Introduction

After a period of relative neglect, following the pioneering studies reported in the 1970s by Ranson et al. (1977) and by Towler and Coxon (1979), renewed interest is emerging in the United Kingdom in the sociological and psychological understanding of religious professionals, including the publication of significant books based on empirical enquiries among Anglican clergy (Burgess 1998; Francis and Robbins 1999; Warren 2002; Francis et al. 2005), Methodist ministers (Haley and Francis 2006), Pentecostal pastors (Kay 2000), and Roman Catholic priests (Louden and Francis 2003). A key theme running through many of these studies concerns the work-related psychological health of religious professionals, who appear to be attempting to maintain ministry and mission in a largely secular society and with a dwindling workforce.

A second set of recent studies has focused specifically on assessing the work-related psychological health of clergy in the United Kingdom by means of a modified form of the Maslach Burnout Inventory adapted specifically for use among religious professionals. The Maslach Burnout Inventory, designed originally by Maslach and Jackson (1981) for general use among individuals working in the caring professions, conceptualised professional burnout in terms of a balanced-affect model of psychological well-being consistent with the pioneering work of Bradburn (1969). According to this model, professional burnout is characterised by the presence of negative affect and the absence of positive affect. The Maslach Burnout Inventory proposed two measures of negative affect (emotional exhaustion and depersonalisation) and one measure of positive affect (personal accomplishment). Three pioneering studies reported on the use of the modified form of the Maslach Burnout Inventory among 1,071 Anglican clergymen engaged in full-time parochial ministry (Rutledge & Francis, 2004), among 1,468 Roman Catholic priests engaged in parochial ministry (Francis et al., 2004), and 1,278 male stipendiary parochial clergy in the Church of England (Turton 2003).

The three data sets assembled by Rutledge and Francis (2004), Francis et al. (2004), and Turton (2003) have provided an invaluable resource with which various theories concerning both the precipitating causes of burnout and the preventative strategies designed to reduce burnout can be tested. For example, Francis and Rutledge (2000) focused on a contextual factor hypothesised to function as a predictor of clergy burnout, namely the distinctive pressures of rural ministry. Their analyses demonstrated that rural clergy experienced lower levels of positive affect associated with ministry (personal accomplishment) but that rural clergy experienced the same levels of negative affect (emotional exhaustion and depersonalisation) as clergy working in other areas. Turton and Francis (2006) focused on an internal factor hypothesised to function as a predictor of clergy burnout, namely negative parental images. These analyses demonstrated that negative parental images were associated with higher levels of negative affect (emotional exhaustion and depersonalisation) and lower levels of positive affect (personal accomplishment).

Three other studies have focused on personal and professional life-style issues which may serve to reduce professional burnout. Francis and Turton (2004) tested the theory that the reflective practitioner approach to ministry as described by van der Ven (1998) was associated with enhanced work-related psychological health. Their analyses supported this theory. Turton and Francis (2007) tested the theory that confidence in prayer was associated with enhanced work-related psychological health among the clergy. Their analyses supported this theory. Francis et al. (2007) tested the theory that companion animals were associated with enhanced work-related psychological health among celibate Roman Catholic priests. Their data failed to support this theory. While having a cat living in the presbytery emerged as neutral in respect of professional burnout, having a dog living in the presbytery emerged as a predictor of higher levels of professional burnout.

Self-esteem and professional burnout

Against this background, the aim of this present study is to test the theory that recalled self-esteem as a child may function as a significant predictor of individual differences in levels of professional burnout among clergy. This theory is grounded in the synthesis of three areas of research, concerning the definition and measurement of self-esteem, concerning the importance of childhood in shaping self-esteem, and concerning the linkage of self-esteem with psychological distress and work-related psychological health.

The notion of self has been of central interest to psychologists since the emergence of the discipline in the late nineteenth century when James (1890) devoted a chapter to the self in his Principles of Psychology. The self, however, remains an illusive and frequently poorly defined concept. The potential both for clarification and for further confusion has been provided by the generation of a range of compound self-related constructs. Strein (1993) in his review of the literature found at least 15 different ‘self’ terms used by writers, with terms such as self-concept, self-esteem, self-image, self-worth, and self-acceptance being used interchangeably and inconsistently. However, the two most widely used terms were self-concept and self-esteem. Huitt (2004) cites Purkey’s (1970) definition of self-concept as:

The totality of a complex, organised and dynamic system of learned beliefs, attitudes and opinions that each person holds to be true about his or her personal existence. (p. 12)

Huitt (2004) goes on to emphasise that the self-concept is not innate, but is developed by individuals through interacting with their environment and by reflecting on that interaction. Since the self-concept is learned rather than innate it can be modified and changed (Franken 1994). For Huitt, the term self-esteem relates to how we value or feel about ourselves and is the affective or emotional aspect of self. This could be expressed as positive or negative self-esteem.

A growing interest from the 1960s onwards in the assessment and measurement of self-esteem, for clinical, educational and research purposes, led to the development of several good quality instruments (Burns 1979), including two that have stood the test of time: the Rosenberg Self Esteem Inventory (Rosenberg 1965) and the Coopersmith Self Esteem Inventory (Coopersmith, 1967). For example, the Rosenberg instrument has been used recently by Furnham and Cheng (2000), Francis and Kaldor (2002) and Shea and Pritchard (2007), and the Coopersmith instrument has been used recently by Francis (2005), Williams et al. (2006) and Hills et al. (2006). Applications of instruments of this nature have built up a secure base of empirically generated information about the antecedents, correlates and consequences of individual differences in self-esteem throughout the life span (Baumeister et al., 2003; Lipnevich, 2006).

Within this research literature a number of studies have pointed to the importance of the early years of life in shaping levels of self-esteem that may survive into adulthood. For example, Argyle (1969) proposes that from a very early age children reflect on the reactions of others towards them. They learn about themselves through the way that significant adults react to them. Then as the self becomes more strongly differentiated, children begin to compare themselves actively with others or to reflect on comparisons that are made for them. The roles that children play or are allowed to play in childhood will further add to their self-esteem as will identification with the models available to them. Coopersmith’s (1967) detailed study of the self-esteem of 10- to 12-year-old boys is an early example of an investigation into the effects of parental behaviour. The major findings were that the parental behaviours which fostered positive self-esteem were acceptance, defined limits, respect for individual initiative within limits, and an expectation that children would respond constructively to challenge. A more recent study by Furnham and Cheng (2000) showed a correlation between mothers’ ‘reasonable discipline’ and positive self-esteem in their children.

Although an individual’s self-esteem is capable of change, there is a wide consensus in the counselling literature (Rosenberg, 1965; Coopersmith, 1967; Wylie, 1974; Burns, 1979) that the patterns established in childhood can persist and be difficult to modify. Much of counselling practice is based on this assumption, whether it is Cognitive Behavioural therapy (Feltham and Dryden 2004) or practice arising from a psychoanalytic basis such as Transactional Analysis (Stewart & Joines, 2004). This perspective gives added point to research concerned not simply with current self-esteem, but also with recalled self-esteem as a child.

One important recent strand of research has focused on the relationship between self-esteem and aspects of work-related psychological health. For example, Judge et al. (1997) proposed that a positive view of the self consists of four more specific traits, namely self-esteem, generalised self-efficacy, locus of control and low neuroticism. Judge and Bono (2001) then proceeded to carry out a meta-analysis of research studies that examined the relationship of these four traits with job satisfaction and job performance. They concluded that all these traits were significant predictors of job satisfaction and job performance.

The theory that self-esteem may be related to perceived levels of stress within a professional group was tested by Lo (2002) among undergraduate nursing students. The study was conducted among one cohort of students during each year of a 3-year program, receiving responses from 120 students in year one, 112 in year two, and 101 in year three. Self-esteem was assessed by the Rosenberg Self-esteem Inventory (Rosenberg, 1965) and stress was assessed by the 12-item version of the General Health Questionnaire (Goldberg, 1978). The data demonstrated a clear association between the two variables. There are, however, two important caveats regarding the generalisation of these findings to an understanding of work-related psychological health among practising nurses. The first caveat concerns the potential differences between the experiences of trainees and the experiences of seasoned professional practitioners. The second caveat concerns the relevance of the General Health Questionnaire for assessing work-related psychological health, since the 12-item form of this instrument may be conceived as a more general measure of psychological distress (O’Connor et al., 2003).

The relationship between self-esteem and professional burnout may, however, be contaminated by the effects of individual differences in personality. For example, the three major dimensions of personality (extraversion, neuroticism and psychoticism) proposed by Hans Eysenck and assessed by various editions of the Eysenckian personality measures (Eysenck & Eysenck, 1991) have been shown to function as strong predictors of individual differences both in professional burnout (Rutledge & Francis, 2004; Francis et al., 2004) and in self-esteem (Francis et al., 1995; Francis, 1996, 1997, 1998).

The present study proposes, therefore, both to examine the simple bivariate relationship between recalled self-esteem as a child and current levels of professional burnout, and then to calculate the strength of this relationship after taking into account the personality dimensions of extraversion, neuroticism and psychoticism.

Method

Sample

The questionnaire was mailed as part of a larger battery of tests to a random sample, generated by the Church Commissioners’ database, of 2,000 male stipendiary parochial clergy working in the Church of England with at least 5 years experience since ordination to the diaconate. Just 33 of the questionnaires were not successfully delivered, and completed questionnaires were received from 1,278 of the recipients, making an overall response rate of 65%. The respondents comprised one priest in his late twenties, 133 in their thirties, 451 in their forties, 441 in their fifties, 247 in their sixties, three in their seventies and two who did not reveal their age; 87% were married, 10% single, 1% widowed, and 2% separated or divorced and not remarried.

Measures

Burnout

Burnout was assessed by a modified form of the Maslach Burnout Inventory (Rutledge & Francis, 2004) in which each of the three subscales comprised ten items arranged for scoring on a five point Likert scale: agree strongly, agree, not certain, disagree, and disagree strongly. The Emotional Exhaustion subscale included the following items: “I feel burned out from my parish ministry”, and “I feel fatigued in the morning when I get up and have to face another day in the parish.” The Depersonalisation subscale included the following items: “I don’t really care what happens to some of my parishioners”, and “I find it difficult to listen to what some parishioners are really saying to me”. The Personal Accomplishment subscale included the following items: “I have accomplished many worthwhile things in my parish ministry”, and “I feel exhilarated after working closely with my parishioners”.

Personality

Personality was assessed by the short form of the Revised Eysenck Personality Questionnaire (Eysenck et al., 1985) which proposes a 12-item measure of extraversion, a 12-item measure of neuroticism, and a 12-item measure of psychoticism. It also contains a 12-item lie scale. Each item was arranged for scoring on a dichotomous scale: yes and no. Extraversion is assessed by items like: “Do you like mixing with people?” and “Are you a talkative person?” Neuroticism is assessed by items like: “Are your feelings easily hurt?” and “Does your mood often go up and down?” Psychoticism is assessed by items like “Do you prefer to go your own way rather than act by the rules?” and “Would you take drugs which may have strange or dangerous effects?” The lie scale comprises items like “Are all your habits good and desirable ones?” and “Have you ever cheated at a game?”

Recalled self-esteem as a child

Recalled self-esteem as a child was assessed by a new 8-item semantic differential grid (see Table 1 for anchor descriptors), arranged for scoring on a 7-point scale. The following instructions were given to the respondents: “Before 18 years old how would you describe your feelings (on a scale of 1 to 7 circle the number that reflects your experience).”

Table 1 Scale of recalled self-esteem as a child
Full size table

Data analysis

The data were analysed by means of the SPSS Statistical Package employing the frequencies, reliability, Pearson correlation, and regression routines.

Results and discussion

Since the scale of recalled self-esteem as a child is a new instrument, the full set of items for this scale is presented in Table 1, together with the correlations of the individual items with the sum total of the remaining items. The alpha coefficient of .92 confirms the overall homogeneity and internal reliability of this new instrument.

Table 2 presents the scale properties of the three scales of the measure of burnout (emotional exhaustion, depersonalisation, and personal accomplishment), the four scales from the short form of the Revised Eysenck Personality Questionnaire (extraversion, neuroticism, psychoticism, and lie scale) and the scale of recalled self-esteem as a child. In this table and in all subsequent analyses personal accomplishment scores have been reverse coded, so that high scores are indicative of lack of personal accomplishment. All scales apart from the psychoticism scale achieve satisfactory levels of internal homogeneity reliability according to the alpha coefficient (Cronbach, 1951). The lower level of reliability associated with the psychoticism scale is consistent with the recognised difficulties in operationalising this dimension of personality (Francis et al., 1992).

Table 2 Scale properties
Full size table

Table 3 presents the bivariate correlations between the three scales of the measure of burnout (emotional exhaustion, depersonalisation, and personal accomplishment) and the six predictor variables (recalled self-esteem as a child, neuroticism, extraversion, psychoticism, lie scale and age). This table also presents the bivariate correlations between the scale of recalled self-esteem as a child and the three personality variables. These data demonstrate that age is a significant predictor of depersonalisation and emotional exhaustion but not personal accomplishment, with older clergy showing lower levels of emotional exhaustion and depersonalisation; that the personality dimensions of neuroticism and extraversion are particularly significant predictors of individual differences in burnout, with stable extraverts showing lower levels of emotional exhaustion and higher levels of personal accomplishment; and that recalled self-esteem as a child also functions as a significant predictor of individual differences in burnout, with clergy who hold more positive recalled self-esteem as a child showing lower levels of emotional exhaustion, lower levels of depersonalisation, and higher levels of personal accomplishment. The important additional point to note from this table is that the set of personality variables also functions as a significant predictor of individual differences in recalled self-esteem as a child. In particular clergy who record higher scores on the extraversion scale and lower scores on the neuroticism scale (stable extraverts) record significantly more positive recalled self-esteem as a child.

Table 3 Correlations between recalled self-esteem as a child, personality and professional burnout
Full size table

Table 4 employs multiple regression to explore a hypothesised casual path model according to which age and personality are assumed to function as prior determinants of both recalled self-esteem as a child and current levels of professional burnout. The first step in this path model regressed age, extraversion, neuroticism, psychoticism and lie scale scores (in that fixed order) on scores of recalled self-esteem as a child. These data demonstrated that extraversion, neuroticism and psychoticism all function as independent predictors of recalled self-esteem as a child, but that neither age nor lie scale scores added further predictive power. The second step in this path model examined the relationship between recalled self-esteem as a child and the three components of professional burnout in turn, after controlling for individual differences in age and personality. These data demonstrated that, after taking age and personality into account, the variance in professional burnout accounted for by recalled self-esteem as a child was reduced to trivial proportions, and in the case of depersonalisation failed to reach statistical significance. In other words, according to the hypothesised causal model, the apparent association between recalled self-esteem as a child and current levels of professional burnout suggested by the bivariate correlations are an artefact of the influence of personality factors on shaping both recalled self-esteem as a child and current levels of professional burnout.

Table 4 Multiple regression models
Full size table

Conclusion

The present study was designed to build on recent research by Lo (2002) concerning the hypothesised relationship between self-esteem and work-related psychological health. The research design extended Lo’s work in five important ways. First, the measurement of self-esteem was clearly refocused as the assessment of recalled self-esteem as a child in order to capture the formative self-esteem underlying subsequent personal development, uncontaminated by subsequent mature reflection. Second, the measurement of work-related psychological health was operationalised through a balanced affect approach to the conceptualisation and assessment of wellbeing. Third, the research was conducted among professional practitioners rather than among trainees. Fourth, the research context was translated from nurses to clergy. Fifth, the research question was contextualised within the coherent and inclusive model of personality and individual differences proposed by Eysenck’s dimensions of extraversion, neuroticism and psychoticism.

Lo’s hypothesised relationship between self-esteem and work-related psychological health carries two potentially important and practical implications for understanding professional burnout among the clergy. The first implication concerns prediction and preventative strategies. If self-esteem scores can predict those clergy most vulnerable to professional burnout, then preventative strategies may be available to support those most vulnerable. The second implication concerns aetiology and therapeutic strategies. If poor self-esteem as a child can be shown to be not merely correlated with poor work-related psychological health but a factor actually leading to poor work-related psychological health, then therapeutic strategies may be available to improve and to enhance self-esteem.

The data, however, provided only partial support for Lo’s hypothesised relationship between self-esteem and work-related psychological health. On the one hand, the bivariate correlation coefficients demonstrated that poor recalled self-esteem as a child is significantly associated with higher levels of depersonalisation, higher levels of emotional exhaustion and lower levels of personal accomplishment. On the other hand, the multiple regressional model demonstrated that the apparent relationship between self-esteem and work-related psychological health is largely an artifact of the way in which the major dimensions of personality predict individual differences both in work-related psychological health and in recalled self-esteem as a child. After taking these personality dimensions into account, the additional variance in measures of work-related psychological health is reduced to trivial proportions.

Two main scientific conclusions emerge from these findings. First, it remains clear from this study, as well as from the earlier reports by Rutledge and Francis (2004), and Francis et al. (2004), that Eysenck’s dimensional model of personality provided a useful model and mechanism for predicting individual differences in clergy work-related psychological health. If this model were to be routinely applied during initial clergy education and training and during continuing clergy development programmes, those most vulnerable to burnout could be identified. Second, it is clear that recalled self-esteem as a child is not an effective additional variable to pursue in the quest for predicting clergy burnout after taking personality dimensions into account. Lo’s potentially rich insight into the causes of poor work-related psychological health appears to have rested on an over-interpretation of the evidence.

An important practical conclusion then emerges from these scientific conclusions. Therapeutic strategies designed to help clergy deal with vulnerability to poor work-related psychological health would be ill advised to concentrate primarily on reconstructing self-esteem. More effective therapeutic strategies might be based on trying to help clients appreciate why certain personality characteristics (especially low extraversion and high neuroticism) predispose toward poor work-related psychological health. Introverts may need, for example, to learn how to deal more effectively with the demands made by social occasions and by people-orientated aspects of their work. Those who score high on neuroticism may need, for example, to learn how to step back from anxiety inducing situations and to delay engagement with people and issues until their anxiety levels subside.

Overall, this study has provided a good example of how explanatory theories and therapeutic models can be re-evaluated in the light of well-researched empirical evidence. Other theories regarding the aetiology and treatment of poor work-related psychological health and professional burnout among the clergy could be helpfully tested in a similar way by future research.

This study has also introduced and tested a new instrument concerned with the measurement of recalled self-esteem as a child. The development of a new instrument was necessary because the established indices of self-esteem focus on evaluation of current perception not on recalled perception as a child. This instrument has demonstrated good internal homogeneity reliability and can be commended for further exploratory research.