these findings in different populations in order to enable us to generalize them.
Clearly these findings have implications for both the conceptualization issues regarding attributional style and the actual measurement of the concept. The main purpose of this study was a proposition for a more reliable and valid measure of attributional style. This can be accomplished by the omission of the dimension of internality since it has constantly proven to be an unreliable measure, the usage of negative events instead of both negative and positive and, finally, the measurement of the dimension of negative globality which was shown to be a valid measure of pessimism and confidence. The findings of this study demonstrate the reliability and the validity of the causal dimension of negative globality for the measurement of pessimism and confidence. Consequently, it is suggested that more attention be given to develop a measure of attributional style which, we suggest, be based on the extent to which people hold generalized low expectations for future performance.
Abramson, L. Y., Metalsky, G. I. & Alloy, L. B. (1989). Hopelessness depression: A theory-based subtype of depression. Psychological Review, 96, 358-372.
Abramson, L.Y., Seligman, M. E. P. & Teasdale, J.D. (1978). Learned helplessness in humans: Critique and reformulation. Journal of Abnormal Psychology, 87, 49-74.
Ahrens, A. H. & Haaga, D. A. F. (1993). The specificity of attributional style and expectations to positive and negative affectivity, depression, and anxiety. Cognitive Therapy and Research, 17, 83-98.
Antaki, C. (1985). Ordinary explanation in conversation: Causal structures and their defence. European Journal of Social Psychology, 15, 213-230.
Argyle, M. & Henderson, M. (1985). The Anatomy of Relationships. London: Heinemann.
Bradbury, T. N. & Fincham, F. D. (1988). Assessing spontaneous attributions in marital interaction: Methodological and conceptual considerations. Journal of Social and Clinical Psychology, 7, 122-130.
Brewin, C. R. (1988). Cognitive Foundations of Clinical Psychology. London: Erlbaum.
Burns, M. O. & Seligman, M. E. P. (1989). Explanatory style across the life span: Evidence for stability over 52 years. Journal of Personality and Social Psychology, 56, 471-477.
Corr, P. J. & Gray, J. A. (1994). The role of attributional style, socialisation, and cognitive abilities in insurance selling: An argument for interaction regression models in validity estimation. Paper presented at the BPS Occupational Section Conference, Brighton.
Cutrona, C. E., Russell, D. & Jones, R. D. (1985). Cross-situational consistency in causal attributions: Does attributional style exist? Journal of Personality and Social Psychology, 47, 1043-1058.
D'Zurilla T. J. & Nezu, A.M. (1990). Development and preliminary evaluation of the Social Problem-Solving Inventory. Psychological Assessment: A Journal of Consulting and Clinical Psychology, 2, 156-163.
D'Zurilla, T. J. & Sheedy, C. F. (1991). Relation between social problem-solving ability and subsequent level of psychological stress in college students. Journal of Personality and Social Psychology, 61, 841-846.
Furnham, A. (1982a). Explanations for unemployment in Britain. European Journal of Social Psychology, 12, 335-352.
Furnham, A. (1982b). Why are the poor always with us? Explanations for poverty in Britain. British Journal of Social Psychology, 21, 311-322.
Furnham, A., Brewin, C. R. & O'Kelly, H. (1994). Cognitive style and attitudes to work. Human Relations, 47, 1509-1521.
Furnham A., Sadka, V. & Brewin, C. (1992). The development of an occupational attributional style questionnaire. Journal of Organizational Behaviour, 13, 27-39.
Furnham, A. & Steele, H. (1993). Measuring locus of control: A critique of general, children's, health- and work-related locus of control questionnaires. British Journal of Psychology, 84, 443-479.
Heaven, P. C. L. (1994). Occupational attributional style and attitudes to work: An Australian study. Australian Psychologist, 29, 57-61.
Hull, J. G. & Mendolia, M. (1991). Modelling the relations of attributional style, expectancies, and depression. Journal of Personality and Social Psychology, 61, 85-97.
Kelley, H. H. (1967). Attribution theory in social psychology. In D. Levin (Ed.), Nebraska Symposium of Motivation, vol. 15. Lincoln, NB: University of Nebraska Press.
Kelley, H. H. (1983). Perceived causal structures. In J. Jaspars, F. D. Fincham & M. Hewstone (Eds), Attribution Theory and Research: Conceptual, Developmental, and Social Dimensions. London: Academic Press.
Livingstone, S.M. & Lunt, P.K. (1989). Debating drunk driving: The construction of causal explanations in television discussion programs. Paper presented at BPS Social Section Conference, Bristol.
McCrae, R. R. & Costa, P. T. Jr (1987). Validation of the five-factor model of personality across instruments and observers. Journal of Personality and Social Psychology, 52, 81-90.
Miller, F. D., Smith, E. R. & Uleman, J. (1981). Measurement and interpretation of situational and dispositional attributions. Journal of Experimental Social Psychology, 17, 80-95.
Peterson, C. (1991). The meaning and measurement of explanatory style. Psychological Inquiry, 2, 1-10.
Peterson, C., Semmel, A., von Baeyer, C., Abramson, L. Y., Metalsky, G.I. & Seligman, M. E. P. (1982). The Attributional Style Questionnaire. Cognitive Therapy and Research, 6, 287-300.
Peterson, C. & Villanova, P. (1988). An Expanded Attributional Style Questionnaire. Journal of Abnormal Psychology, 97, 87-89.
Rust, J. (1991). Rapid Personality Questionnaire: Technical Manual. London: Personality Systems.
Seligman, M. E. P. (1990). Learned Optimism: How to Change your Mind and your Life. London: Pocket Books.
Seligman, M. E. P. & Schulman, P. (1986). Explanatory style as a predictor of productivity and quitting among life insurance sales agents. Journal of Personality and Social Psychology, 50, 832-838.
Spector, P. (1988). Development of the Work Locus of Control Scale. Journal of Occupational Psychology, 61, 335-340.
Weiner, B. (1986). An Attributional Theory of Motivation and Emotion. New York: Springer-Verlag.
Wimer, S. & Kelley, H. H. (1982). An investigation of the dimensions of causal attribution. Journal of Personality and Social Psychology, 43, 1142-1162.
|Printer friendly Cite/link Email Feedback|
|Author:||Xenikou, Athena; Furnham, Adrian; McCarrey, Michael|
|Publication:||British Journal of Psychology|
|Date:||Feb 1, 1997|
|Previous Article:||Illusory self-assessments - can they be reduced?|
|Next Article:||The account taken of statistical power in research published in the 'British Journal of Psychology.'|