Professional Documents
Culture Documents
Frustration Discomfort Scale
Frustration Discomfort Scale
Assessment
INTRODUCTION
Rationalemotive behaviour therapy (REBT) is unique in clearly distinguishing between two categories of dysfunctional belief (Ellis, 1979, 1980). The rst category involves the intolerance of frustration and discomfort, whilst the second involves the global rating of self-worth. Whilst these two categories of belief interact, they are also hypothesized to have an independent and distinct relationship with specic psychological problems.
However, although frustration intolerance1 is a central concept in REBT theory, and one of REBTs distinguishing features, investigation of these beliefs has been limited (Neenan & Dryden, 1999). Evidence from existing scales indicates that frustration intolerance and self-worth do form separate categories (DiGiuseppe, Leaf, Exner, & Robin, 1988). Yet beyond this, the content and factor structure of frustration intolerance beliefs, and their relationship with specic problems, has remained largely unexplored.
* Correspondence to: Dr. Neil Harrington, Stratheden Hospital, Cupar, Fife, KY15 5RR, UK E-mail: [email protected]
Traditionally the term low frustration tolerance has described this belief process. Since this has been criticized as being ambiguous (Neenan & Dryden, 1999) frustration intolerance is used in this paper.
Development of the Frustration Discomfort Scale One reason is that many earlier irrational belief measures, such as the Belief Scale (Malouff & Schutte, 1986), were based on inadequate and outdated denitions of irrational belief, and gave only an overall score. Contemporary REBT theory focuses on four belief processes, with demands (e.g. I must do well) as the central process, and self-worth, frustration intolerance, and awfulizing secondary to this belief. Some assessment scales, such as the Survey of Personal Beliefs (Demaria, Kassinove, & Dill, 1989), have employed this model. However, by focusing on belief processes the content of beliefs has been largely ignored, although recent evidence suggests that both belief processes and their content are important. In particular, the series of experimental studies by Bond and Dryden indicates that the content of self-worth and frustration intolerance beliefs, rather than demandingness, is central in determining psychological disturbance (Bond & Dryden, 2000). The focus on belief content, rather than irrational processes, brings REBT closer to other cognitive therapy approaches. However, whilst assessment scales in the general literature have focused on belief content, they do not clearly distinguish between self-worth and frustration intolerance. For instance, achievement beliefs on the Dysfunctional Attitude Scale (DAS) (Weissman & Beck, 1978) almost entirely refer to issues of selfworth. Likewise, the Multidimensional Perfectionism Scale (Frost, Marten, Lahart, & Rosenblate, 1990) fails to separate self-worth and frustration intolerance beliefs. In regard to perfectionism, REBT would argue that high standards may be independently associated with self-worth (If I fail it means Im a failure as a person) and frustration intolerance beliefs (I cant stand not achieving my absolute best). Even if irrational belief processes are fundamental to emotional problems, it may be argued that these are too generalized and all encompassing to enable specic predictions. Furthermore, although frustration intolerance has been treated as unidimensional, the range of frustration intolerance beliefs described in the REBT literature suggests a multidimensional construct. These beliefs include the intolerance of hassles, effort, unfairness, and uncomfortable emotions (Dryden & Gordon, 1993). However, there is no existing belief scale specically aimed at assessing the content of frustration intolerance beliefs. The General Attitude and Belief Scale (GABS) (DiGiuseppe et al., 1988) was designed to assess both belief content and process, and factor analysis of the GABS indicated that
Copyright 2005 John Wiley & Sons, Ltd.
375 items were best grouped in terms of content (Shaw, unpublished masters thesis). In line with theoretical models (Walen, DiGiuseppe, & Dryden, 1992), these content factors reected themes regarding the need for achievement, approval, and comfort, with further items added later to form a fairness sub-scale (Bernard, 1998). However, the General Attitude and Belief Scale has signicant limitations. First, the sub-scales do not clearly differentiate between frustration intolerance and self-worth beliefs, with some sub-scales including items from both categories. Second, the frustration intolerance content range is relatively limited with many similar worded items, features that are likely to distort the factor structure. Therefore, Neenan and Dryden (1999) have highlighted the need to develop a multidimensional measure of frustration intolerance. Likewise, from a cognitive therapy perspective, Power et al. (1994) have recommended the development of multidimensional belief scales, with content areas based on theoretical as well as statistically grounds. The authors argue that the failure to adequately assess specic belief domains is one reason for inconsistent evidence linking core beliefs with enduring vulnerability to depression. This is because only a small sub-set of beliefs will be relevant to specic problems, and therefore global measures of beliefs will merely act as general indicators of distress. Similarly, a fundamental weakness of REBT theory has been the difculty in distinguishing between distinct psychological problems in terms of specic irrational beliefs (Kendall et al., 1995). The purpose of the present study was to develop a multidimensional scale regarding the content of frustration intolerance beliefs, and to clarify its factor structure. Two separate studies were involved in the development of the scale, both employing clinical and student samples. The rst study reports the construction and exploratory factor analysis of a preliminary Frustration Discomfort Scale (FDS). Based on these results, a revised scale was developed in the second study, and the factor structure tested by conrmatory factor analysis.
376 adult clinical psychology department (39% men and 61% women). The mean age of the clinical group was 39 years (range 1874). The sample was recruited from a diverse geographical and social area, and represented a broad range of nonpsychotic problems, with anxiety disorders (32%), depression (23%), and anger (14%) the most common classications. Problems were largely moderate to severe, with 50% of clients experiencing these difculties for three years or more and 44% rating the effect on their everyday life as severe. There was a response rate of 44%. A previous pilot study had indicated that nonresponders did not differ from responders on diagnostic or demographic characteristics. The non-clinical group consisted of undergraduate psychology students from three separate courses in abnormal psychology (30% male and 70% female). Missing data was randomly spread across the items, with a mean percentage missing values for each item of 0.32%. Nine clinical cases and one student case with more than 10% of missing values were removed. With remaining cases containing only a small percentage of missing data, a group mean was employed. Six extreme outliers were also removed due to problematic patterns of responses. Thus, following preliminary data analysis, 329 cases were included in the factor analysis: 242 in the clinical group and 87 in the student group.
N. Harrington tionally described by the phrase I cant stand it) and self-worth beliefs were separately rated for each demand statement: I absolutely must not be kept waiting Because I cant tolerate waiting 0 1 2 3 4 It would totally lower my self-esteem 0 1 2 3 4
Scale Development
It was important that the structure of items conformed to current REBT theory. Irrational beliefs have been described as being best represented as a compound sentence involving both a demand and a secondary belief (DiGiuseppe, 1996). It has also been argued that the absolute nature of the demand statement needs to be emphasized, and that the word should is insufcient in this regard (Robb & Warren, 1990). Following these recommendations, the items were designed as compound sentences with demand statements prefaced by the phrase absolutely must. It was also necessary to differentiate frustration intolerance from self-worth beliefs. This presented some difculty, since these two categories of belief are closely interrelated. Therefore, Dryden (1996) has argued that the strength of self-worth relative to frustration intolerance needs to be assessed to determine which type of belief is most salient. To address this, both frustration intolerance (tradiCopyright 2005 John Wiley & Sons, Ltd.
Reverse-scored items were not used, since the usefulness of this method of reducing response bias has been questioned (Schriesheim & Hill, 1981). In particular, it has been noted that only ten percent of respondents are required to mistakenly answer the reversed questions the wrong way to produce a spurious reverse-scored factor (Schmitt & Stults, 1985). Items referring to affect were avoided to reduce cross contamination and to insure the scale primarily assessed cognition rather than emotion (Smith, 1989). To enable adequate investigation of the underlying factor structure, and satisfy content validity, care was taken to include all relevant aspects of frustration intolerance beliefs. Possible content domains were suggested by REBT theory, and included demands for achievement, approval, comfort, fairness, immediate gratication, and control/certainty (Neenan & Dryden, 1999; Walen et al., 1992). An item pool was created using examples of frustration intolerance in the REBT literature (e.g. Dryden & Gordon, 1993; Neenan & Dryden, 1999). Seventy-four items were selected from this larger group to form the preliminary Frustration Discomfort Scale. A small pilot study (n = 30) was conducted to allow revision of items that had poor comprehension or extreme frequency distributions. Content validity of the scale was reviewed by Albert Ellis (personal communication, 1999), who commented that the important aspects of the concept had been thoroughly described.
Measures
Frustration Discomfort Scale (preliminary version). This contained 74 items, rated on a ve-point Likert scale. Individuals were asked to rate the strength of belief, with the following anchors: (0) absent, (1) mild, (2) moderate, (3) strong, (4) very strong. Using examples, the instructions highlighted the difference between frustration and selfesteem beliefs.
Clin. Psychol. Psychother. 12, 374387 (2005)
Development of the Frustration Discomfort Scale Rosenberg Self-Esteem Scale (Rosenberg, 1965). This scale was only completed in study I by the clinical group. It is a 10-item Likert-type scale with satisfactory internal reliability (0.87) and temporal stability (0.86). Schmitt and Stults (1985) have highlighted the importance of careful preliminary screening and the deletion of careless responses due to reversed item response error. Following this recommendation, the present study identied outliers in terms of the difference between the negative and positive self-esteem sub-scale scores, eliminating nine cases and leaving 232 for analysis. The mean and standard deviation of the clinical group was 23.39 (5.79), with higher scores representing high self-esteem.
377 discriminative validity were used. First, substatement scores on the FDS were examined to screen for those items with higher self-worth substatement scores relative to frustration intolerance scores. Paired t-tests indicated that only four items had signicantly higher self-worth sub-statement scores. Second, correlations between FDS items and the Rosenberg Self-Esteem Scale were examined. Comparison of corrected itemtotal correlations showed that three items had signicantly lower correlations with remaining items on the FDS than with the Rosenberg Self-Esteem Scale. Two of these items (intolerance of personal aws and of upsetting others) were clearly associated with the concept of self-worth. These two items and six further items displaying multiple problems were removed, leaving 66 items to be included in the factor analysis.
378 and factor IV for 7.02%. All items had salient loadings on at least one factor. Reassuringly, the results of the four-factor oblique rotation closely matched the orthogonal rotation. The pattern of factors and item order was essentially the same in both rotations. Attempts at oblique rotations with ve and six factors failed to converge after 50 iterations, further supporting the reliability and robustness of the four-factor solution.
N. Harrington approval and more with the control and subservience of others. Indeed, compared with the other FDS sub-scales, entitlement had the lowest correlation with self-esteem (Table 1). The central theme of this factor might be summarized as the belief that ones desires must be met and that other people should indulge and not frustrate these desires. The third factor, labelled discomfort intolerance, referred to demands that life should be easy, comfortable, and free of hassle. The highest loading item was Tasks that I attempt absolutely must not be too difcult. Otherwise, I cant stand doing them (0.680). Such beliefs have traditionally been central to the REBT concept of frustration intolerance. Whilst in the GABS these beliefs were combined with emotional intolerance items to form a need for comfort sub-scale, the present analysis indicated they were best represented by a separate factor. Factor four was labelled achievement, with items referring to task related frustration. The highest loading item was I absolutely must not leave work unnished. I cant bear to leave work unnished (0.745). These types of belief have been included in several belief scales, including the DAS achievement sub-scale and the multidimensional perfectionism scales. However, the present achievement items aimed to assess the intolerance of frustration, as opposed to loss of self-worth, following achievement goal failure. The relatively weak correlation of the achievement sub-scale with self-esteem, compared with its correlation with other FDS subscales, supports the argument that this aim was successful (Table 1).
Factor Labelling
The rst factor was labelled emotional intolerance, with items reecting intolerance of emotional distress. The item with the highest loading on this factor was I absolutely must be free of distressing feelings as quickly as I can. I cant stand for them to continue (0.716). Items included beliefs regarding the uncertainty, controllability, and aversiveness of emotion. REBT has long emphasized the importance of secondary disturbance, such as anxiety about anxiety, and its relationship to frustration intolerance (Ellis, 1979). The second factor was labelled entitlement. The highest loading item was I absolutely must not be taken for granted. I cant stand being unappreciated (0.661). Included within this factor were demands for immediate gratication and for fairness. Whilst many of the items referred to personal relationships, analysis indicated that these beliefs were not associated with self-worth or gaining approval. Thus, the item regarding upsetting others had a negligible correlation (r = 0.09) with entitlement, suggesting that individuals scoring high on this factor were less concerned with
Table 1. Correlations between the revised FDS and Rosenberg Self-Esteem Scale (preliminary FDS scores in brackets) Rosenberg Self-Esteem Scale Discomfort intolerance Entitlement Emotional intolerance Achievement Full four-factor scale Gratication Fairness Full ve-factor scale -0.43*** (-0.36***) -0.20*** (-0.17*) -0.49*** (-0.44***) -0.29*** (-0.18**) -0.43*** (-0.36***) -0.28*** -0.16* -0.40*** FDS 0.65*** (0.72***) 0.67*** (0.73***) 0.70*** (0.69***) 0.64*** (0.57***) 0.79*** 0.68***
N = 232 (self-esteem), N = 242 (preliminary FDS). N = 248 (self-esteem), N = 254 (revised FDS). *** p < 0.001. ** p < 0.01. * p < 0.05.
379 the remaining FDS items (i.e. minus the sub-scale in question). The discomfort and emotional intolerance sub-scales had the strongest correlations with self-esteem and entitlement and achievement the weakest. Signicant differences in FDS mean scores were found between the clinical and student groups (Table 3). All sub-scales, except entitlement, were higher in the clinical population. There was no signicant correlation between the full scale score and age (r(240) = -0.02, ns) or gender (t(240) = 0.62, ns), or between these variables and individual subscales. Although 15 nal scale items did not signicantly differentiate between the two groups, it was expected that some beliefs would have elevated scores in the student group. Thus, items referring to task obstruction, task hassle, task interest, and task persistence all showed elevated scores, possibly reecting increased concerns regarding academic work.
Sub-Scale Formation
The nal selection of items to form sub-scales was primarily based on relative loadings, using a cut-off point of >0.45. The need to take into account communalities and to include a representative content range led to three items below this loading criterion being retained. Only three of the nal subscale items were classed as complex variables, with loadings greater than 0.45 on other factors. Corrected itemtotal correlations were computed to highlight those items with poor reliability regarding designated sub-scales. All items showed correlations above 0.45 on their own scale, and above 0.30 with the full scale. Alpha values were (with the mean inter-item correlation in brackets) emotional intolerance 0.91 (0.43), entitlement 0.88 (0.37), discomfort 0.90 (0.40), achievement 0.82 (0.36), and full-scale score 0.95 (0.29). The full scale comprised 47 items, with each of the rst three sub-scales containing 13 items, and the achievement sub-scale eight items. As expected, the FDS dimensions were substantially intercorrelated (Table 2).
Table 2. Revised FDS: sub-scale intercorrelation matrix (preliminary scale in brackets) Sub-scales 1 2 3 4 Discomfort intolerance Entitlement Emotional intolerance Achievement 1 0.56 (0.65) 0.62 (0.68) 0.48 (0.47) 2 3 4
0.60 (0.51)
Table 3.
Preliminary FDS: mean scores and standard deviations Clinical Student 72.6 (27.1) 16.6 (8.4) 24.1 (9.3) 18.5 (9.4) 13.5 (5.9) t 5.50*** 4.31*** 0.69 8.64*** 4.04***
95.1 (34.5) 22.4 (11.5) 25.0 (10.9) 30.8 (12.2) 16.9 (7.0)
380 best described the data, and yielded meaningful sub-scales showing high internal consistency. Additional factors appeared residual and poorly dened. However, the original content categories suggested by REBT theory were not entirely replicated. Control beliefs were associated with a number of content areas rather than forming a separate certainty/control factor. Discomfort intolerance and emotional intolerance were also found to form separate factors. It could also be argued that the equivalent dimension to the need for approval within the domain of frustration intolerance was that of entitlement. A further nding was that, rather than forming separate factors, items referring to immediate gratication and fairness mainly loaded on entitlement. Interestingly, whilst loading on entitlement, the two positive gratication items (the need for a buzz or excitement) had weak relationships with remaining items on the FDS (0.34 and 0.24 respectively). This was unexpected, since short-term hedonism is described as a core aspect of frustration intolerance (Dryden & Gordon, 1993). On the other hand, the item referring to intolerance of gratication delay had a relatively strong correlation (0.58) with the remaining scale. One possible explanation is that the two excitement items reected functional extraversion rather than dysfunctional impulsiveness. It might further suggest that frustration intolerance items were better worded in terms of negative states, such as deprivation or aversiveness, rather than as a need for positive gratication.
N. Harrington Scale. This suggested that the overlap with selfesteem was relatively small. Therefore, it was decided to use a single frustration intolerance statement for each item and dispense with the additional demand and self-worth statements: I cant stand doing things that involve a lot of hassle 1 2 3 4 5 These changes substantially simplied the question structure and improved readability (Flesch reading score of 84% compared to 64%). There was also a reduction in missing data and spoiled responses (2% compared to 5%). Furthermore, considerably fewer responses in the absent category (12% compared to 21%) suggested greater careless responding on the preliminary scale, possible due to item complexity. All but two preliminary scale items were transferred to the new scale. Of 47 items included in the revised scale, 37 were used with relatively minor changes in wording. Overlapping and redundant item content was reduced and the achievement scale was lengthened to make it equal with other sub-scales. Since it was unclear whether fairness and gratication beliefs were separate dimensions or facets of entitlement, further items were generated to enable entitlement to be split into two smaller sub-scales. Study II aimed to examine how well the hypothesized factor structure tted the data using a conrmatory factor analysis.
The participants and procedure followed that of the preliminary study. The clinical sample consisted of 254 patients, with the response rate (45%) comparable to the previous study. Mean age of the clinical group was 38.6 (41% male and 59% female). Six cases were removed due to spoilt or missing data. The non-clinical population (n = 79) came from two classes of undergraduate psychology students (17% male and 83% female). This gave an overall sample of 333 cases for inclusion in the conrmatory factor analysis. The mean percentage of missing values for each item was 0.48%, and these were replaced by the group mean.
Measures
Frustration Discomfort Scale (revised version). The initial scale consisted of 47 items rated on a veClin. Psychol. Psychother. 12, 374387 (2005)
Development of the Frustration Discomfort Scale point Likert scale with the following anchors: (1) absent, (2) mild, (3) moderate, (4) strong, (5) very strong. The instructions were as follows: Listed below are a number of common thoughts and beliefs that people may have when they are distressed or frustrated. Please read each statement and decide how well this usually describes your own beliefs. Circle the number that best indicates the strength of this belief. Following analysis, the nal number of items was reduced to 28 items on the four-factor scale, with each sub-scale containing seven items. The ve-factor scale contained seven additional items to form separate gratication and fairness sub-scales. Rosenberg Self-Esteem Scale. The Rosenberg scale was completed by clinical and student groups. Preliminary screening resulting in the removal of three clinical cases showing careless responding, leaving 248 clinical, and 77 student replies, for further analysis.
381 had substantially higher correlations with the overall FDS than with self-esteem. Final selection for sub-scales took into account the strength of corrected item-total correlations and the need to have an acceptable range of item content. Sub-scale alpha coefcients were (mean interitem correlation in brackets) discomfort intolerance 0.88 (0.50), entitlement 0.85 (0.42), emotional intolerance 0.87 (0.49), and achievement 0.84 (0.43), with gratication 0.84 (0.43), and fairness 0.83 (0.42). The full-scale mean inter-item correlation for the four-factor scale was 0.94 (0.34) and the vefactor scale 0.95 (0.33). These values indicate good internal consistency, with the mean inter-item correlation giving a clearer measure of item homogeneity since this is not inuenced by scale length. Although the sub-scales were designed to measure relatively narrow factors, no mean inter-item correlation exceeded 0.50, which represents the item redundancy cut-off point suggested by Briggs and Cheek (1986). Furthermore, the full-scale scores are within the optimal range of item homogeneity (0.200.40) that Briggs and Cheek suggest offers the best compromise between bandwidth and delity. Finally, all items had corrected itemtotal correlations above 0.48 with the full scale, and above 0.50 with sub-scales (Table 4). Overall, these results suggest the revised scale has captured the range and complexity of frustration intolerance beliefs within an acceptable level of item redundancy.
382 non-signicant c2 indicating good t; (2) the c2 divided by the degrees of freedom, with a ratio of between two and three suggesting a good t (Carmines & McIver, 1981); (3) the comparative t index (CFI; Bentler, 1990), with values above 0.95 indicating a good t; (4) the root mean square of approximation (RMSEA; Browne & Cudeck, 1993), with values over 0.10 leading to rejection of the model, those from 0.05 to 0.08 acceptable, and values below 0.05 indicating a good t; (5) the TuckerLewis Index (TLI; Tucker & Lewis, 1973), which takes into account the degree of parsimony, with scores of above 0.90 regarded as a reasonable t. Fit indices are presented in Table 5. Although all models had signicant c2 values, this frequently overstates the lack of t in larger samples. However, the one- and two-factor models also displayed a poor t on the other t indices, with both the c2 ratio and the RMSEA values failing to reach acceptable levels. On the other hand, the four- and ve-factor models were both good ts to the data. The CFI and TLI values indicate a better t for the four-factor model, although the difference was relatively small. All items loaded on their designated factors. The standardized factor loadings for both the ve- and four-factor models were all substantial and signicant, ranging from 0.554 to 0.826 (Table 4).
N. Harrington correlations with remaining FDS items were substantially higher than with self-esteem, supporting the divergent validity of the scale. Whilst the comparisons in Table 1 refer to clinical groups, the relationship between the FDS sub-scales and student self-esteem scores showed evidence of even greater separation (correlations with remaining FDS items in brackets): discomfort intolerance -0.38 (0.57), entitlement -0.11 (0.67), emotional intolerance -0.38 (0.70), achievement -0.24 (0.68). Means and standard deviations for both the four and ve-factor models are presented in Table 6. Independent t-tests indicated signicant differences in FDS scores between the clinical and student groups. As might be expected, emotional intolerance showed the greatest separation and achievement frustration the least. Analysis of individual items indicated that nine items did not distinguish between the two groups. As with the preliminary scale, these items largely reected the increased importance of task orientation in a student population (Q2, 4, 5, 8, 9, 10, 16, 24, and 25). Four of the seven additional items included in the fairness and gratication sub-scales also failed to discriminate between clinical and student groups (Q30, 31, 32, and 33). Interestingly, the gratication sub-scale had the strongest correlation with remaining FDS items, underlining the importance of these beliefs. There were no gender differences for the FDS total score in the clinical or student groups. However, the emotional intolerance sub-scale was signicantly higher for females in the clinical group (t(252) = 2.53, p < 0.05). Unlike the preliminary scale, there was a weak but signicant negative relationship between the FDS total score and age (r(248) = -0.15, p < 0.05), with age signicantly correlating with entitlement (r(248) = -0.24, p < 0.001) and emotional intolerance (r(248) = -0.15, p < 0.05). This indicates that entitlement and emotional intolerance decrease with greater maturity.
Sub-Scale Intercorrelations
The examination of the intercorrelations between sub-scales was helpful in comparing the four- and ve-factor models (Table 2). The correlation corrected for attenuation due to unreliability should be larger than the sub-scale reliabilities, and if this is not the case then the two scales lack discrimination. All four sub-scales, although moderately intercorrelated, met this criterion. However, the fairness and gratication scales had corrected correlation (0.87) higher than their respective reliabilities (0.83 and 0.84), suggesting that these two scales are practically indistinguishable.
Discussion
The conrmatory factor analysis supported the multidimensional model of frustration intolerance described in study I. Although the four- and vefactor models both tted the data well, the magnitude of the difference in t between these two models was small and the degree of interrelationship large. Furthermore, the fairness and gratication sub-scales had corrected correlations higher than their reliability coefcients. This suggests
Clin. Psychol. Psychother. 12, 374387 (2005)
Table 4. Factor I Factor II Factor III Factor IV Factor II (a) Factor II (b)
Revised FDS: Standardized factor loadings and corrected item-total correlations Corrected itemtotal correlation 0.724 0.701 0.637 0.666 0.633 0.602 0.623 0.756 0.716 0.707 0.697 0.627 0.600 0.559 0.826 0.799 0.702 0.685 0.636 0.632 0.619 0.683 0.632 0.625 0.665 0.581 0.547 0.499 0.756 0.728 0.646 0.613 0.585 0.600 0.570
Factor I Discomfort intolerance 21 I cant stand doing things that involve a lot of hassle 13 I cant stand having to push myself at tasks 25 I cant stand having to persist at unpleasant tasks 5 I cant stand doing tasks that seem too difcult 17 I cant stand the hassle of having to do things right now 1 I need the easiest way around problems; I cant stand making a hard time of it 9 I cant stand doing tasks when Im not in the mood
Factor II Entitlement 10 I cant bear it if other people stand in the way of what I want 6 I cant stand it if other people act against my wishes 18 I cant stand having to give in to other peoples demands 22 I cant stand having to change when others are at fault 26 I cant tolerate criticism especially when I know Im right 14 I cant tolerate being taken for granted 2 I cant stand having to wait for things I would like now
Factor III Emotional intolerance 19 I cant bear disturbing feelings 11 I cant bear to have certain thoughts 3 I must be free of disturbing feelings as quickly as possible; I cant bear if they continue 15 I cant stand situations where I might feel upset 27 I cant stand to lose control of my feelings 7 I cant bear to feel that I am losing my mind 23 I cant get on with my life, or be happy, if things dont change
383
384
Factor IV Achievement 28 I cant tolerate any lapse in my self-discipline 24 I cant stand feeling that Im not on top of my work 20 I cant stand doing a job if Im unable to do it well 4 I cant stand being prevented from achieving my full potential 8 I cant bear the frustration of not achieving my goals 16 I cant bear to move on from work Im not fully satised with 12 I cant tolerate lowering my standards even when it would be useful to do so
Factor II (a) Gratication 10 I cant bear it if other people stand in the way of what I want 6 I cant stand it if other people act against my wishes 18 I cant stand having to give into other peoples demands 29 I cant tolerate being overlooked 34 I cant bear being deprived now of things I lacked in the past 32 I cant stand giving up immediate pleasures for the sake of a distant goal 2 I cant stand having to wait for things I would like now
Factor II (b) Fairness 22 I cant stand having to change when others are at fault 14 I cant tolerate being taken for granted 33 I cant tolerate being treated with disrespect 26 I cant tolerate criticism especially when I know Im right 35 I cant tolerate other peoples bad or stupid behaviour 31 I cant stand being left in the dark with no explanations 30 I cant bear to have been treated unjustly
N. Harrington
385
Table 6.
Revised FDS: means and standard deviations Clinical Student 75.7 (15.9) 17.6 (4.7) 19.6 (4.4) 18.3 (5.1) 20.2 (5.2) 95.4 (19.4) 17.2 (4.1) 22.2 (5.1) t 6.82*** 4.22*** 3.87*** 10.39*** 3.28*** 6.57*** 4.43*** 3.27***
Four-factor score Discomfort intolerance Entitlement Emotional intolerance Achievement Five-factor score Gratication Fairness
90.8 (20.7) 20.4 (6.5) 22.1 (6.2) 25.6 (6.6) 22.7 (6.1) 113.2 (25.3) 19.9 (6.3) 24.6 (5.9)
that they are correlated to such an extent that little would be gained by their separation, and that they are best interpreted as highly related facets of entitlement (John & Benet-Martinez, 2000). However, whilst the four-factor model is preferred on the grounds of parsimony it may be premature to conclude that the distinction between gratication and fairness is unimportant. One limitation of the present study is that, whilst representing a wide range of disorders, the numbers of individuals in some specic diagnostic categories was relatively small. Problems expected to have particular associations with frustration intolerance, such as alcohol and substance abuse, were underrepresented, and it would be useful for further research to determine the generalizability of these factors to different populations. The revised scale did show an increased correlation with self-esteem, suggesting that the more complex preliminary scale structure helped to separate these belief categories. However, the increase in correlation is relatively small, and balanced against this the revised scale shows improvement in readability, acceptability, and psychometric properties. Since individuals with low frustration tolerance would be less likely to complete difcult tasks, these are not minor considerations. DiscrimCopyright 2005 John Wiley & Sons, Ltd.
ination between the clinical and student groups was also improved, with all sub-scales signicantly higher in the clinical group. These results are of clinical and theoretical importance. Regarding the REBT theory of psychological disturbance, they support a model of irrational belief in which frustration intolerance and self-worth beliefs form separate categories of disturbance (DiGiuseppe, 1996). However, comparison of the preliminary and revised scales indicates that restricting the wording of irrational beliefs to frustration intolerance statements, and omitting absolute demandingness, did not change the factor structure or reduce the relationship with psychological disturbance. Whilst this does not imply that imperative demands are irrelevant to disturbance, it does suggest that other belief processes may be equally important. This is also consistent with recent evidence that irrational beliefs may be best phrased in terms of secondary belief content (Bond & Dryden, 2000). From a clinical perspective, effective therapy depends on being able to identify which cognitions are related to what problems. However, although frustration intolerance beliefs are hypothesized as being central to many difculties, such as therapy resistance (Ellis, 1985), the empirical evidence is
Clin. Psychol. Psychother. 12, 374387 (2005)
386 sparse. In particular, it is unclear whether the dimensions of frustration intolerance are uniquely related to particular problems or if these beliefs have a more general inuence. There is initial evidence from further validation data that specic dimensions of frustration intolerance are associated with distinct emotional problems, independent of any shared relationship with self-esteem (Harrington, unpublished doctoral thesis). It is hoped that the development of a multidimensional measure will encourage the further investigation of these beliefs.
N. Harrington
ing. Poster session presented at the World Congress of Behaviour Therapy, Edinburgh. Dryden, W. (1996). When musts are not enough. In W. Dryden (Ed.), Inquiries in rational emotive therapy (pp. 4750). London: Sage. Dryden, W., & Gordon, W. (1993). Beating the comfort trap. London: Sheldon. Ellis, A. (1979). Discomfort anxiety: A new cognitive behavioral construct. Part I. Rational Living, 14, 38. Ellis, A. (1980). Discomfort anxiety: A new cognitive behavioral construct. Part II. Rational Living, 15, 2530. Ellis, A. (1985). Overcoming resistance: Rational emotive therapy with difcult patients. New York: Springer. Frost, R.O., Marten, P., Lahart, C., & Rosenblate, R. (1990). The dimensions of perfectionism. Cognitive Therapy and Research, 14, 449468. John, O.P., & Benet-Martinez, V. (2000). Measurement: Reliability, construct validation, and scale construction. In H.T. Reis, & C.M. Judd (Eds), Handbook of research methods in social and personality psychology (pp. 339369). Cambridge: Cambridge University Press. Kendall, P.C., Haaga, D.A.F., Ellis, A., Bernard, M., DiGiuseppe, R., & Kassinove, H. (1995). Rational emotive therapy in the 1990 and beyond: Current status, recent revisions, and research questions. Clinical Psychology Review, 15, 169185. Malouff, J.M., & Schutte, N.S. (1986). Development and validation of a measure of irrational belief. Journal of Consulting and Clinical Psychology, 54, 860862. Neenan, M., & Dryden, W. (1999). Rational emotive behaviour therapy: Advances in theory and practice. London: Whurr. Power, M.J., Katz, R., McGufn, P., Duggan, C.F., Lam, D., & Beck, A.T. (1994). The Dysfunctional Attitude Scale (DAS): A comparison of forms A and B and proposals for a new sub-scaled version. Journal of Research in Personality, 28, 263276. Robb, H.B., & Warren, R. (1990). Irrational belief tests: New insights, new directions. Journal of Cognitive Psychotherapy, 4, 303311. Rosenberg, M. (1965). Society and the adolescent self-image. Princeton, NJ: Princeton University Press. Schmitt, N., & Stults, D.M. (1985). Factors dened by negatively keyed item: The result of careless respondents? Applied Psychological Measurement, 9, 367373. Schriesheim, C.A., & Hill, K.D. (1981). Controlling acquiescence response bias by item reversals: The effect on questionnaire validity. Educational and Psychological Measurement, 41, 11011114. Smith, T.W. (1989). Assessment in rational emotive therapy: Empirical access to the ABCD model. In M. Bernard, & R. DiGiuseppe (Eds), Inside rational emotive-therapy: A critical appraisal of the theory and therapy of Albert Ellis (pp. 135153). San Diego: Academic. Tucker, L.R., & Lewis, C. (1973). A reliability coefcient for maximum likelihood factor analysis. Psychometrika, 38, 110. Walen, S.R., DiGiuseppe, R., & Dryden, W. (1992). Practitioners guide to rationalemotive therapy (2nd ed.). Oxford: Oxford University Press.
ACKNOWLEDGEMENTS
This paper is based on research submitted to the University of Edinburgh in part fullment of a Doctorate of Philosophy degree. I would like to thank Professor Mick Power for his help and encouragement throughout the research.
REFERENCES
Arbuckle, J.L., & Wothke, W. (1999). Amos 4.0 users guide. Chicago, IL: SPSS. Bernard, M.E. (1998). Validation of the General Attitude and Belief Scale. Journal of RationalEmotive and CognitiveBehavior Therapy, 16, 183196. Bentler, P.M. (1990). Comparative t indices in structural equation models. Psychological Bulletin, 107, 238246. Bond, W.B., & Dryden, W. (2000). How rational beliefs and irrational beliefs affect peoples inferences: An experimental investigation. Behavioural and Cognitive Psychotherapy, 28, 3343. Briggs, R.S., & Cheek, J.M. (1986). The role of factor analysis in the development and evaluation of personality scales. Journal of Personality, 54, 106148. Browne, M.W., & Cudeck, R. (1993). Alternative ways of assessing model t. In K.A. Bollen and J.S. Long (Eds.), Testing structural equation models (pp. 136162). Newbury Park, CA: Sage. Carmines, E.G., & McIver, J.P. (1981). Analyzing models with unobserved variables. In G.W. Bohrnstedt, & E.F. Borgatta (Eds), Social measurement: Current issues. Beverly Hills: Sage. Cattell, R.B. (1978). The scientic use of factor analysis. New York: Plenum. Demaria, T.P., Kassinove, H., & Dill, C.A. (1989). Psychometric properties of the survey of personal beliefs: A rationalemotive measure of irrational thinking. Journal of Personality Assessment, 53, 329341. DiGiuseppe, R. (1996). The nature of irrational and rational beliefs: Progress in rational emotive behavior therapy. Journal of Rational Emotive and Cognitive Behavior Therapy, 14, 528. DiGiuseppe, R., Leaf, R.C., Exner, T., & Robin, M. (1988). The development of a measure of irrational/rational think-
387
Zwick, W.R., & Velicer, W.F. (1982). Factors inuencing four rules for determining the number of components to retain. Multivariate Behavioral Research, 17, 253269.