The effect of client characteristics on burnout among helping professionals


Material Information

The effect of client characteristics on burnout among helping professionals
Physical Description:
xi, 174 leaves : ill. ; 28 cm.
McIntosh, Nancy Jones, 1934-
Publication Date:


Subjects / Keywords:
Burn out (Psychology)   ( lcsh )
Job stress   ( lcsh )
Stress (Psychology)   ( lcsh )
bibliography   ( marcgt )
theses   ( marcgt )
non-fiction   ( marcgt )


Thesis (Ph. D.)--University of Florida, 1984.
Includes bibliographical references (leaves 162-173).
Statement of Responsibility:
by Nancy Jones McIntosh.
General Note:
General Note:

Record Information

Source Institution:
University of Florida
Rights Management:
All applicable rights reserved by the source institution and holding location.
Resource Identifier:
aleph - 000473779
notis - ACN8988
oclc - 11665491
System ID:

Full Text







Dedicated to the memory



Whose life challenged, whose integrity inspired, and

whose friendship healed.


Writing a dissertation appears to be, by definition, a long, arduous task.

However, the process can be made more tolerable when support is available

through all phases of work. I want to thank some of the many who have offered

that support to me.

During the long haul, I have had the loving support of my very unusual

family, who although they must often have wondered why I was doing what I

was doing, have always tolerated my doing it--to myself and to them. I love

and thank them. In addition, many Lakeview friends and my colleagues at the

University of South Florida have offered continuing encouragement. Jeannie and

John, with each of whom I have shared laughter, work, and despair, are important

and appreciated.

In the final, rushed, long days of project completion, I have had the

additional, very practical help, and warm, fun support of Judy, who appeared as

if by magic; and of Gert Anderson, who not only typed the dissertation but did so

in constant good humor, offering encouragement at every turn.

Finally, I sincerely appreciate the work of the members of my committee.

Professors William M. Fox (Chairman), H. Joseph Reitz, and James T. McClave

were in all ways helpful, prompt, and challenging in their responses to my work.








I INTRODUCTION..... ......................1

Conceptualization and Definition of Stress ................ 3
Stress as Stimulus-Response . ... .... 3
Stress as an Area of Study ...... .............. 4
Stress as Arousal and Threat. . . 4
Models of Stress ................... .........6
Stress Research Paradigm . . .7
Definitions ................... ............7
Relationships............................... 9
Problems in Stress Research . . ... .11
Research Methods. ............ ..............11
Research Settings .......................... 13
Overview of Study................... ......... 13
Proposed Study ............................ 13
Research Questions. . .... 14
Study Variables .......................... 17
Summary and Preview ......................... 17

II LITERATURE REVIEW.................... ..... 19

The Study of Work-Related Stress . .... 22
Variable Specification, Literature Review, And Hypotheses. 24
Clients Characteristics. . . .. 24
Burnout ..................... ........... 32
Client Characteristics-Burnout Relationship ... 35
Work Load ............................... 36
Social Support. ......................................40
Anxiety ..... ..... ..... ........ .... ........... .. 46

Physical Symptoms. ................. ............... 51
Absence ...................................... 56
Intervening Variables. .............................. 60
Technology ..................................... 63
Task Characteristics ................................ 68

III METHOD a. . . . 74
III METHOD................................ 74

Research Questions and Hypotheses. . ...... .74
Research Design ............................77
Research Strategy .................................. 77
Independent Variables ........................77
Dependent Variables .........................78
Intervening Variables ...... ................. 78
The Sample ....... ...... .... ........... 78
Procedures. . . . .. 84
Instruments .... ............ ....... ..... 86
Dependent Variables.... ................ ......... 86
Intervening Variables ..................... ...90
Independent Variables ....................... 91
Analysis ................... .............93
Unit and Level of Analysis .................... 93
Analytic Procedures ....................... 94

IV RESULTS ................................ .96

Factor Analyses ................... ..........96
Client Characteristics. . . .. .96
Physical Symptoms ................... ... .... 98
Statistical Description of Scales. . . ... 98
Means and Standard Deviations: Dependent Variables .....100
Summary Descriptive Statistics. . . .. 102
Bivariable Correlations ....................... 103
Dependent Variables ..........................103
Independent and Intervening Variables . 103
Reliabilities ................... .......... .104
Test of Hypotheses ................... ....... .107
Client Characteristics . . ... 108
Burnout ............................ ...108
Staged Regressions ..... ............. ...... 123
Effects of Control Variables . .... .127
Summarized Results of Tests of Hypotheses . ... 128

V DISCUSSION ................................ 129

Methodological Concerns . .... 129
Measurement of Client Characteristics . .131
Measurement of Burnout . .. .. .131
Implications of Findings Relevant to Research Questions ..... .132
Research Question 1 . . ... 132
Research Question 2 .................. ...... 134
Research Question 3 . . .. .135

Research Question 4 . . ..136
Summary ............................. 138
Implications of Results for Administrative Practice .. 138
Implications of Results for Conceptualization and Modeling. 140
Implications of Research for Further Research ... .141
Conclusion ................... .............. 142


PARTICIPATION ..........................143




E MASLACH BURNOUT INVENTORY ................... .148

PROFESSIONALS .......................... 149



I PHYSICAL SYMPTOMS SCALE ............ ...... ..153

3 JOB DIAGNOSTIC SURVEY ......................155

K TECHNOLOGY INSTRUMENT ........ .......... 156


M WORK LOAD SCALE .........................160

N SOCIAL SUPPORT INSTRUMENT ................ .. 161

REFERENCES ................... ................ .162

BIOGRAPHICAL SKETCH. ...............................174


3-1 Demographic Characteristics of Respondents. ............ ..80

3-2 Job-Related Characteristics of Respondents ... .. 82

4-1 Orthogonally-Derived Primary Factors of Client . .97
Characteristics Scale

4-2 Orthogonally-Derived Factors of Physical Symptoms 98

4-3 Means, Standard Deviations, and Ranges for Independent, .... .100
Intervening, and Dependent Variables

4-4 Bivariate Correlations: Independent, Intervening, and .. 104
Dependent Variables

4-5 Reliabilities for Independent, Intervening, and. . .106
Dependent Variables

4-6 Interrater Reliabilities on Technology Scale . .. 107

4-7 Regression of Burnout on Client Characteristics . .111

4-8 Regression of Burnout on Client Characteristics. . .112
and Work Load

4-9 Regression of Burnout on Client Characteristics,. . .114
Work Load, and Interaction of Client
Characteristics with Work Load

4-10 Regression of Burnout on Client Characteristics and. .115
Work Load with Range for Client Characteristics
Restricted to Scores Above Median

4-11 Regression of Burnout on Client Characteristics and. 116
Social Support

4-12 Regression of Burnout on Clients Characteristics .... .118
Social Support, and Interaction

4-13 Regression of Burnout on Client Characteristics and ... 119
Social Support, with Range for Client Characteristics
Restricted to Scores Above Sample Median

4-14 Regression of Anxiety on Client Characteristics. . .121
and on Manageability, Likeability, and Treatability

4-15 Regression of Physical Symptoms on Client Characteristics. ..... ..122

4-16 Staged Regressions of Technology on Client. . ... ..125
Characteristics and of Burnout on Technology and
Client Characteristics

4-17 Staged Regressions of Task Characterisitcs on . ... ..126
Burnout, and of Burnout on Client Characteristics



1-1 General Stress Research Paradigm . .... .8

1-2 Direct Relationship Investigated . . 15

1-3 Relative Effects of Various Stressors on Single Strain ... .15

1-4 Variable Prediction of Strains by Single Stressor. .. .16

1-5 Direct and Indirect Effects of Stressors on Strain ... 16

2-1 Effect of Particular Stressor on Specific Strain .. .20

2-2 Relative Effects of Particular Stressor on Single Strain...... 20

2-3 Variable Prediction of Specific Strains by Single Stressor ..... .21

2-4 Direct and Indirect Effects of Particular Stressor on. 21
Specific Strain

Abstract of Dissertation Presented to the Graduate Council
of the University of Florida in Partial Fulfillment of the Requirements for the
Degree of Doctor of Philosophy.



Nancy Jones McIntosh

April, 1984

Chairman: William M. Fox
Major Department: Management and Administrative Sciences

The study was a systematic exploration of the relationship of a single

stressor, client characteristics, to a single strain, burnout, among helping

professionals. In order to view the relationship from a number of angles, client

characteristics and two other stressors, work load, and social support, were

investigated as predictors of burnout. In addition, the differential predictive

effects of client characteristics on burnout and three other strains, anxiety,

physical symptoms, and absence behavior, were examined, and the possibility of

indirect effects of client characteristics on burnout through technology and/or

task characteristics was explored.

Data were gathered by questionnaire from nurses (n = 99) in a children's

hospital in a southeastern state. The surveys were mailed to nurses' homes and

completed by respondents on their own time. The response rate was 55%.

Previously-validated scales were used for the measurement of all variables except

absence frequency, client characteristics, and physical symptoms. A factor-based

scale for measurement of client characteristics was developed for the study.

The physical symptoms scale was used simultaneously in another study for which

it was developed (Gaines, in process). Reliability for the client characteristics

scale was .81 and for the physical symptoms scale was .87.

Results indicate that the primary dimensions of client characteristics are

manageability, likeability, and treatability. Burnout is predicted by client

characteristics (inversely) and family support (inversely), and by work load. The

effects of work load and family support are neutralized when client characteristics

are positively evaluated. Client characteristics are related (inversely) to physical

symptoms, but unrelated to anxiety and absence frequency. Client characteristics

affect burnout indirectly through task characteristics, but not through technology.



As early as 1848, an area of study that the French labeled social medicine

"investigated the contribution of poverty, working conditions, nutrition, housing

and other social factors to disease and mortality" (Garfield, 1979, p. 34). Much

more recently, many behavioral scientists have begun to direct their research

efforts toward understanding quality of work life issues. It is asserted that

improvements in the quality of work life, defined as "the individual's psychological

well-being at work or avowed happiness with work" (Brief, Schuler, and Van Sell,

1981, p. 13) will lead to more positive evaluation of the worker toward self and

the job, to lower rates of turnover and absenteeism, and to improved physical

and psychological health (Hackman & Suttle, 1977). Job stress is one of the

quality of work life issues that affects the psychological and physical well-being

of organizational members.

Although as Kasl (1978) has commented, stress at work, in terms of being

a perceived problem, may not be as major an issue as unemployment, chemical

and physical health hazards in the workplace, or quality of working life in general,

it is an issue that is commanding increasing attention in both the popular press

and academic journals and meetings (Beehr & Newman, 1978; Brief et al. 1981;

Schuler, 1980; Wallis, 1983).

The heightened interest in the phenomenon of stress may be accounted

for in large measure by the seriousness of its consequences and the magnitude

of its costs to individuals and to organizations. House (1981) has asserted that

"a growing body of evidence indicates that psychosocial forms of occupational

stress have deleterious effects on a wide range of physical and mental health

outcomes" (p. 8). More specifically, these outcomes, which may be behavioral,

psychological, physiological or cognitive, and may be temporary and transient or

more long-lasting in nature, include accident proneness, absenteeism, increased

turnover, increased smoking, anxiety, depression, anger, lowered self-esteem,

increased blood and urine catecholamines, increased heart rate and blood pressure,

gastrointestinal disorder, inability to make decisions, and increased experience

of mental blocks (Beehr & Newman, 1978; Cooper & Marshall, 1976; House 1981;

Ivancevich & Matteson, 1980). These consequences, in turn, are associated with

physical illnesses, including peptic ulcer, chronic respiratory diseases, and

cardiovascular diseases, and with mental ill health.

In addition to the obvious human costs to individuals suffering the

consequences of job stress, these reactions are thought to lead to a reduction

in productivity and lowered motivation among stressed workers, which in turn

leads to economic costs for organizations (Zaleznik, Kets de Vries, and Howard,

1977). Efforts to determine the dollar value of these economic costs have

produced estimates that two-thirds of the visits made to family physicians are

prompted by stress-related symptoms which are thought to result in absenteeism,

company medical expense, and lost productivity with estimated costs of $50-75

billion, or $750.00 per worker, annually (Wallis, 1983). Peptic ulcer and

cardiovascular disease, two diseases with a documented relationship to stress,

result in costs of $45 billion annually (Schuler, 1980). Ivancevich and Matteson

(1980), after considering contributory factors including costs of loss of decision-

making effectiveness, of physical and mental health problems and their care, and

of lost work, have estimated the cost of stress in the United States at $75-90

billion annually--or nearly 19% of the gross national product.

The evidence regarding the importance of stress research obviously

justifies what appears to be a surge of interest and research effort on the part

of behavioral scientists and medical researchers. Unfortunately, the need for

information about causes and effects of stress is much more clear than are the

conceptualization of the stress questions and the research methodology for

pursuing answers to them.

Conceptualization and Definition of Stress

Many of the reviews of the stress literature include a lament similar to

Schuler's (1980) statement that "stress remains a term without conceptualization

and without definitional agreement" (p. 187). The emphasis here must be on lack

of agreement, since the literature abounds with conceptualizations and definitions.

An effort to classify the variety of approaches and to identify important

dimensions of the stress concept is presented in order that the present study

may be placed within the context of stress research in general.

Stress as Stimulus-Response

Stimulus. Several researchers, including Hall and Mansfield (1971), specify

that stress is a stimulus. Cooper and Marshall (1976), for example, designated

as occupational stress those "negative environmental stresses or factors associated

with a particular job" (p. 11). In addition to these straightforward views of

stress as stimulus, a more complex approach conceptualizes the stress stimulus as

a lack of congruence among designated variables. For example, French (1974)

defines stress as lack of person-role fit which leads to experiences of strain;

McGrath (1976) as the perception of an environmental demand which might exceed

the perceiver's capability for meeting it; and Zaleznik et al. (1977)

as "disparity between what an individual needs and what the environment offers

in terms of gratification and reward" (p. 160).

Response. Selye (1976), known as the father of stress research,

has defined stress as a "nonspecific response of the body to any demand made

upon it" (p. 14). While Selye's position has been accepted by Levi (1974) and

others, Frankenhaueser (1976) and Mason, Maher, Hartley, Mougey, Perlow, and

Jones (1976) contend, and have demonstrated experimentally, that physiological

responses are specific.

Stimulus and Response. At least two approaches to the

conceptualization of stress include both stimulus and response. Lazarus (1966)

indicated that there is, at least in the exploratory stages of researching stress,

a circularity because "the stress stimulus is defined by the reaction, and the

stress reaction by its relationship with the stress stimulus" (p. 5). In a

somewhat different view, Beehr and Newman (1978) and Margolis and Kroes

(1974) define stress as an interaction of factors in the work situation and the

worker (stimulus) that result in a change in the individual's functioning (response).

Stress as an Area of Study

Another approach to the conceptualization of stress was offered by

Lazarus (1966), explicitly supported by McLean (1974), and implicitly subscribed

to by many researchers and writers in the field of occupational stress. Lazarus

(1966) stated that stress is not "a stimulus, response or intervening variable, but

rather a collective term for an area of study" (p. 27).

Stress as Arousal and Threat

In addition to difficulties and differences encountered by stress

researchers in specifying the denotation of "stress," there is also lack of agreement

about the connotation of the term. While a number of the streams of stress

research (for example, studies dealing with life-event stress or job stress) appear

to treat stress as a negative outcome, Selye (1974) asserted that the absence of

stress is death. Selye (1976) termed negatively valued stress responses "distress"

and positively-valued stress responses "eustress."

One useful approach to distinguishing between the positive and negative

connotations of stress as studied in the organizational literature may be to

conceptualize it as arousal or as threat. Although Lazarus (1966) asserted that

"arousal concentrates on what is general to all kinds of activation, while threat

concerns a particular kind of psychological state which, through the stimulation

of coping processes, has the capacity to arouse but is not equivalent to arousal"

(p. 390), in the job stress literature arousal appears to have a positive connotation.

Scott (1966), in an application of activation theory, discussed the positive

relationships that can be expected among arousal, non-routine task, and

performance. McGrath (1976) reported a positive relationship between experienced

stress or arousal, (operationalized as pulse rate, breathing rate, and behavioral

activity) and performance.

The negative connotation of stress as it is conceptualized in the job

stress literature is reflected in McLean's (1974) conceptualization of stress as

threatening events. McLean (1974) identified three types of events as threatening,

including the threat of losing self-control, the threat to conscience reflected in

experiences of guilt, and the threat of personal physical harm. Lazarus (1966)

indicated that "psychological stress is distinguished by the intervening

variable of threat. Threat implies a state in which the individual anticipates a

confrontation with a harmful condition of some sort" (p. 25). These definitions

of stress as threat are congruent with the implications and the emphasis of much

of the occupational stress literature (Beehr & Newman, 1978; Brief et al., 1981;

House, 1981; Ivancevich & Matteson, 1980; McLean, 1974).

Models of Stress

The study of stress has produced numerous models, as might be expected

in a research domain which is of interest to a variety of disciplines and for

which boundaries are not clearly delineated. In many of the models (House,

1981; Ivancevich & Matteson, 1980; Levi, 1974; Schuler, 1980), the term stress

is used for a class of intervening variables, usually hypothesized to condition the

effect of environmental stressors on strain. French and Caplan (1973) use the

term stress for the independent variable in their model, while others (Beehr &

Newman, 1978; Brief et al., 1981; Kahn, 1974; McGrath, 1976) reserve the term

to refer to the overall process or sequence of events modeled.

Among the models used to research stress in organizations, a few (Beehr

& Newman, 1978; McGrath, 1976; Payne, 1979) emphasize process, while most

(e.g., Brief et al., 1981; Cooper & Marshall, 1976; House, 1981; Kahn, 1974; Levi,

1974) are concerned with content. Process models attend to the processes by

which individuals are hypothesized to progress from one event in the stress

sequence to another. For example, Payne (1979) hypothesized a perception

process, an appraisal process, and an action process, while McGrath (1976),

although he included content variables in his model, placed primary emphasis on

the explication and exploration of the appraisal, decision, performance, and

outcome processes.

Content models specify the sequence of events or conditions hypothesized

to constitute the stress experience. In general, these models include environmental

events or conditions, frequently labeled stressors, individual responses, often

labeled strains, and outcomes, or consequences. In addition, most content models

include individual differences as intervening (Cooper & Marshall, 1976) or

moderating (Brief et al., 1981; Ivancevich & Matteson, 1980; Kahn, 1974) variables.

Stress Research Paradigm

Despite the proliferation of stress models in the past decade, a general

stress research paradigm seems to have emerged. The content model, which is

presented in Figure 1-1 and which follows closely the work of Caplan, Cobb,

French, Jr., Van Harrison and Pinneau (1975) and House (1974, 1981), is

parsimonious while being adequately inclusive of the critical classes of variables

and specifications of relationships among them. This model is used to guide the

present study as a model of job stress, although the model is sufficiently general

that it is useful as a guide for research across disciplines and areas of interest.


Stress is used, following Lazarus (1966), McGrath (1976), and McLean

(1974), as a general term for the area of study and to designate the overall

experience, including antecedents and consequences, being researched. This

experience may be conceptualized as a transactional process between an individual

and a situation that involves physical or psychological threat for the individual

(Lazarus, 1966).

Stressor refers to any job dimension or characteristic that poses a physical

or psychological threat to an individual. Stressors may be objectively threatening,

in which case they are termed environmental stressors, or subjectively determined

to be threatening, in which case they are termed perceived stressors. Shinn

(1979) distinguished between two types of stressors in the work environment:

stressful job events and ongoing stressors. Among the many characteristics of

the work place that have been considered stressors in theoretical and empirical

stress studies are task difficulty and ambiguity, role overload, minimal decision-

making power, interpersonal disagreement, and isolation (Cooper & Marshall, 1978;

Maslach, 1982; McGrath, 1976).










Strain refers to the relatively short-term psychological, physiological and

behavioral responses of an individual to the presence of a stressor. Strain is

sometimes modeled as an intervening variable through which stressors exert an

indirect effect on outcomes. Illustrative of psychological strains are anxiety,

depression, job dissatisfaction, and somatic complaints; of physiological strains

are heart rate, respiration, blood pressure, serum cholesterol and adrenaline levels;

and of behavioral strains are increased smoking, absence, loss of appetite, accident

proneness, and change in performance level or level of work involvement (Beehr

& Newman, 1978; Caplan, et al. 1975; House, 1981).

Outcomes refer to the long-term psychological, physiological and/or

behavioral consequences to the individual and/or the organization of stress

experience. Among the consequences of primary interest are decrement in

organizational effectiveness, and changes in physical and mental health, including

incidences of cardiovascular disease, gastrointestinal problems such as peptic

ulcers, and schizophrenia.

Conditioning variables refer to those characteristics of the individual or

situation that moderate the direct relationship between other variables in the

model. Several models of job stress (Brief et al., 1981; French & Caplan, 1973;

Ivancevich & Matteson, 1980; Kahn, 1974) suggest that individual differences,

including Type A personality, flexibility-rigidity, self-esteem, education, and age,

moderate the relationships between stressor and strain and between strain and

outcome. In addition, situational factors such as social support appear to moderate

these relationships (House, 1981).


In the model (Figure 1-1), each arrow that connects one box to another

(1, 2, 3, 4, 5, 6) represents a hypothesized direct relationship. In addition, a

sequence of arrows (1-2-3; 4-3; 1-5) represents a hypothesized indirect relationship

between variables in the first and last boxes connected by the arrows. Further,

conditioning variables are hypothesized to alter the direct relationships

represented by the arrows to which the conditioning variable arrow points.

For example, environmental stressors are hypothesized to have a direct

effect on perceived stressors (arrow 1), on strain (arrow 4), and on outcomes

(arrow 6). In addition, environmental stressors are hypothesized to have an

indirect effect on psychological strain through perceived stressors (sequence 1-

2), on outcomes through perceived stressors (sequence 1-5), and on outcomes

through perceived stressors and strain (sequence 1-2-3). Further, it is hypothesized

that individual and situational variables will moderate each direct and indirect

effect of environmental stressors (arrows 7, 8, 9, 10, 11).

Many of the relationships hypothesized by the stress model are supported

by empirical evidence from the job stress literature. Although a review of the

evidence for relationships among classes of variables included in this study is

presented in Chapter II, examples are noted here. Sales (1970) found that

environmental stressors and perceived stressors (arrow 1) are significantly related,

and Caplan and Jones (1975) reported that both objective and subjective stressors

predict strain (arrows 1 and 4), although subjective stressors are the better

predictor class. Zaleznik et al. (1977) offered evidence supporting the

environmental stressor to strain and the environmental stressor to outcome

relationships (arrows 4 and 6). Frankenhaueser and Gardell (1976) and Kohn and

Schooler (1973) found evidence supporting the environmental stress to

psychological strain and environmental stress to physiological strain relationships

(arrow 4) respectively. Cooper and Marshall (1976) cited a growing body of

evidence linking working conditions (environmental and perceived stressors) to

mental and physical ill health outcomes (arrows 5 and 6), and perceived stressors

to strain (arrow 2). Levi (1974) reported evidence linking mechanisms (strains)

to disease (outcomes; arrow 3).

Problems in Stress Research

Research Methods

Beehr and Newman (1978) observed that relatively few of the variables

from each class of variables in the stress model have been investigated empirically,

and called for a systematic approach to the study of relationships among variables.

Response to this appeal requires that researchers in the area of stress deal with

a number of relatively complex methodological issues, including clear specification

of types of variables, explicit statement and support of assumptions of causality,

and use of multivariate analyses to ascertain the explanatory contribution of

multiple predictor variables and the nature of relationships between variables.

Cooper and Marshall (1976), Kast (1978), and Parkes (1982) have

emphasized the necessity of avoiding confounding perceptions of the work place

with responses to perceptions, and/or of confounding independent and dependent

variables due to lack of conceptual and operational clarity. Lazarus (1966),

having pointed out that stress researchers sometimes emphasize antecedent

conditions of stress and other times emphasize response aspects of stress, declared

that "one aspect of the problem (the antecedents of stress) can never really be

separated from the other (the nature of the stress reaction). Stress theory

requires an analysis of the mechanisms and conditions that link them" (p. 9).

However, as Cooper and Marshall (1976) have observed, much of the

stress research has relied on correlational analysis which does not permit

inferences about direction of causality nor about intervening variables. In order

to specify independent, dependent, and intervening variables, assumptions about

or demonstrations of direction of causality need to be made explicit. The

conclusion drawn by Cooper and Marshall (1976) that their extensive review of

the stress literature provided "seminal evidence to support the notion that work

environment and model organizations have an impact on the physical and mental

health of their members" (p. 25) is strengthened by subsequent longitudinal studies,

including those of Karasek (1979) and Parkes (1982), which confirm the job-to-

strain direction of causality inferred previously from (primarily) cross-sectional


Caplan et al. (1975) asserted that once the causal role of job stress has

been established, it is required that specific stressors for particular jobs be

identified. However, specification of the impact of each stressor will require

application of more sophisticated multivariate analysis techniques, since there is

frequently correlation between and among various predictors (Kasl, 1973). In

addition to ascertaining the independent effects of stressors, it is important to

determine the type of functional relationships between stress and strains. Levi

(1972) has suggested that the relationship between stress and stimulation level is

of the inverted U form; McGrath (1976) found a positive, linear relationship

between arousal and performance when task difficulty was controlled for; and

Hackman and Oldham (1976) reported a positive linear relationship between

desirable task characteristics and satisfaction. The need for considering other

than linear relationships is underscored by Schuler's (1980) conclusion, based on

a review of stress literature, that stressors may have different patterns of

relationships with different patterns of stress. A related problem in the conduct

of stress research is that of restricted range of values for the dependent variable


(McLean, 1974). This restriction occurs in experimental studies because ethical

considerations, along with the subjects' expectations that they will be unharmed

in experimental situations, preclude inducing high levels of stress. It may occur

in natural field studies because those most affected by stress select themselves

out of the situation.

Research Settings

Schuler (1980) concluded, based on his review of research on the

relationship of occupational type and stress, that, when other organizational

factors are held constant, the most stress occurs for individuals in managerial

positions and health care professions. Randolph, Price, and Collins (no date)

have reviewed literature documenting high levels of stress in hospital systems.

Hay and Oken (1972) and Vreeland and Ellis (1969) reported on the particular

stress experiences of Intensive Care Unit (ICU) nurses, Leatt and Schneck (1980)

on those of head nurses, and Parkes (1982), on those of student nurses. Despite

the appearance of a few studies (Gray-Toft & Anderson, 1981; Leatt & Schneck,

1980; Parkes, 1982) in recent years concentrating on stress among nurses, many

questions remain unanswered. For example, no research studies appear to have

been done in hospital settings with nurses as subjects and burnout as the dependent

variable (Perlman & Hartman, 1982; Randolph et al., no date).

In addition to the need for systematic research studies of stress among

nurses in hospital settings, Beehr and Newman (1978) and House (1981) have

called for stress research with female subjects, after noting that virtually all

research on stress has used at least predominantly male samples.

Overview of Study

Proposed Study

The need for more systematic explorations of relationships among variables

in the stress research paradigm is reflected in the foregoing summary of the

stress research literature. A review of the literature relevant to antecedents

and types of strain among helping professionals, detailed in Chapter II, suggests

that characteristics of clients and client-helper relationships lead to strain among

helpers. The review further suggests that an important manifestation of strain

among helpers is a syndrome that has been labeled burnout. Consequently, the

present study attempts to respond to both the methodological and content issues

of interest by systematically exploring the relationship of the particular job

stressor, client characteristics (see page 27 for definition) to the particular strain,

burnout (see page 35 for definition). The systematic exploration involves

examining the direct relationship of client characteristics to burnout; examining

the relative effects of client characteristics and other strains on the single

strain, burnout; determining whether the single stressor, client characteristics,

predicts burnout and other variables of the strain class of variables differentially;

and testing for indirect effects of client characteristics on burnout through

intervening mechanisms. The study also represents an additional test of the

hypothesized relationship between stressor and strain within the stress research

paradigm, and explores the possibility that the model needs to be expanded to

allow for intervening mechanisms to explain the means by which stressors affect

strain (Lazarus, 1966). A general representation of the research approach is

presented in Figures 1-2, 1-3, 1-4, and 1-5.

In summary, the purpose of this study is to clarify empirically, through

systematic application of the stress research paradigm, the effect of client

characteristics on burnout among helping professionals.

Research Questions

The research questions addressed by the study include:

1. What are the effects of client characteristics on burnout?

FIGURE 1-2 Direct Relationship Investigated

* Client Characteristics
* Burnout

FIGURE 1-3 Relative Effects of Various Stressors on Single Strain



FIGURE 1-4 Variable Prediction of Strains by Single Stressor



FIGURE 1-5 Direct and Indirect Effects of Stressor on Strain

Client Characteristics

2. What are the comparative effects of client characteristics

and other stressors on burnout?

3. Do client characteristics predict burnout and other strains


4. What is the mechanism through which client characteristics affect


5. Is the stressor-strain (arrow 2, Figure 1-1) relationship

supported by the study data?

Study Variables

Independent Variables. Independent variables in the study are

representative of perceived stressors in the stress research paradigm and include

client characteristics, workload and social support.

Dependent Variables. Dependent variables in the study, representing the

psychological, physiological and behavioral components of the strain class of

variables in the stress research paradigm, include burnout, anxiety, physical

symptoms and absence.

Intervening Variables. The intervening variables explored are technology

and task characteristics.

Summary and Preview

Chapter I has presented an overview of the general area of stress research.

It has been noted that stress can be conceptualized as stimulus, as response, as

a dynamic phenomenon involving both stimulus and response, or as a general

rubric representing a field of study. Stress can also be classified using the

dimensions of arousal, which generally has positive connotations, or threat, which

has negative connotations.

Several stress models, representing general stress models and job stress

models, and classified as process or content models, were reviewed briefly.

Figure 1-1 presents a general content model of the stress phenomenon. Drawn

largely from Caplan et al. (1975) and House (1981), the model, equivalently

referred to as the stress research paradigm, comprises four classes of variables

and hypothesizes relationships among them. The model also provides for examining

the moderating effects of situational and/or individual conditioning variables on

the various direct and indirect relationships between and among the environmental

stressors, the perceived stressors, strain, and outcomes.

Research problems salient to the study of stress in organizations were

reviewed, including the need to specify conceptually and operationally the

variables in a study in a way that prevents confounding of the independent and

dependent variables, the need to make explicit assumptions about the direction

of causality, the need to specify intervening variables, the necessity of

multivariate analysis to facilitate more systematic research in the area, and the

importance of explaining not only the existence but also the form of relationships

among variables. The purpose of the present study was introduced, and research

questions and variables specified.

Chapter II presents a rationale for the selection of variables to be studied,

reviews the literature pertinent to the variables, and evolves hypotheses based

on the literature.



A review of the job stress literature revealed a need to approach stress

research systematically to determine whether available models are adequate to

predict and explain relationships of interest. Much of the job stress research

has examined relationships in isolation, making comparison, interpretation, and

extrapolation difficult. This study uses the core relationship of one stressor of

primary interest (client characteristics) to one strain of primary interest (burnout)

as a basis for examining relationships hypothesized by the general stress paradigm

(Figure 1-1).

In addition to the basic research question regarding the relationship

between client characteristics and burnout, four related research questions are


1. What are the comparative effects of client characteristics

and other stressors on burnout?

2. Do client characteristics predict burnout and other strains


3. What is the mechanism through which client characteristics

affect burnout?

4. Is the stressor-strain relationship supported by the study data?

Additional variables specified for the study are work load and social

Figure 2-1

Effect of Particular Stress on Specific Strain



Figure 2-2 Relative Effects of Particular Stressors on Single Strain






Charac teristicss
Social Support
Work Load


Figure 2-3 Variable Prediction of Specific Strains By Single Stressor

STechnology I I


Figure 2-4 Direct and Indirect Effects of Particular Stressor on Specific






support (stressors), anxiety, absence and physical symptoms (strains), and task

characteristics and technology (intervening variables). Models representing the

framework used for exploring the relationship among the specified variables are

represented by figures 2-1, 2-2, 2-3, and 2-4.

In this chapter, following a brief review of the general literature

demonstrating the relationships of work-related factors to stress, a review of

the more specific literature is used to develop the rationale for the inclusion of

each variable in the study, to present results of empirical studies that included

the variables, and to evolve hypotheses relevant to the research questions. An

effort has been made to organize the literature review so that studies relative

to the core relationship being examined (client characteristics to burnout) are

presented first, followed by those dealing with the other specified stressors and

their relationship to burnout. The literature relevant to each additional specified

strain and its relationship to client characteristics is then introduced. Finally,

the intervening variables and their relationships to the specified strains are


The Study of Work-Related Stress

Researchers' efforts to specify causes and consequences of work-related

stress have resulted in a diversity of approaches to study of the phenomenon.

One general approach (Kasl, 1978) involves comparison of the relative prevalence

of stress among groups whose members are alike on some dimension, such as

occupation, occupational level, unit membership or type of organizational role

(e.g., line or staff). Having determined that stress is more prevalent in one

group than another, the researchers search for explanations of the differences.

For example, Zaleznick et al. (1977) found that managerial personnel, compared

to staff or operations personnel, experience lower levels of stress, and attributed

the difference to what they termed the "bureaucratic effect". Gore (1978) found

that rural unemployed experienced lower levels of stress than urban unemployed,

and attributed the difference to varying degrees of social support. Kornhauser

(1965) found that mental health is directly related to occupational skill level

and, because neither "prejob" personal factors nor selection procedures appear

to account for the difference, attributed it to the nature of the work.

Cooper and Marshall (1976) noted that a trend in stress literature appeared

to be toward looking at significant job components instead of toward broad

occupational level or discrete occupational groupings for predictors of stress.

This approach involves researchers' using theory, previous empirical findings

and/or their own observations to identify factors in the work environment thought

to be related to stress, and stress responses thought to be related to factors in

the working environment. Gavin and Axelrod (1977), for example, found that

scores on an affective scale which included measures of anxiety, depression and

irritation, increased with increases in ambiguity, role conflict, and work load,

and with decreases in job security, participation, and skill utilization. In a series

of studies undertaken through the University of Michigan's Institute for Social

Research (IRS), researchers have used this approach to identify work-related

stressors including role ambiguity (Kahn, Wolfe, Quinn & Snock, 1964), role

conflict (Caplan, 1971; Kahn et al., 1964; Sales, 1970), and overload (French &

Caplan, 1973). French and Caplan (1973), in a summary of the IRS studies of

job stress, indicated that they have identified nine strains: job dissatisfaction,

job tension, lower self-esteem, threat, embarrassment, high cholesterol levels,

increased heart rate, skin resistance, and increased smoking.

Despite the large "gaps" in our knowledge about stressors and strains,

and the relationships between and among them (Cooper & Marshall, 1976), and

regardless of the research approach chosen, the evidence clearly indicates that

relationships exist between factors in the work place and stress experiences of

those who work there.

Variable Specification, Literature Review, and Hypotheses

Client Characteristics

Concept and Defintion. Because they are, in essence, both consumers

of service and the raw material upon which the service is performed, clients in

human service organizations have a critical impact on organizational functioning.

In fact, Hasenfeld (1978) has asserted that "the essential character of any human

service organization will be manifested through the patterned relations between

its clients and staff, relations which are the raison d' etre of the organization"

(p. 184). These patterned relations constitute a social system (Katz and Danet,

1973) in which the interpersonal event (Jermier, 1982) is the primary and definitive


The client-organization or client-helper interaction may be viewed from

a number of perspectives. Hasenfeld (1978) noted that one research tradition

has explored client attributes as determinants of client's utilization of available

services, and that a second has focused on organizational influences on client-

staff interaction. Other approaches (Danet, 1981; McKinlay, 1975) have been

primarily concerned with the distribution of control in the relationship, reflecting

a concern that clients will not be served by agencies in which they have no

influence. Wills (1978) approached the review of literature relevant to the client-

helper interaction with the declaration that "the helper's perception of a client's

personality is an important aspect of therapeutic relationships" (p. 968).

Each of these approaches implies a primary concern with the effect of

the client-staff interaction on the client. An alternative approach explores the

consequences for the helper of various traits of the client and characteristics

of client-helper transactions. Mennerick (1974) discussed the client-staff

relationship as one that produces conflict for the helper, and the typing of clients

as a process useful for coping with the stress resulting from the conflict. Maslach

(1978) emphasized the demanding nature of the client-helper interaction for the

helper, noting that "the providers of human services are usually required to work

intensely and intimately with people on a large-scale, continuous basis. They

learn about these people's physiological, social and physical problems and are

expected to provide aid or treatment of some kind" (p. 112). These required

interpersonal events, which are frequently emotion-laden, can be very stressful

to client and helper. Recent work by Cherniss (1980), Maslach (1978; 1982), and

Jones (1981) emphasized the high costs to helpers of these client-professional


A number of authors have suggested dimensions of clients' identities that

influence responses of professionals. Danet (1981), in her review of the relevant

literature, identified sex, race, appearance, and similarity of client to helper

among factors that affect helper response. McKinlay (1975) contended that,

because the status of clients is reflected in the status of professionals serving

them, the status of an individual is an important determinant not only of type

of helper response, but also of whether the person will be accepted for treatment.

Lefton and Rosengren (1974) noted that, although the helping agency usually

provides services for only one aspect of the client's person, the whole person is

presented. They suggested that the specific and contrasting interests that

organizations have in clients may be characterized on two dimensions: laterality

and longitudinality. The former dimension refers to the biographical space:

whether it involves many or a single aspect of the client's life; and the latter

refers to biographical time: whether it involves the full history and future of

the individual or a brief, circumscribed period. Shinn's (1979) observation that

there are two types of stressors in the work environment, stressful events and

ongoing stressors, is related to the longitudinal dimension. Client problems may

be emergent and acute, or they may be chronic.

Wills (1978), after reviewing the literature relevant to helpers' perceptions

of clients, concluded that helpers categorize clients along three underlying

dimensions (manageability, treatability and likeability), based on the helper's

response to three questions about the client: "To what extent will the client pose

a management problem? To what extent will the client improve? To what extent

will I like the client?"

Maslach (1978) identified three sources of stress for the helper in the

client-helper relationship: "type of problems facing the client, the nature of

the staff member's relationship to the client and rules regulating the relationship,

and the client stance and reactions to staff" (p. 114). Problems of the client

impact the relationship because they cause the focus of the interaction to be

on negative aspects of the client. In addition, the number, intensity, and timing

of interactions required to address the problem, the degree of stress implied by

the nature of the problem, and the probability that the problem is solvable all

relate to the type and intensity of the impact that the problem will have on

the client-helper relationship.

Based upon this summary of a review of the relevant (primarily) conceptual

literature, a measure of client characteristics is used as the independent variable

of primary interest in this study. In this context, the client characteristics

concept refers to the evaluative and descriptive assessment of aspects of the

client's identity that are relevant to his/her role as client.

Empirical Evidence. Several studies that relate client characteristics to

affective or general stress responses were found. Caplan (1972) reported that

responsibility for people appears to be associated with cigarette smoking and

with diastolic blood pressure, both of which increase risk of coronary heart

disease. In a survey of British male employees, Cherry (1978) found that contact

with people was a significant predictor of nervous strain, even with the level of

functioning, the most powerful predictor used, in the model. Kahn (1974) reported

finding that dealing simultaneously with people inside and outside the organization

(boundary spanning) was productive of conflict, while Miles (1980b) found in a

study of research and development roles, that the linking and coordinating

activities of boundary spanners led to role conflict when information gathering

and transfer were controlled, but that information gathering and transfer activities

were unrelated to role conflict when linking and coordinating activities were

controlled. Futhermore, both information gathering and transfer activities and

linking and coordinating activities were positively related to satisfaction. It

appears, then, that the client interaction, even when conceptualized in a relatively

unfocused manner (i.e., as responsibility for, contact with) is related to strain

and to other affective outcomes. The Miles (1980b) results signal the importance

of recognizing that a single work factor may have both negative and positive


Evidence also exists in the literature that more narrowly focused

conceptualizations of client characteristics are related to stress among helpers.

Difficult clients were among the leading five causes of emotional stress listed

by caregivers (Koocher, 1979), and nurses identified changed physical or

psychological condition of patients as a major stressor (Vreeland & Ellis, 1969).

Pratt (1979), in a study of primary grade teachers, found that hostile acts and

non-cooperative acts of students were the two most commonly identified stressful

events, and that financial hardship of children's homes is strongly related to

teachers' viewing their jobs as stressful and difficult. In a study conducted

among policemen, Jermier (1982) found that environmental danger, which is

defined by the nature and distribution of interpersonal events and was measured

objectively, was a significant predictor of performance, but not of commitment.

Jermier, Gaines, and McIntosh (1983), however, found that perceived physical

danger was negatively related to commitment and pay satisfaction and unrelated

to intrinsic satisfaction.

In summary, a variety of empirical approaches has provided general

support for a client characteristics-general stress relationship, but the relation-

ships are sometimes (Miles, 1980b; Jermier, 1982) not in the expected direction

and/or not of the expected intensity.

Relationship of Stress Among Nurses. Maslach (1978) suggested that work

place stressors for health professionals include a frequent need to work with too

many patients, the necessity for making critical decisions with inadequate

information, and the potential seriousness and costs of making poor decisions.

From her review of relevant literature, Marshall (1980) identified a number of

stressors in the nursing job itself: heavy physical work, work overload, the actual

unpleasantness of the tasks, the need to use sophisticated equipment skillfully,

and lack of privacy in the work place.

It seems reasonable to assume that because nursing by definition involves

direct contact with clients, and the contact may be emotionally charged (Maslach,

1978) and/or physically demanding, nurses are particularly vulnerable to client-

related stress in the work place. Several factors, related to the concept of

stress as threat, appear to contribute to this vulnerability. Nurses experience

a threat to competence (Cherniss, 1980; Kramer, 1974, Marshall, 1980). In fact,

in a list of potentially stressful conditions and events generated from and rated

by nurses, insecurity about knowledge and competence was rated as being among

the three most stressful items (Jacobson, 1978). The nurse's sense of competence

may be threatened because she lacks information that would be helpful to patients

and their families (Marshall, 1980; Maslach, 1978) or because she fears failure.

Koocher (1979) observed that the idea of failure is noxious to health care

professionals, and that for workers whose training and socialization require them

to value healing, untreatable patients represent failure.

Exacerbating these related stressors of threat to competence and threat

of failure is the high degree of visibility inherent in the nurse's work. She is

not spared, as Zaleznick et al. (1977) demonstrated that managers are, from

continuous evaluation. Ironically, her failures are more readily evident to her

and to others because of their immediacy and visibility, and her successes often

leave and are never heard from again. The costs for nurses of the simultaneously

occurring conditions of threat and visibility (with its corollary condition of lack

of privacy) are compounded by the expectation that nurses will show no symptoms

of stress among patients, families, physicians or colleagues; the fact that stress

experienced by nurses is not an acceptable area of concern; and the requirement

that the nurses assist patients and their friends and families in dealing with their


The patient-nurse relationship is central to the potential stress experience

of nurses. It is within the context of this relationship that competence is

demonstrated or failure occurs and it is with the patient that these results are

most visible. Pines and Kafry (1978) observed that "most everyday human

relationships are symmetrical, but the therapeutic relationship is complementary:

one person gives, the other receives" (p. 499). While it is obvious that satisfaction

also accrues from the patient-nurse relationship, a number of stressful aspects

of the interaction have been identified.

Cobb (1974) identified responsibility for persons--particularly for the

future of persons--as a stressor, and suggested that the less distance between

the worker and the persons for whom she is responsible, the more strain will

result. Similarly, Maslach (1978) asserted that the frequency of contact with

clients is related to the amount of burnout experienced.

French and Caplan (1973) summarized findings that demonstrated that

boundary spanning activities among workers are related to experienced role

conflict. The necessity for nurses to interact frequently with persons external

to the unit--including families and friends of patients, providers from within the

hospital of auxiliary services, and physicians--suggests vulnerability to stress from

this source.

Nurses must also deal with the stress that accompanies giving personal

care to patients whose conditions they may find extremely unattractive. In a

study by Vreeland and Ellis (1969), nurses identified the changed physiological

condition and/or psychological impact of a patient's illness or treatment as the

most stressful part of the job. Hay and Oken (1972), cited above, described

the ICU as "a situation that involves intimacy with the frightening, repulsive

and forbidden. Stimuli are present to materialize literally every conflictual area

at every stage of psychological development" (p. 122).

Another class of patients whose problems are particularly stressful for

nurses is that of the dying patient. Dealing with the dying provides a constant

reminder to the nurse of her own mortality, presents her with a failure in terms

of her training and commitment to heal, and places extreme demand on her for

both emotional control and giving emotional support. Marshall (1980) observed

that the trends (hospices notwithstanding) toward dying in hospitals and toward

reducing the number of nurses with whom each patient interacts lead to more

constant dealing with dying patients by some nurses.

Clients can represent physical danger to nurses in several respects. On

some units, nurses are actually subject to exposure to communicable diseases.

They are frequently required to do heavy lifting and other physically difficult

work, and are sometimes subject to erratic behavior of irrationally ill patients.

A somewhat less obvious stress that results for nurses as a result of the

patient-nurse relationship involves ethical issues. For example, with the

legalization of abortion, nurses, who may or may not believe that abortion is

morally acceptable, are caught in a conflict between the role of life-saver and

the act of life-taking.

These and other obvious and subtle implications of the patient-nurse

relationship, coupled with the vulnerability of nurses to threat, support Marshall's

(1980) conclusion that "the job of nurse incorporates several distinctive features

which set it apart as a special case in the stress literature" (p. 20).


Concept and Definition. Although there is not general agreement about

whether it gives a name to a problem that has been prevalent over the years,

describes a problem of recent development, or creates a problem (Maslach, 1978;

Paine, 1982), the term "burnout" has appeared with increasing frequency in both

scholarly and popular publications in the last decade. The burnout phenomenon

can be conceptualized in terms of persons who are at risk, predisposing conditions,

general stressor antecedents, the process by which it occurs, and the indicators

of its presence.

Persons considered most at risk of experiencing the burnout phenomenon

are those in professions with a high degree of people contact (Perlman & Hartman,

1982) who work "intensely and often intimately with people troubled by physical,

psychological and/or social stress" (Jackson & Maslach, 1982, p. 64). These

professions include nurses, physicians, police, counselors, psychologists, teachers,

and ministers.

Freudenberger's (1975) description of the committed worker outlines

conditions that seem to predispose workers to the experience of burnout. First,

the desire to help creates, in itself, a pressure that pushes the worker to try

harder, but in reality can lead to lowered effectiveness. Second, the populations

served by helping professionals are often in extreme need and so "continually

take, suck, demand" (Freudenberger, 1975, p. 75). In addition, work which was

initially exciting to the professional becomes dull and monotonous. These

conditions may well render the helper vulnerable to a variety of stressors,

including problems associated with self-doubts about competence, impatience with

lack of client motivation, bureaucratic interference, and difficulties with peer

relationships (Cherniss, 1980).


Maslach and Pines (1977) have described the process by which the people,

conditions and stressors interact to result in burnout. Helping

professionals are required to work intensely and intimately with
people on a large-scale and continuous basis. They learn about
these people's problems and are expected to provide treatment
of some kind. .This type of professional interaction arouses strong
feelings of emotion and personal stress which can be disruptive and
incapacitating. In order to perform efficiently and well in such
situations, the professional may defend against these strong emotions
through techniques of detachment. By treating one's clients or
patients in a more remote, distant way, it becomes easier to perform
the necessary interviews, tests or operations without suffering strong
psychological discomfort (p. 100).

Among indicators that the burnout process has in fact occurred, according

to Carrol and White (1982), are a significant decrease in the quality of services

provided customers or clients, poor staff morale, and increased negative behaviors

such as employee theft, drug abuse and accidents.

Burnout has been defined as a special and distinctive kind of emotional

exhaustion (Maslach, 1978), and in terms of outcomes (Carrol & White, 1982).

Although Kahn's (1978) inclusion of (inappropriate) attitudes toward clients and

toward self is similar to many definitions, his specification of physical symptoms

is not. Maslach (1982) has noted that despite the variety in the definitions of

burnout, there are three areas of general agreement: 1) "burnout is a phenomenon

that occurs at the individual level"; 2) "burnout is an inner psychological

experience involving feelings, attitudes, motives and expectations"; and 3) "burnout

is a negative experience for the individual, in that it concerns problems, distress,

discomfort, dysfunction and/or negative consequences" (p. 32). Maslach has also

suggested that there is basic agreement on three underlying dimensions of burnout:

exhaustion or fatigue; a change toward more negative responses to others; and

a change toward more negative evaluation of oneself. For the purposes of this

study, burnout is defined "as a syndrome of physical and emotional exhaustion,

involving the development of negative self concept, negative job attitudes, and

loss of concern and feeling for clients" (Pines and Maslach, 1978, p. 233).

Empirical Evidence. Although a large number of articles and books have

been published on burnout, there have been relatively few empirical studies using

the burnout variable with a variety of work place stressors. Cherniss (1980), in

a longitudinal, qualitative study of 28 new professionals, discovered that in his

sample, which included male and female subjects, increased burnout was associated

with lack of goal clarity, low intellectual stimulation, and decreased autonomy

at work. Contrary to expectations created by the non-empirical burnout literature,

increased client contact was associated with increased positive attitudes. Pines

and Kafry (1978), in a study of social workers in which the population was 85%

female, found that worker relations and work sharing were negatively related to

tedium. Tedium is a strain that is defined in terms of emotional exhaustion and

burnout, and related to a desire to leave the job, negative attitudes toward

clients, and negative work attitudes. In a study of child abuse workers conducted

by Berkeley Planning Associates (1977), poor program leadership, low staff

communication, little supervisory responsibility, and high levels of rule

formalization were associated with higher levels of burnout. Barad (1979) found

that poor quality of leadership was a predictor of burnout among Social Security

field contact employees, also. In similarly designed studies of day care workers

(Maslach & Pines, 1977) and of mental health workers (Pines & Maslach, 1978),

the researchers found that the number of time outs possible for the helper was

negatively related to burnout. This possibility of respite had the concurrent

effect of reducing the relative amount of direct client contact. Jones (1981),

however, found that increased unauthorized work breaks were associated with

higher incidence of burnout among a sample of nurses. In the day care setting

increased program structure was related to less positive appraisal of the working

situation and to decreased burnout.

In summary, it appears that among professionals lack of some degree of

control or influence in the work place (autonomy, Cherniss, 1980a; time out

possibilities, Maslach & Pines, 1977, and Pines & Maslach, 1978; rule formalization,

Berkeley Associates, 1977), low intrinsic motivating properties of the job

(stimulation, Cherniss, 1980; supervisory responsibility, Berkely Associates, 1977),

and unsatisfactory social interaction at work (poor superivsion, Barad, 1979; low

staff communication, Berkeley Associates, 1977; work relations and work sharing,

Pines & Kafry, 1978) are related to increased incidence of burnout. On the

other hand, clear direction is related to lower incidence of burnout (program

structure, Maslach & Pines, 1977; goal clarity, Cherniss, 1980). The only

unexpected finding was the negative relationship between client contact and

burnout (Cherniss, 1980), and the conflicting findings regarding time out (Maslach

& Pines, 1977; Pines & Maslach, 1978; Jones, 1981). It may be that time away

from the job serves either an escapist or a preventive function.

Client Characteristics-Burnout Relationship

Empirical Evidence. Of the relatively few empirical studies done on

burnout, several report on the effect of some aspect of the client-professional

relationship on burnout. Pines and Kafry (1978) found that number of clients

served by the social workers in the study was significantly related to the incidence

of burnout. Similarly, Maslach and Pines (1977) and Pines and Maslach (1978)

found that the ratio of clients to staff was a significant correlate of burnout,

with an increase in the ratio associated with an increase in burnout. Pines and

Maslach (1977) also found that contact time with patients and increased incidence

of schizophrenia among patients were related positively and significantly

with burnout. Barad (1979) reported that unrealistic client expectations are

associated with increases in burnout. Koocher (1979), in his descriptive study

of caregivers, found that the subjects rated "difficult patients" among the top

five correlates of burnout. In his study of new professionals, Cherniss (1980)

reported that the degree of client motivation and ability, the manipulativeness

of clients, and the role of clients as critics were all factors that could produce

either satisfaction or strain for the helpers.

Client characteristics, it appears, are related to burnout across at least

three dimensions: 1) characteristics of the contact with the client (Cherniss,

1980a); 2) type of problem (Barad, 1979; Koocher, 1979; Pines & Maslach, 1978);

and 3) number of clients (Maslach & Pines, 1977; Pines & Kafry, 1978; Pines &

Maslach, 1978).

Hypotheses. Hypotheses regarding the relationship between client

characteristics and burnout are:

Hla: The underlying dimensions of relevant client

characteristics are manageability, treatability,

and likeability.

Hlb: Positive evaluation of client characteristics is

negatively and significantly related to burnout.

Work Load

Concept and Definition. Work load has been conceptualized by Coburn

(1975) in terms of job-worker congruence. Coburn suggested that when demand

to perform exceeds capacity to perform, work overload exists, and when capacity

exceeds demand, work underload exists. The former creates stress; the latter

is accompanied by frustration. Beehr and Newman (1978) interpreted research

on role overload in terms of ability, indicating that the experience of role

overload might actually reflect on the ability of the employee to do the work

assigned in the allotted time period.

In a series of studies conducted through the University of Michigan

Institute for Social Research, the concept of work load has been refined within

the overall framework of role conflict. Sales (1970), defined role overload as a

condition in which "the focal person is faced with obligations which, when taken

as a set, require him to do more than he is able to do in the time available"

(p. 593). Later, the work of French, Tupper, and Mueller (1965) which had

differentiated between qualitative (having work that is too difficult or too easy)

and quantitative (having too little or too much to do) work load was replicated

in research at Goddard and Kennedy Space Centers, resulting in a general

agreement that work load refers to the amount and the difficulty of the set

of tasks assigned to an individual, relative to the time allotted for their


Empirical Evidence. Five studies of the association of work load and

strain, based on surveys of large samples, are reviewed. Margolis and Kroes

(1974) found that role overload is related to an increase in escapist drinking, and

a decrease in self esteem, motivation to work, and frequency of suggestions

made to employers. Cherry (1978) reported that work load added significantly

to the prediction of strain, even when level of functioning, which explained two-

thirds of the strain in the sample, was already entered in the model. Kohn and

Schooler (1973) found that a composite measure of job pressure predicted job

satisfaction but not organizational commitment. Time and dirtiness, two factors

of the composite measure, were negatively related to job satisfaction and

unrelated to commitment, while heaviness of work was unrelated to either strain.

House, Wells, Landerman, McMichael & Kaplan (1979) reported that even with

demographics and risk-related factors including smoking behavior, exposure to

chemicals, obesity, and physical activity controlled, work load was a significant

predictor of neuroticism, a strain measure including anxiety and depression.

Caplan et al. (1975) reported that subjective quantitative work load was correlated

with depression and irritation. A reanalysis of these data by La Rocco, House,

and French, Jr. (1980) revealed that there was also a work load by coworker

support interaction effect on worker dissatisfaction. In an analysis of Swedish

and American worker survey data, Karesek (1979) found that increased job demand

is positively related to increased exhaustion and depression. However, when

decision latitude is congruent with job demand, the result is task directed activity

instead of stress.

A number of other studies have used different methodologies to investigate

the work load-strain relationship. Frankenhaueser and Gardell (1976), who are

attempting to bring together field study and laboratory methodology in the study

of the effects of work place stressors on strain, asserted that work overload and

work underload result from advanced technology. For example, the machine-

paced job represents overload because of a push for production that requires

constant attention, and underload because the work is machine-paced, allowing

the worker no discretion. Their findings confirmed that machine-paced in contrast

with man-paced jobs led to increased production of catecholamines and increases

in self-reported strain.

Hay and Oken (1972), in their descriptive study of ICU's, observed that

increased work load led to increased stress for nurses. It was their perception

that the noxious stimuli present in the work place lowered the overload threshold

for ICU nurses. Jacobson (1978) suggested that adequate staffing permits nurses

to take other pressures in stride.

In a longitudinal study using college students as subjects, Caplan and

Jones (1975) found that changes in work load caused changes in anxiety and

tension, but not in heart rate. Sales (1970), in a laboratory study, found that

objectively measured overload caused increased errors, productivity and tension,

and decreased self-esteem. Subjective overload caused higher experienced time

pressure, greater feelings of tension and anger, and higher enjoyment of task.

The objective measures, as might be expected, appeared to be better predictors

of "hard data" outcomes while the subjective measures predicted self-report


The literature reviewed indicates that work load is positively related to

psychological, physiological and behavioral strains. Included among the

psychological strains were general strain (Cherry, 1978; Frankenhaueser & Gardell,

1976), neuroticism (House et al. 1979), exhaustion and depression (Karasek, 1979),

anxiety and tension (Caplan & Jones, 1975; Sales, 1970), and depression and

irritation (Caplan et al. 1975). Physiological strains included in the studies were

catecholamines (positively related, Frankenhaueser & Gardell, 1976) and heart

rate (unrelated, Caplan & Jones, 1975). Behavioral strains positively related to

increased work load were escapist drinking (Margolis and Kroes, 1974) and

productivity (Sales, 1970).

Only one empirical study relating work load specifically to burnout was

located, although speculation about the extent of the relationship abounds in the

conceptual literature. Berkeley Planning Associates (1977), in a correlational

study conducted among child abuse workers, found that control over work load

and burnout were inversely related.

Hypotheses relative to work load and burnout are:

H2a: Work overload is positively and significantly

related to burnout.

H2b: When client characteristics are positively

evaluated, work load is unrelated to burnout.

Social Support

Concept and Definition. In the social support literature "social

relationships are hypothesized to expand or contract the individual's capacity for

managing stress" (Karesek, Triantis & Chaudhry, 1982, p. 182). Social support,

characterized by Gore (1978) as a relatively stable characteristic of the

individual's interaction with his social environment, derives from a number of

sources, including coworkers, supervisors, peers, and family and friends.

The effort to delineate the dimensions underlying the social support

construct has produced a number of typologies. For example, Karasek (1979)

identified trusting social relations and instrumental support as primary dimensions

while Cobb (1976) has suggested that the dimensions are emotional, esteem and

network. House (1981) has proposed four dimensions that are conceptually

inclusive of all of the other typologies (Gaines, in process): emotional,

instrumental, informational and appraisal. La Rocco et al. (1980) asserted that

"although there are many potential types of social support the data in

most studies of social support largely tap emotional supportiveness" (p. 204).

In addition to the issues of sources and types of social support is the

question of the type of relationship that social support has to stress outcomes

of interest. Analyses of social support have treated the construct as both an

independent variable and as a moderating or conditioning variable. When treated

as an independent variable, social support is hypothesized to exert a direct or

main effect by functioning as a negative stressor or strain, and/or as a positive

health outcome. When treated as a moderating variable, social support is most

often hypothesized to exert a buffering influence on the relationships between

stress and outcomes. For example, high social support would be expected to

reduce the impact of a stressor on a strain (House, 1981). More recently, Seers,

McGee, Serrey, and Graen (1983), following a review of the mixed results of

studies testing the buffering hypothesis, suggested the coping hypothesis as an

alternative conceptualization of the moderating effect of social support on strain.

The coping hypothesis suggests that, in the presence of high levels of stress,

social support will increase levels of positive and decrease levels of negative

outcomes. Seers et al. (1983) tested the three hypotheses regarding the effect

of social support on stress: H1: Social support affects strain directly and

simultaneously with other independent variables; H2: Stress and social support

interact to affect strain, with stress negatively related to strain for individuals

with high support and unrelated to strain for those with low support; or H3:

Stress and social support interact to affect strain, with support negatively related

to strain for individuals in a high stress condition and unrelated to strain for

individuals in a low stress condition. H3 is the coping hypothesis. Seers et al.

(1983) found moderate support for the direct effects hypothesis and strong support

for the coping hypothesis. The buffering hypothesis was unsupported.

Although Cobb (1976) asserted that researchers should not expect to find

significant main effects of social support, most researchers agree with House's

assertion "that main effects are obvious to most people" (1981, p. 32).

Empirical evidence regarding main effects of social support is presented below.

There is, however, considerable disagreement regarding buffering effects in both

the conceptual and the empirical literature. One source of disagreement involves

the choice of statistical approach. Although moderating effects on a variable

are frequently analyzed by the addition of an interaction term to a moderated

regression model, the results of this approach are difficult to interpret. Because,

as Seers et al. (1983) pointed out, the term is symmetrical, it is difficult to

infer with any certainty which is the predictor and which is the moderator

variable. An alternative approach to testing for buffering effects involves

subgrouping based on social support scores and analyzing for differences between

subgroups. The evidence regarding buffering effects of social support is confusing.

Evidence supporting the buffering effect was found in studies conducted by Gore

(1974), House and Wells (1978), and La Rocco et al. (1980), while a notable study

which was among the first to test the buffering hypothesis (Pinneau, 1976)

revealed that the number of significant interactions between social support and

predictor was not more than the number of interactions that might be expected

to occur by chance. Similarly, Gavin and Axelrod (1977), in their study of 95

management level employees in an underground mine, found no support for the

buffering effect of social support on the stress-strain relationship.

Although social support has been incorporated into the study of stress as

a "legitimate" construct, and has generated considerable speculation regarding

the dimensionality and the form of its relationship to the stress experience, a

commonly accepted definition of the term has not yet been specified. Social

support has been defined as a feeling, as information, and as social relationship.

Moss (1973), who defined support as "the subjective feeling of belonging, of being

accepted, of being loved, of being needed all for oneself and not what one can

do" (p. 237), emphasized the fact that social support is not conditional on

acheivement and thus provides an alternative source of a sense of worth. Cobb's

(1976) definition of social support as "information leading the subject to believe

that he is cared for and loved, esteemed, and a member of a network of mutual

obligation" (p. 300) indicates that the essence of social support is external to

the subject (information) as opposed to internal to the subject (feeling). For

the purposes of this study social support refers to the belief or feeling that

assistance, acceptance, and esteem are available from significant others.

Empirical Evidence. A number of studies examine main effects of social

support on stressors, general measures of strain, and physiological strain. Caplan

et al. (1975), in their study of a stratified sample of 318 representatives of 23

occupations, found that each of their four support scales (supervisor, colleagues,

family and friends, and participation) was negatively related to stressors (role

ambiguity, underutilization of skills) and to psychological strains (dissatisfaction,

depression), but not physiological strains (heart rate, cholesterol, uric acid). This

finding is particularly intriguing because French (1974) reported significant

conditioning effects of social support on physiological but not on psychological

strains. Pinneau (1976) found that support, in general, was negatively correlated

with affective strains and support specifically from supervisors was negatively

correlated with some job stressors, including role conflict and ambiguity,

utilization of ability, complexity, and unwanted overtime. Pinneau (1976) found

an insufficient ratio of significant to insignificant interaction effects to infer a

buffering effect. Seers et al. (1983), in their study cited above of predominantly

female respondents (N = 104) who all had the same job title in a large federal

government agency, examined direct and interactive effects of four types of

social support (family and friends, coworkers, branch manager, and unit manager)

in a design that included role conflict and role ambiguity as stressors, and four

facets of job satisfaction as strains. Support measures correlated negatively and

significantly with stressors in three of eight relationships. Support measures

correlated significantly and positively with satisfaction measures in five of twenty

relationships tested.

Karasek et al. (1982), in an extension of their analysis of the survey data

of workers from Sweden and the United States, were concerned primarily with

the conditioning effect of social support. They interpreted correlations between

their independent variable, task strain composite, and their social support measures

as indicators of independence of the constructs examined. These correlations

between social support and stressors have been considered direct effects in other

studies. Although significance levels are not reported, given the large sample

sizes it appears that eight of the fourteen relationships examined are significant,

which could be interpreted as evidence supporting the existence of a main effect

of social support on stressors. The authors examined correlations between support

variables and strain measures as evidence of a direct effect of support on strain,

and concluded that, in general, the effects are strong and in the predicted


The Karasek et al. (1982) study provides support for a hypothesis of

direct effects of social support on stressors and strains. Limited support is also

provided by the other studies reviewed, despite the fact that many of them were

undertaken primarily as a test of the buffering hypothesis. The findings of

Caplan et al. (1975) that social support has a direct effect on psychological but

not physiological strains, and the findings reported by French (1974) that

interactions of stressor and social support affect physiological but not

psychological strains suggest one possible approach to sorting out the confusion

relative to results of social support research. Another approach is suggested by

the introduction by Caplan et al. (1975) of the question of substitutability of

one form of social support for another. It is possible that a general measure

of support, without specification of referent, would yield useful information.

Four studies were reviewed that examined the main effects of social

support on burnout. None of these authors reported testing for buffering effects.

In his descriptive study of 28 new professionals, Cherniss (1980) found that new

professionals value collegial support of an emotional, informational and/or

evaluative nature, but only five of the twenty-eight felt that they had received

such support. Cherniss (1980) concluded that this lack of support contributed to

the high incidence of burnout among the professionals.

In their study of child care workers, Maslach and Pines (1977) included

number and importance of staff meetings among the predictor set, with

dehumanization, a variable similar to the depersonalization dimension of burnout,

as a dependent variable. Interviews indicated that staff meetings were viewed,

among other things, as enabling "the staff to socialize informally and to give

each other support. ." (p. 109). Number of staff meetings was positively and

significantly correlated with positive evaluation of work conditions and with

worker relations. Pines and Kafry (1978), in their study of 129 social service

workers, 84% of whom were female, measured social support as social feedback,

work relations, and work sharing. The dependent variable in the study was

tedium, which is very similar in its conceptualization to the emotional exhaustion

dimension of burnout. Pines and Kafry (1978) found that each support measure

was negatively and significantly related to tedium. Maslach and Jackson (1982),

using a nursing sample, found that support and recognition from coworkers and

from supervisors was negatively and significantly correlated with emotional

exhaustion, and that being able to discuss work with friends and perceived support

from friends were positively associated with personal accomplishment.

Among these studies is evidence that social support has a direct effect

on strain. Cherniss (1980) offered qualitative evidence that lack of social support

contributes to the overall burnout syndrome; the findings of Maslach and Pines

(1977) support the hypothesis that social support reduces depersonalization; Pines

and Kafry (1978) demonstrated that social support reduces emotional exhaustion;

and Maslach and Jackson (1982) add evidence that social support reduces emotional

exhaustion and provides support for the hypothesized positive relationship between

social support and a sense of personal accomplishment. It is important to note

that in none of these studies was social support measured using the Institute for

Social Research items, as it has been in many studies relating social support to

other dependent variables. Hypotheses suggested by the social support literature

are as follows::

H3a: Social support is negatively and

significantly related to burnout.

H3b: When client characteristics are positively

evaluated, social support is unrelated to



Concept and Definition. Margolis and Kroes (1974) identified, as among

five types of job-related strain, those which are short term and occur in "close

temporal proximity" (p. 15) to specific job stressors. Among these short term

strains is anxiety. Anxiety, as other emotional reactions, may be enduring over

time, in which case it is recognized as a relatively enduring personality

characteristic termed "trait anxiety". Lazarus (1966) stated that "anxiety is the

inevitable accompaniment of being threatened" (p. 322). In response to the

contention of some psychologists that anxiety is an affective state with no

external referent, Lazarus (1966) insisted that "the emphasis on the internal

source of anxiety misses the point that there must also be some counterforce in

the environment or internalized from the environment to make an impulse

threatening" (p. 66). May (1983) and Spielberger, Lushene and McAdoo (1977)

labeled the counterforce in the environment (which in the stress research paradigm

used in this study is termed stressorr") stress, and concurred that the cognitive

evaluation of the objective reference as dangerous produced threat. Anxiety,

then, may be conceptualized as "how the individual relates to stress, accepts it"

(May, 1983, p. 142). The degree and duration of the affective response, which

is unpleasant and may interfere with adaptive activities, reflect the degree of

threat perceived.

For purpose of this study, state anxiety is defined as "subjective,

consciously perceived feelings of tension, apprehension and nervousness

accompanied by or associated with activation of the autonomic nervous system"

(Spielberger, 1975, p. 137).

Empirical Evidence. Although anxiety has been operationalized in a

variety of ways, a review of studies of anxiety in work organizations suggests

that job related stressors are evaluated as threatening. Several studies used a

measure developed by Cobb (1970) for use in the Institute for Social Research

stress research program. This approach to the operationalization of anxiety

reflects a conceptualization of the construct that is very similar to Spielberger's

(1975). For example, items used in the Caplan et al. (1975) adaptation of Cobb's

(1970) approach include "I feel nervous" and "I feel jittery." Using this measure,

Caplan and Jones (1975), who obtained three measures of responses of 73 male

computer users to the impending and actual shut-down of computer services,

found that change in work load and change in role ambiguity were significantly

related to change in anxiety of the subjects. Furthermore, anxiety was found

to covary with stress to a greater degree than either resentment or depression.

Caplan et al. (1975) also used the Cobb (1970) operationalization of anxiety in

their survey of employees from 23 occupations and found that work load was

positively and significantly related to anxiety. La Rocco et al. (1980), after

their reanalysis of the Caplan et al. (1975) data, reported that two additional

work stressors, role ambiguity and role conflict, as well as two other outcomes,

boredom and job dissatisfaction, are positively and significantly related to anxiety.

Using similar anxiety items, but reporting results based on inclusion of items on

depression and irritation also, Gavin and Axelrod (1977), in a study of 95 managers

from all levels, found that role conflict, role ambiguity and variation in work

load were positively and significantly related, and that job security, participation,

and underutilization of skills were negatively and significantly related to their

composite strain measure. House et al. (1979) also reported that, among the

blue collar tire and rubber factory employees surveyed in their study, a composite

measure termed neuroticism and including the anxiety items from the Cobb (1970)

operationalization was related to work load.

Several authors reported results based on anxiety scales that were included

in more extensive surveys of mental health. Billings and Moos (1982) surveyed

214 employed male and 115 employed female members of families randomly

selected from the San Francisco population. They reported that for men lack

of autonomy on the job, lack of clarity, and work pressure were associated with

increased anxiety, while low job clarity was the only predictor of anxiety for

women. Interestingly, involvement, work group cohesion, and supervisory support

were negatively related to anxiety for men, but these social support variables

were unrelated to anxiety for women.

Gentry, Foster and Froehling (1972), in a study of 34 nurses in ICU and

non-ICU units in a veteran's administration and a general medical center hospital,

found that anxiety was higher among the medical center ICU nurses than among

either the veteran's hospital ICU nurses or non-ICU nurses in the medical center.

They interpret their findings to mean that, although ICU nursing can be highly

stressful, as is commonly thought, it is not necessarily so, with the difference

resulting from such factors as the relative amount of help available and the

availability of continuing education opportunities. In a survey of 1054 employees

of Swedish industries reflecting a variety of technologies, Gardell (1976) found

that workers in jobs that were low in job discretion and skill requirements, and

in jobs with low pay, showed higher levels of anxiety than those whose work

offered more discretion and higher pay. In their study designed to explore the

relationships of dimensions of occupations to psychological functioning, Kohn and

Schooler (1973) found that closeness of supervision, time pressure, likelihood of

dramatic change, frequency of being held responsible for things outside one's

control, and risk of losing one's job or business were all related to anxiety.

Although the State-Trait Anxiety Inventory (STAI) instrument designed

by Spielberger, Gorsuch, and Lushene (1970) to measure anxiety has been widely-

used in psychological and educational research, its use appears much less prevalent

in organizational studies. Ganster, Mayes, Sime and Tharp (1982) used the STAI

to assess the effectiveness of stress management training for reducing levels of

anxiety experienced by their social work subjects. Results of the experimental

study indicated that stress management training can be effective in reducing

levels of anxiety, and that the change persists over time. Abdel-Halim (1982)

reported that role conflict and role ambiguity were positively related and leader

and work group support were negatively related to anxiety as measured by the

STAI for his sample of 89 male managers from a large equipment manufacturing


Several studies can be interpreted as reflecting the relationship of client

characteristics to anxiety. Gow and Williams (1977) reported that nurses in

agencies offering care to the chronically ill were more anxious than those in

community health agencies or in agencies offering care to the terminally ill.

Mayer and Rosenblatt (1974) found, in an anecdotal study of 39 social work

students, that client-worker relationships characterized by too much distance or

too much closeness led to increased anxiety, as did client hostility toward the

worker. Using an experimental design in which she examined the effects of sex

of patients, type of unit (medical or surgical), and time order of assignment

(first or second), Parkes (1982) reported that among student nurses anxiety was

not predicted by either type of nursing (medical or surgical) nor by sex of patient,

although anxiety was significantly related to a sex by time period interaction.

It appears safe to conclude that work-place stressors having to do with

role conflict, role ambiguity, and work load may be expected to affect the level

of anxiety experienced by members of an organization. The effects of

characteristics of the client and the client-worker relationship on anxiety are

less clear from the organizational literature. While the Gow and Williams (1977)

and Mayer and Rosenblatt (1974) studies seem to provide support for assuming

such a relationship, Parkes' (1982) study of student nurses does not. Also relevant

to the point are the conclusions of Spielberger et al. (1977), based on their

review of the psychological literature, that threat of physical danger and threat

to self-esteem are associated with increased levels of state anxiety. Both types

of threat appear to be a part of the client-helper relationship.

Although the lack of literature on the topic and the lack of agreement

in findings in the few studies that are available make them somewhat exploratory,

the following hypotheses are tested. Because burnout is a long-term, and anxiety

a short-term psychological strain, it is expected that the two appear successively

and not concurrently. It is also expected that anxiety will be related to the

same stressors expected to be related to burnout.

H4a: Anxiety and burnout are negatively and

significantly correlated.

H4b: Positive evaluation of client characteristics

is negatively and significantly related to


H4c: Manageability and treatability are better

predictors than likeability of anxiety.

Physical Symptoms

Conceptualization. A relationship between work-place stressors and

physiological outcomes is hypothesized in a number of the stress models (Beehr

& Newman, 1978; House, 1974; Ivancevich and Matteson, 1980; Schuler, 1980).

Levi (1972) stated that "we know that psychosocial stimuli cause physiological

changes, which in turn could lead to precursors and disease" (p. 19). As Ivancevich

and Matteson (1980) indicated, the suggestion is not that diseases of adaptation

to the physiological changes are "uniquely stress-induced or even primarily stress-

based" (p. 19), but that stressors may lead to negative health consequences.

Selye (1976) suggested that these consequences could include high blood pressure,

heart and blood vessel diseases, kidney disease, rheumatoid athritis, digestive and

metabolic disease, allergies, nervous and mental disease, and possibly cancer. It

is thought that stressors affect these strains by altering mental processes,

endocrine processes, lymphatic processes, and immunoreatice processes (Levi,

1972). The flight or fight response identified by Cannon (1929) is produced by

the release of hormones when a situation is appraised as threatening. When

unrestricted, this response triggers appropriate behavior "which eliminated the

cause or source of the stressful correlation. It is becoming less feasible for

individuals in organizations, however, to engage in either a flight or fight behavior,

resulting in dysfunctional physiological reactions" (Schuler, 1980, p. 201).

Although House (1974) suggested that some types of occupational stress

appear to relate specifically to particular diseases or clusters of diseases, Cassel

(1979) concluded that most evidence supports the viewpoint that particular

psychosocial events or conditions "will not be etiologically specific for any given

disease. The clinical manifestation of this enhanced susceptibility will not be a

function of the particular psychosocial stressor, but of the psychochemical or

microbiological disease agents harbored by the organism or to which the organism

is exposed" (p. 47).

Physical symptoms represent a manifestation of physiological changes

thought to be induced by threatening psychosocial conditions, including

occupational stressors. House et al. (1979), for example, suggested that variables

involving interpersonal tensions at work appear to affect the incidence of ulcers,

and Cassel (1976) suggested that "not receiving adequate evidence (feedback) that

. actions are leading to anticipated consequences" (p. 113) can constitute a

disease-enhancing social situation. Each of these work-place stressors is

potentially inherent in the client-helper relationship.

Empirical Evidence. A number of studies have found evidence supporting

a hypothesized relationship between organizational stessors and physiological

strains, although in some cases, the evidence is inconsistent. Zaleznick et al.

(1977) found general support for an occupational stressor-physiological strains

linkage. A comparison across occupations revealed that respondents from the

operations and staff groups, which were the higher stress occupations in the

organization surveyed, had a significantly higher incidence of physical disorders

than did respondents from the management group, which was a low stress

occupation. Moreover, the predominant disorders were different for the two

stressed groups. Operations personnel had a high prevalence of medication use

and allergy and respiratory problems, while staff personnel showed the highest

incidence of cardiovascular and gastro-intestinal disorders.

Several studies addressed the relationship between work-place stressors

and specific physiological measures. Sales (1970), found that objectively measured

work overload led to greater increases in heart rate than work underload. When

work load was measured subjectively, there was no significant difference in heart

rate change between overload and underload conditions. O'Brien, Smith,

Goldsmith, Fordham, & Fan (1979) observed the relationship of heart rate and

oxygen consumption to activity among assembly line workers and nurses. For

both groups, oxygen consumption increased with activity. However, although

increases in heart rate could be accounted for by the increased activity among

assembly workers, the same relationship was not found among nurses.

Frankenhaueser and Gardell (1976), in their study of sawmill workers, found that

increased catecholamine production was associated with machine-paced work that

was characterized by short work cycle and restricted posture.

Illustrative of studies exploring the relationship between social support

and physiological stress are those of Gore (1978) and Billings and Moos (1982).

Gore (1978), in her longitudinal study of the effects of a plant's closing, found

that cholesterol levels and physical symptoms were significantly higher among

subjects who did not receive social support, although these strains were not

significantly related to a sense of economic depression. Billings and Moos (1982)

found that, among the male subjects in their study, physical symptoms were

directly related to low autonomy, control, and a composite work stressor measure,

and were inversely related to involvement, cohesion, supervisor support, and a

composite resources (social support) measure. For females in the study, physical

symptoms were predicted only by low autonomy, with all social support variables

unrelated to physical symptoms. Researchers have also found physical symptoms

related to other forms of strain. Caplan and Jones (1975) found that anxiety

and heart rate were positively related, and Cherry (1978) found that

while only 16.6% of subjects with moderate work-related strain reported physical

symptoms, 49.1% of those with high work-related strain reported physical


Several studies relate work-place stressors to general measures of physical

disorder. Rousseau (1978a), in a study cited more fully in the discussion of task

characteristics, found that physical stress was positively related to role conflict,

and negatively related to autonomy, feedback, variety, and learning, but not to

task identity or task significance. Margolis and Kroes (1974), in their interviews

with 1496 workers over the age of 16, found that overall job stress, under-

utilization of skills, resource inadequacy, insecurity, and non-participation were

associated with decrements in physical health, but overload and ambiguity were

not. Similarly, Gavin and Axelrod (1977) found that among management personnel

in an underground mine, underutilization of skills and lack of job security were

related to an increase in psychosomatic complaints, although role conflict, role

ambiguity, and overload were not. Coburn (1975) found that, although moderately

complex jobs were related to improved health, jobs high or low in complexity

were related to poor health.

House et al. (1979) found that, among hourly, non-managerial workers in

a tire and rubber manufacturing company, 18 of 25 correlations between stressors

(responsibility pressure, role conflict, work load, quality concern, and job vs. non-

job conflict) and physical symptoms (angina, ulcers, neurosis, itch and rash, and

cough and phlegm) were significant, and as the authors stated, not trivially so.

Each symptom was predicted by at least two stressors, and each stressor predicted

at least three symptoms. House et al. (1979) concluded that "the pervasive

impact of stress across all self-reported health outcomes is consistent with the

hypothesis that stress increases the susceptibility of blue collar workers to a

wide range of health problems and diseases" (p. 147). The authors further note

that there are some specificities among the stress-strain relationships, in that

only certain stressors affect ulcers and angina: ulcers seem to be related to

stressors that suggest interpersonal conflict and angina to stressors that indicate

task orientation.

Thus, there is convincing evidence that work-place stressors are related

to physiological conditions, although the evidence on specific stressor-strain

relationships is somewhat confusing. For example, work load was a predictor of

physical symptoms in the Sales (1970) and House et al. (1979) studies, but not in

the Gavin and Axelrod (1977) or Margolis and Kroes (1974) studies. Similarly,

the effect of social support on physical symptoms is confusing (Billings & Moos,

1982; Gore 1978).

No studies were found that dealt specifically with the effect of client

characteristics on physical symptoms. However, if client characteristics are

conceptualized as interpersonal events (Jermier, 1982), and assuming some validity

in the conclusion of House (1981) that interpersonal tensions affect health, and

of Cassel (1976) that feedback is related to disease susceptibility, it seems

reasonable to expect a relationship between client characteristics and physical

symptoms of helpers.

H5a: Positive evaluation of client characteristics

is negatively and significantly related to

self-reported physical symptoms.

H5b: Client likeability is negatively and sig-

nificantly related to self-reported

gastrointestinal problems.

H5c: Client treatability is negatively and significantly

related to self-reported cardiac symptoms.


Conceptualization. The inclusion of absenteeism as a behavioral strain in

a number of the stress models (Beehr & Newman, 1978; Brief et al. 1981;

Ivancevich & Matteson, 1980; Schuler, 1980) with virtually no conceptual or

empirical rationale given for its inclusion reflects the assumption that seems to

underlie the absence literature: that absenteeism is a "hard" criterion variable

(Johns, 1978) or "clear-cut act" (Porter & Steers, 1973), with obvious definition

and importance. Muchinsky (1977), however, contended "that 'absenteeism' is a

particularly ambiguous concept and this ambiguity has clouded the exact meaning

of many studies that investigated the relationship between absenteeism and other

variables" (p. 317). Critical to clarification of the ambiguity is specification of

the motivation for studying absenteeism. Typically, researchers and reviewers

have seemed to be interested in absenteeism because it has been assumed to be

costly to organizations (Steers & Rhodes, 1978), implying a need for managerial

control and leading to inclusion of suggestions of absence reduction in research

reports and reviews (Muchinsky, 1977; Nicholson & Goodge, 1976). An alternative

approach is to consider absenteeism a quality of work life issue, "which requires

that the gap between behavioral scientists' accounts of absence as a social

problem and its experiential reality to the worker be closed" (Johns &

Nicholson, 1982, p. 128).

Although absence has been conceptualized as withdrawal (Beehr & Gupta,

1978; Muchinsky, 1977) or flight (Ivancevich & Matteson, 1980), it may also be

true that absence from work sometimes represents a positive choice to be

elsewhere; for example, to fulfill family obligations (Johns & Nicholson, 1982)

or pursue leisure activities. Absence in response to stress may represent

withdrawal because the extent of the stress has become unbearable or coping

behavior designed to minimize current costs of stress and increase the worker's

future capacity for dealing with stress (Maslach & Jackson, 1982).

Despite the fact that previous research has identified a number of

correlates of absence, including both organizational and personal characteristics,

the amount of explained variance has been relatively small and unstable (Johns

& Nicholson, 1982) and the classes of correlates, although conceptually

independent, have been statistically redundant (Johns, 1978), suggesting that

important correlates of absence may have been overlooked. Exploration of the

meaning of absence from work to the stressed and absent worker may help

identify additional correlates.

Empirical Evidence. Most of the studies that include absenteeism as a

dependent variable examine how it is related to organizationally related factors,

such as task or role characteristics or features of the work context; to personal

characteristics; or to some combination or interaction of organization and person.

Several of the studies reviewed examined the effects of organizational stressors

on absence.

Ivancevich (1974), in his experimental study of the effect of a change

to four-day work week on absence among operating and managerial personnel,

found no relationship between the change and patterns of absence. Parkes (1982),

in the experimental study cited above of the effect of patient type, patient sex

and period of assignment on attitudes and behavior of 164 student nurses, found

that none of the independent variables predicted absence behavior. Although

these relatively objective characteristics of the work context did not predict

absenteeism, Hackman and Oldham (1976) reported, as had Hackman and Lawler

(1971), that the subjective perception of task characteristics was negatively and

significantly related to absenteeism for 658 subjects from 62 occupations.

Spencer and Steers (1980), in a study of 200 hospital clerical and service

workers, compared the influence of work experiences and personal characteristics

on absenteeism, which was measured as the total number of days absent in a

nine-month period. Because work experiences (group attitudes, met expectations,

job challenge, personal importance, and organizational dependability) were highly

correlated (median r=.57), partial correlation coefficients were used in the

analysis. Results indicated that personal characteristics related significantly to

absenteeism were tenure in the organization (directly), tenure in position and

age (inversely), and sex (females absent less often than males), and that the only

work environment factor related to absenteeism was job challenge (negatively).

The model explained 18% of the variance in absenteeism. Mowday and Spencer

(1981) found that a single motivating potential score measure of task

characteristics was negatively related to absenteeism, measured as total number

of incidents of absence in a one-year period. The task measure interacted with

a personal measure to influence absenteeism such that high need for autonomy

and high job scope were associated with decreased absence, and low need for

achievement and high job scope were associated with increased absence.

Watson (1981) compared the effects of job satisfaction, job situation and

personal characteristics on absence, and found that the model explaining the

most variance included job situation and personal characteristics. Watson (1981)

also found that the full model explained 47% of the variance for females but

only 18% for males, with marital status and tenure as the significant predictors.

Nicholson and Goodge (1976) found that various types of absence (sickness absence,

unsanctioned absence, holiday absence, and casual absence) were differentially

predicted among the 343 female workers in their sample. For example, workers

in more enriched jobs had less casual and unsanctioned absence, but had no

significant difference in sickness absences from those in unenriched jobs. In

addition, they found that younger workers and those with longer tenure have

higher levels of casual holiday and unsanctioned absence than older workers or

those with less tenure, respectively. Johns (1978) tested the effects of four sets

of predictors (job satisfaction, personal characteristics of workers, leadership

style, and job content) on the frequency of absence incidents among 208 workers

(24% female) in a manufacturing organization. Using stepwise regression, he

found that age, sex, task identity, consideration, and feedback were significant

predictors, explaining 9.4% of the variance, but further analysis revealed that

the identity effect was an artifact of sex. Johns concluded that there is a great

deal of redundancy in the effects of conceptually independent predictors on


The review of studies that explored both work-related factors and personal

characteristics as possibles correlates of absence behavior reveals that

subjectively-measured characteristics of the task itself, and age, tenure, and sex

of worker, consistently contribute to prediction of absence behavior.

Several studies examined the effects of known stressors on absence.

Karasek et al. (1982) found that, among Swedish workers, task strain (measured

as the job demand-job decision latitude relationship) was positively and

significantly related to absence. Coburn (1975) found that individuals in conditions

of role overload had the highest absence rates in his sample of 1143 males.

Margolis and Kroes (1974) reported that underutilization of skills, role overload,

resource inadequacy, and non-participation, but not role ambiguity, were positively

and significantly related to absence.

Three studies related burnout to absenteeism. Cherniss (1980) found

that increased burnout was associated with increased absenteeism among new

professionals. Maslach and Jackson (1981), in their study of the measurement

of burnout, found that absenteeism was positively related to depersonalization.

Maslach (1982) reported that "absenteeism is more prevalent with burnout, and

some staff people routinely take their maximum sick leave" (p. 80).

Although the studies cited above seem to indicate that there is a

relationship between some stressors and absence, the relationship between client

characteristics and absence is expected to be complex. The Parkes (1982) study

found no relationship between patient sex and absenteeism of nurses. Hay and

Oken (1972) stated that "group loyalty reinforces work pressure in stimulating

guilt about any absence" (p. 127). Maslach and Jackson (1982) suggested that

one way of coping with the emotional exhaustion and depersonalization of burnout

(and so, presumably, with the stresses that produce them), is by getting away,

with absence being one way of withdrawing.

H6a: Positive evaluation of client characteristics

is negatively and significantly related to

absence frequency.

H6b: Manageability and likeability are better

predictors than treatability of absence


Intervening Variables

Lazarus (1966) has called upon stress researchers to identify and analyze

the mechanisms and conditions that link antecedents of stress and stress reactions.

While a number of the stress models mentioned above specify conditioning or

moderating variables, specification of intervening variables that may explain

and/or elaborate the relationships of stressor to strain is seldom observed in

either the conceptual or theoretical literature. Inclusion of intervening variables

in research models permits identification of both direct and indirect effects of

stressor on strain. For these reasons, the study specifies two constructs through

which client characteristics are expected to influence burnout: technology and

task characteristics.

Gillespie and Mileti (1977) observed that studies of technology typically

take one of two approaches: detailing the effect of technology on the attitudes

or behavior of individuals, or examining its effect on organizational structure.

They further suggested that the technology construct may reasonably be considered

a dependent variable, subject to the nature of demands made by some components

of the organization's environment, and that "the conditions that define its status

in lines of causality need to be spelled out and empirically validated" (p. 14).

Figure 2-4 suggests that technology is a dependent variable relative to

client characteristics. Clients are the raw material of people-changing

organizations. Perrow (1965) stated that technology is influenced by the cultural

definitions of this human material that hospitals attempt to alter. Lefton and

Rosengren (1974) observed that many client characteristics considered relevant

could have an impact on the functioning of organizations and the "organization

must select and define those client characteristics which are salient for their

purposes" (p. 474). Similarly, Hasenfeld (1974) has concluded that client

characteristics are critical to the organization's definition of its technology.

While inclusion of technology in the model as a dependent variable relative to

client characteristics is a departure from the usual approach to technology

research, its inclusion as an independent variable relative to individual outcomes

is more consistent with a typical treatment of the technology construct (Gilespie

& Mileti, 1977). Frankenhaueser and Gardell's (1976) statement that thereee is

a growing recognition of the risk that maladjustment--as manifested in alienation,

emotional disturbances and psychosomatic disorders--may develop as side effects

of advancing technology" (p. 35) provides support for the expectation of a

relationship between technology and burnout. Perrow (1965) observed that,

because treatment and diagnostic facilities are now key resources of hospitals,

they must be controlled by those who use them (the doctors). It appears

reasonable that this manifestation of technology constitutes a stressor for nurses,

who retain little control.

In addition to the direct effect on burnout, and indirect effects through

technology, client characteristics may have indirect effects on burnout as a result

of their effect on characteristics of tasks performed by individuals. Because

technology and task characteristics are sometimes considered different labels for

the same construct (for example, Miles, 1980a), specification of a task

characteristics-burnout relationship requires delineation of the conceptual and

empirical differences between the technology and task characteristics constructs,

in addition to explicating the rationale for expecting that client characteristics

are related to individual outcomes.

Although Pierce and Dunham (1978) demonstrated that the

operationalization of technology and task characteristics does not discriminate

between the two, they asserted that "organization structure, technology, and job

design represent three distinct constructs" (p. 410). Technology defines the kind

or nature of the work of an organization or unit, while job design refers to more

micro-level features of the work, described by Van de Ven and Ferry (1980) as

the "analytical properties of individuals' jobs or positions" (p. 206). A difference

between the two constructs is implied by Griffin's (1982) assertion that technology

can either constrain or facilitate job redesign. Empirical evidence suggests that

task characteristics sometimes vary within technologies (for example, Abdel-

Halim, 1981) and sometimes appear to differentiate among technologies (for

example, Rousseau, 1977).

The relationship between job characteristics and individual outcomes has

been often and explicitly treated in the organizational behavior literature. Job

characteristics are viewed as important in the determination of quality of work

life for organizational members (Griffin, 1981; Hackman & Suttles, 1977) and

have been shown to be related to affective and behavioral outcomes (Emery &

Trist, 1960; Hackman & Lawler, 1971; see Griffin, 1982, for review). More

specifically relevant to the relationship of task characteristics to burnout is the

work of Jermier (Gaines & Jermier, 1983; Jermier et al., 1983) which

conceptualized physical danger, a manifestation of the client worker relationship,

in interaction with departmental context, to emotional exhausion. Rousseau

(1978b) interpreted the results of her study of the effect of departmental,

positional, and individual characteristics on individual attitudes and behavior as

evidences that "perceived job characteristics are critical links of organization-

related characteristics to individual response" (p. 535).

It appears, then, that client characteristics help define the context within

which work is done, and thus affect workers' perceptions of their tasks, which

in turn affect affective outcomes for the individual. In fact, Johns and Nicholson

(1982) concluded that "subjective perceptions of job dimensions and work attitudes

should be restored to their logical research roles as individual-level intervening

variables, lying between the structure of work environments and worker behavior,

rather than treated as intrapsychic independent variables without reference to

their situational antecedents" (p. 131).


Concept and Definition. Woodward (1965) conceptualized technology in

terms of the type of production process used by the organization (small batch

or unit, mass, or process production), while Thompson (1967) based his typology

on "the manner in which individuals or work groups are arranged" (Griffin, 1982,

p. 107). The three technological types identified by Thompson (1967) are long

linked, mediating, and intensive. The Aston group (Hickson, Pugh & Pheysey,

1969), basing their typology on workflow integration, identified operations

technology, materials technology, and knowledge technology as the categories

useful for classifying the technology of an organization. While the Woodward

(1965) and Thompson (1967) models suggest discrete and exclusive categories, the

Aston group (1969) model suggests that organizations can be classified at any

point along a continuum that represents the degree to which the dimension is

descriptive of an organization's technology. Gillespie and Mileti (1977) have

combined several of these dimensions in their definition of technology as the

"types and patterns of activity, equipment and material, and knowledge or

expertise, used to perform tasks" (p. 8).

Perrow's (1965, 1967) suggestion that technology could most usefully be

conceptualized with reference to the raw material used by the organization or

unit has provided an approach to the conceptualization of technology that has

general applicability in organizational settings, and that has impacted much of

the thinking and research on technology. Glisson (1978) observed that i n

human service organizations, which attempt to produce cognitive, affective or

behavioral changes in clients, the state of knowledge is determined by the extent

to which the raw material (human beings) is understood and the extent to which

outcomes applied to change them can be predicted is extremely limited" (p. 383).

Overton, Schneck and Hazlett (1977) found this approach to the conceptualization

of technology useful for thinking about the technology of nursing, and concluded

that "techniques for transforming the raw materials, that is changing the health

states of patients, are the specific actions taken by nurses" (p. 205). It should

be noted that "equipment is not included in the definition; equipment is a tool

of technology, but technology rests on knowledge of the raw material" (Perrow,

1965, p. 916). The amount and sophistication of technological equipment necessary

to implement the technology are sometimes used as a surrogate measure of

technology (Leatt & Schneck, 1981).

For purposes of this study, technology is defined as "a technique or

complex of techniques employed to alter 'materials' (human or nonhuman, mental

or physical) in an anticipated manner" (Perrow, 1965, p. 915).

Technology and Individual Outcomes: Empirical Evidence. Several studies

were discovered that address the question of the relationship of technology to

affective outcomes, or to stress specifically. Abdel-Hamin (1981), in a study

which used Thompson's (1967) categories to measure technology, found a main

effect of organization technology on satisfaction in addition to the effects of

role conflict, role ambiguity, and role overload, respectively. In this sample of

managers and staff personnel from a manufacturing organization (long linked

technology) and managers and non-managers from a bank (mediating technology),

subjects working in the bank experienced higher levels of satisfaction than those

working in the manufacturing organization.

Two studies assessed the relationship of technology to mental health.

Gardell (1976) reported that among 640 mass production workers and 414 process

production workers in Sweden, individuals doing machine-controlled, repetitive

work had lower mental health than craftsmen or individuals operating more

complicated mechanical systems. Kornhauser (1965), in his study of 655 male

factory workers, found that mental health was poorer for subjects doing machine-

paced work than for those doing repetitive work, and poorer for those doing

repetitive work than for those in skilled jobs. Mental health was operationalized

by Kornhauser (1965) as the sum of scores on indices of anxiety, self-esteem,

hostility, sociability, life satisfaction, and personal morale.

In a study of 271 (65% female) employees, representing a total of 19

departments of an electronics firm and a radio firm, Rousseau (1978a) found

that technology, operationalized as mechanization, was negatively and significantly

related to psychological and physical stress and to absence. However, when

technology was measured using the Thompson (1967) typology, mediating

technology was negatively and significantly related only to absence and

psychological stress, while long-linked technology was positively and significantly

related only to absence.

Several studies have dealt with the effect of technology on individual

outcomes among nurses. Sheridan and Vredenburgh (1978) found that technology

(labeled task structure in their study and measured using Lynch's 1974 scale)

was not significantly related to job tension among 216 nurses in a metropolitan

hospital. Gentry et al. (1972) administered standardized self-report instruments

to 34 nurses, and compared results based on ICU/non-ICU assignments. ICU

nurses, compared to non-ICU nurses, reported greater depression, hostility, and

anxiety. In their descriptive study of ICU nursing, Hay and Oken (1972) reported

that ICU nurses found that the need to function in a situation that was both

routine (due to the need for repetitive monitoring and data collection) and

potentially urgent (due to the high probability of life-threatening incidents)

contributed to the stressfulness of the work. Parkes (1982) found that type of

unit (which may be considered a surrogate for technology) was a significant

predictor of depression and work satisfaction among her sample of 164 student

nurses. Her results suggest that nursing on medical wards compared to surgical

wards produces more depression and less satisfaction for student nurses.

Leatt and Schneck (1980), in a study using data from 153 head nurses,

from nine specialties and 25 hospitals, found that, while sources of stress were

the same across specialties, the frequency with which certain types of stress

were experienced varied across specialties. Nurses working in medical, ICU, and

auxilliary units reported the highest levels of stress associated with patient-based

stressors. Similarly, Gray-Toft and Anderson (1981), in their study of 122 nurses

on five patient care units, found that, while sources of stress were the same

among nurses, the amount of stress experienced differed among units, with those

on the medical unit experiencing the highest levels of stress.

Apparently only Maslach and Pines (1977) have examined the relationship

of a technology-type variable to burnout. In their study of day care workers,

they found that, although a less structured program was related to perceptions

of better working conditions and more feelings of control over work, it was also

related to increased experience of emotional exhaustion.

It appears, based on the studies reviewed, that more routine technology

is generally associated with lower levels of stress among nurses. Such a conclusion

is speculative, since in most studies technology is not measured directly but

inferred from the surrogate measure of nursing subunit. The definition by Leatt

and Schneck (1980) of a nursing subunit as "a geographic inpatient area of a

hospital having a technological domain, an assigned number of beds, its own

regular complement of nursing staff with shared goals, and a formal hierarchical

structure" (p. 35) seems to support such an inference. In addition, Overton et

al. (1977), Filley, House and Kerr (1976) and Leatt and Schneck (1981) found

that their technology scale differentiated among subunits in their sample of 154

subunits in 24 hospitals. This relationship is tested directly in the study.

H7a: Client treatability is positively and

significantly related to routineness

of technology.

H7b: Routineness of technology is negatively

and significantly related to burnout.

Task Characteristics

Concept and Definition. Historically, conceptualization of task design

seems to have reflected the broader context within which work was accomplished.

This interpretation is captured in the Filley et al. (1976) classification of task

design approaches on dimensions of degree of specialization and historical period.

Prior to the industrialization of work, tasks were generally of a craft nature,

with little specialization. By early in the twentieth century, with the emphasis

on separation of management and labor and imposition of efficiency as the

primary criterion for work activity, most of the crafts-type of jobs, which had

become increasingly specialized, gave way to jobs designed according to the

principles of scientific management. During the period from the 1940's to the

1960's, largely in response to studies (Guest, 1957; Roethlisberger & Dickson,

1939) purporting to demonstrate that factors other than mechanical efficiency

in the design of the task affected productivity, efforts were made to enlarge

and rotate jobs. At about the same time, the Tavistock Institute group (Trist

& Bamforth, 1951) recognized the importance to productivity and affective

outcomes of the social arrangements of jobs.

More currently, with the addition of concern for quality of work life to

concerns for productivity, increasing effort has been directed toward understanding

the meaning of work to workers (for example, Work in America, 1973) and to

exploring means that will enable workers to experience meaningfulness from

work. Current issues relative to job design include consideration of whether the

critical variable is represented by the objective characteristics of the job (Schwab

& Cummings, 1976) or by the perceived job characteristics (Hackman & Oldham,

1976) and the means by which perceptions of task are developed (O'Reilly,

Parlette, & Bloom, 1980; Salancik & Pfeffer, 1978). The early work of Hackman

(Hackman & Lawler, 1971) and the more recent study by Jermier et al. (1983)

indicate a considerable degree of congruence between objective and perceived

task characteristics.

Assuming with Rousseau (1977) t hat there exists a set of job dimensions

which affect the level of employee performance and attitudes toward work" and

that it is the employee's perception of these dimensions that affects burnout,

the job characteristics theory of job design developed by Hackman and Oldham

(1975; 1976) is utilized. Hackman and Oldham suggested, and demonstrated

empirically, that the characteristics of the work affect the intervening motivating

states of experienced meaningfulness of work, experienced responsibility for

outcomes of the work, and knowledge of the actual results of the activities. In

turn, these psychological states favorably affect personal and work outcomes

including internal work motivation, quality of work performance, satisfaction with

the work, and absenteeism and turnover. Task variety, task significance, task

identity, autonomy, and feedback have been identified through a series of empirical

studies (Hackman & Lawler, 1971; Hackman & Oldham 1975; 1976; Turner &

Lawrence, 1965) as dimensions of task design, although the stability of the

dimensionality of task characteristics is questionable (Dunham, Aldag, & Brief,

1977). Aldag et al. (1981) have suggested that there may be additional sample-

specific or general dimensions useful for defining the characteristics of tasks.

Pioneering work in the study of burnout (Cherniss, 1980; Freudenberger, 1975;

Maslach & Pines, 1977) suggests that this may be so among helping professionals.

In particular, Maslach's (1978) discussion of the role of the client in burnout

suggests that client characteristics may represent such a dimension, and/or may

affect workers' perceptions of task characteristics relative to other dimensions.

Task characteristics refer to workers' evaluative perceptions of activities

and interactions that comprise their jobs.

Task Characteristics and Individual Outcomes: Empirical Evidence. Pierce

and Dunham (1976), in their review of the job characteristics research, concluded

that there is considerable evidence across studies to support the hypothesized

job characteristics-satisfaction relationship. Results of several recent studies

provide additional support for their conclusion. Hackman and Oldham (1976), in

a study of 658 respondents from 62 occupations, found that all task characteristics

were positively and significantly associated with internal motivation, general

satisfaction and growth satisfaction. Abdel-Halim (1981) compared the effects

of role and task (measured with the Job Diagnostic Survey) on satisfaction in

an organization with mediating technology (a bank) and an organization with

long-linked technology (a production firm). Satisfaction was regressed on role

conflict and job complexity, role ambiguity and job complexity, and role overload

and job complexity. In each case, job complexity had a main effect on satisfaction.

In addition, there was a three-way interaction of role, task, and technology such

that in long-linked organizations high role ambiguity and low task complexity

were related to a lower degree of satisfaction and in mediating technologies

high role ambiguity and high job compleixty were related to a lower degree of

satisfaction. Griffin (1981) reported that in a longitudinal study of a non-

unionized manufacturing plant, job characteristics were positively and significantly

related to satisfaction both at the time of initial data collection and in the

follow-up study conducted three months later. A number of studies have explored

the relationship of task characteristics to other affective outcomes. French et

al. (1975) found that job complexity, assessed in terms of Hackman and Oldham's

conceptualization, was negatively and significantly related to boredom. Aldag

and Brief (1975) found, in a sample of 104 employees of a division of correction,

that 33 of 40 correlations between core job dimensions and affective outcomes

were positive and significant. Outcomes measured included satisfaction, work

involvement and internal motivation. In a study of employees whose work was

directly related to the production process in 19 units of 13 organizations, Rousseau

(1977) found that all task characteristics were negatively and significantly related

to alienation, and positively and significantly related to involvement and

satisfaction. Karasek (1979) found that decision latitude, which is similar to the

autonomy construct of the job characteristics theory, was negatively and

significantly related to depression and exhaustion in a survey of Swedish workers.

In an extension of the task design approach, Karasek (1979) also found an

interactive effect of job decision latitude and job demand on depression. Only

one study was located that examines the relationship of task characteristics to

burnout. Jermier et al. (1983) found, in a sample of 169 employees of a police

department, that a composite measure of task characteristics was negatively and

significantly related to emotional exhaustion, one dimension of the burnout

construct. Pines and Kafry (1978) found that task variety, task significance, and

autonomy were not significantly related to tedium among social workers, but

that feedback was negatively and significantly related to tedium, a construct

similar to burnout, among 129 (predominantly female) social workers. Caseload

was related positively and significantly related to tedium.

It appears, then, that the relationship of perceived task

characteristics to affective outcomes has empirical support, although only one

of the studies cited above deals directly with burnout as a dependent variable.

Several studies appear to offer support for the relationships among variables

suggested by this study. Figure 2.4 indicates that a stressor (client characteristics)

affects a strain (burnout) both directly and through an intervening variable, task

characteristics. When Beehr (1976), in a study designed to explore situational

moderators of the role stress-role strain relationship, collected data from 651

workers (51% male) employed in five midwestern work organizations, he found

that autonomy was correlated significantly and negatively with role ambiguity

(a stressor) and significantly and negatively correlated with job dissatisfaction,

life dissatisfaction, low self-esteem, and depressed mood (strains). In addition,

autonomy was the only one of the three situational variables tested that moderated

the role ambiguity-strain relationships.

Jermier et al. (1983) found that among the 171 employees of a

police department, perceived physical danger, which may be conceptualized as a

stressor and/or as a task characteristic, predicts emotional exhaustion, and,

inversely, commitment and pay satisfaction (strains). Although physical danger

was not significantly related to motivating potential score (a composite measure

of task characteristics), the positive correlation approached significance. Task

motivating potential predicted intrinsic satisfaction, pay satisfaction,

commitment, and, inversely, emotional exhaustion. Rousseau (1978a) tested a

model similar to that proposed for this study. In a study of 271 workers from

19 departments in two organizations, she investigated the direct effects of

department characteristics on affective and behavioral outcomes, and examined

task characteristics as a possible mediating variable through which the independent

variables would have indirect effects on the outcomes. She found that measures

of structure were positively related to psychological stress and that measures of

technological mechanization were negatively related to stress. Autonomy and

feedback from agents and from the job were negatively and significantly related

to psychological stress. In addition, 62% of the common variance between

departmental characteristics and individual affective and behavioral responses

was accounted for by task characteristics, indicating support for the mediating

role of that variable. Maslach and Jackson (1981) found that among 605 people

from a variety of health and service occupations (44 percent female), feedback

was significantly correlated with all dimensions of burnout: inversely with

emotional exhaustion and depersonalization, and positively with personal

accomplishment. Maslach and Jackson (1982) found knowledge of results and

feedback inversely related to emotional exhaustion and depersonalization among


The Pines and Kafry (1978), Jermier et al. (1983), Rousseau (1978a),

Maslach (1981), and Maslach and Jackson (1982) studies appear to provide support

for expecting indirect effects of client characteristics on burnout through the

intervening variable, task characteristics.

H8a: Client manageability is positively

and significantly related to task


H8b: Client treatability is positively and

significantly related to task charac-


H8c: Task characteristics are negatively

and significantly related to burnout.



Research Questions and Hypotheses

The study has systematically explored the relationship of a particular job

stressor, client characteristics, to a specific strain, burnout, by examining the

direct relationship of client characteristics to burnout, the relative effects of

client characteristics and other stressors (workload and social support) on burnout,

the differential predictive value of client characteristics for burnout and other

strains (anxiety, physical symptoms, absence), and the indirect effects of client

characteristics on burnout through intervening variables (technology and task


Research questions specified were:

1. What is the relationship of client characteristics

to burnout?

2. What are the comparative effects of client

characteristics and other stressors on burnout?

3. Do client characteristics predict burnout and

other strains differentially?

4. What is the mechanism through which client

characteristics affect burnout?

Daft (1983) emphasized the importance of recognizing that, if research

results are to contribute to new knowledge, uncertainty must be built into research

designs. He suggested that "one should start with incomplete facts, with

ambiguity, and plan experiments on the basis of probability, even a bare hunch,

rather than certainty. Then look for surprise" (p. 540). Because there has been

little attention to the effects of client characteristics on service deliverers,

there was considerable uncertainty relative to the answers to the research

questions posed above, and thus the potential for a great deal of surprise.

However, on the basis of the review of relevant literature and some bare hunches,

the following hypotheses were proposed.

Hla: The underlying dimensions of relevant

client characteristics are manageability,

treatability, and likeability.

Hlb: Positive evaluation of client characteristics

is negatively and significantly related to burnout.

H2a: Work overload is positively and significantly

related to burnout.

H2b: When client characteristics are positively evaluated,

work load is unrelated to burnout.

H3a: Social support is negatively and significantly related

to burnout.

H3b: When client characteristics are positively

evaluated, social support is unrelated to burnout.

H4a: Anxiety and burnout are negatively and

significantly correlated.

H4b: Positive evaluation of client characteristics is

negatively and significantly related to anxiety.

H4c: Client manageability and client treatability are

better predictors than likeability of anxiety.

H5a: Positive evaluation of client characteristics

is negatively and significantly related to

self-reported physical symptoms.

HSb: Client likeability is negatively and significantly

related to gastrointestinal symptoms.

H5c: Client treatability is negatively and significantly

related to self-reported cardiac symptoms.

H6a: Positive evaluation of client characteristics is

negatively and significantly related to absence


H6b: Client manageability and client likeability are better

predictors than treatability of absence frequency.

H7a: Client treatability is positively and significantly

related to routineness of technology.

H7b: Routineness of technology is negatively and

significantly related to burnout.

H8a: Client manageability is positively and

significantly related to task characteristics.

H8b: Client treatability is positively and significantly

related to task characteristics.

H8c: Task characteristics are negatively and significantly

related to burnout.

Research Design

Research Strategy

The study was designed to respond to methodological issues identified

above, as well as to substantive issues in stress research. Because the substantive

issue of central concern was the prediction and occurrence of burnout among

helping professionals, particularly in relation to characteristics of their clients,

it was considered important to investigate the phenomenon in relation to particular

work situations. The hypotheses were tested using a cross-sectional field study.

Although the cross-sectional design does not permit definitive conclusions

regarding cause-effect relationships, previous studies (i.e. Gore, 1978; Karasek,

1979; Parkes, 1982) using longitudinal designs confirm the suggestion that job

dimensions lead to stress among workers.

In further response to methodological concerns regarding stress research,

variables included in the analyses have been clearly identified as representative

of various components of the stress model. Anticipated relationships were

specified in advance and staged analysis was used to separate direct and indirect

effects of predictor variables on outcomes. In addition, multiple regression

analyses were used to study simultaneously the effects of several factors on

dependent variables of interest. This analytic technique also permitted

statistically controlling for factors which might confound results but which cannot

be controlled for by randomization in a field-study design.

Independent Variables

Independent variables used as control variables in analysis of data included

age, tenure in the organization, tenure on the unit, education of subject and

unit. Variables investigated as predictors of strain were client characteristics,

work load, and social support.

Dependent Variables

Dependent variables of interest to the study included burnout, anxiety,

absence, and physical symptoms.

Intervening Variables

Intervening variables explored in the study were unit-level technology and

task characteristics.

The Sample

Determination of the staff size needed to produce a sample size which

would permit the desired analyses required examination of the number of subjects

required for the proposed analyses and an estimate of the response rate that

might be expected.

Kerlinger and Pedhazur (1973) have suggested that [a] ny multiple

regression analysis, and especially those with many independent variables, should

have at least 100 subjects. ." (p. 446). A commonly used rule-of-thumb regarding

the number of cases necessary for multiple regression analysis, the primary

statistical technique used in the study, is that for each independent variable in

a study there should be ten subjects. In order to meet this rule-of-thumb

requirement, the present study required 80 or more subjects. Cohen and Cohen

(1975) suggested that determination of an appropriate number for a study requires

making judgments regarding the relative costs and benefits of being able to

determine the presence of effects of a designated magnitude. More specifically,

they suggested that, in a study involving five to fifteen independent variables,

researchers establish a sample size which would enable detection of an effect

of "medium" magnitude, or R = .30, and that the desired power of a test be set

at .80. Application of these conventions suggested by Cohen and Cohen (1975)

indicated that the present study required a sample size of 109.

Babbie (1973) reported that there is great variety of response rates in

survey research. Although Kerlinger (1966) reported that response rates on mail

surveys of 50 to 60% are among the best to be expected, he urged researchers

to make every effort to achieve response rates of 80 to 90%. Babbie (1973),

however, suggested that a response rate of "50% is adequate for analysis and

reporting. A response rate of at least 60% is good. And a response rate of 70%

is very good" (p. 165). Consequently it was concluded that the present study

required a minimum of 80 respondents and that a staff of at least 170 nurses

would be required to maximize the probability that the minimum number of

responses would be received. Therefore questionnaires were mailed (see details

below) to the entire 200 member RN and LPN nursing staff of a hospital for


Located in a southeastern state, the 113-bed hospital, which employs 513

persons, offers extensive general and specialized services to local residents, and

is one of the eight centers in its state designated to treat premature or

congenitally ill babies. In addition, cardiac surgery services are provided for

adult patients. Of the 200 questionnaires mailed, six were returned to sender

as undeliverable due to improper addresses, and 106 were returned by nurses,

representing a 55% return rate. Of those returned, seven were not usable due

to extensive missing data. The report of results of analysis includes specific

information regarding treatment of other missing data.

Ninety-nine percent of the respondents were female (Table 3-1). 45.3%

were between 21 and 30 years old, 39.6% were 31 to 40 years of age, 9.4% were

41 to 50 years of age, and 4.7% were 51 to 60 years of age. 2.9% of the

respondents had been in nursing less than one year, 53.4% for one to ten years,

28.6% for 10 to 19 years, and 15.2% for over 19 years. The highest nursing

education level attained by respondents was LPN, 6.6%; (Table 3-2) diploma,

Demographic Characteristics of Respondents

































TABLE 3-1--continued

Characteristics Percent


LPN 6.9

Diploma 29.2

Associate 32.1

Baccalaureate 22.6

Masters 9.4

Nursing Tenure

Less than 1 year 2.9

1-3 years 8.6

4-6 years 23.8

7-9 years 21.0

10-13 years 12.4

14-16 years 8.6

17-19 years 7.6

20-22 years 3.8

over 22 years 11.4

Job-Related Characteristics of Respondents

Characteristics Percent


Staff Nurse 59.0

Charge Nurse 11.4

Patient Care Coordinator 4.8

Other 24.8

Unit Tenure

less than 6 months 6.7

6 months to 1 year 11.4

1-3 years 32.4

3-5 years 27.6

5-10 years 15.2

over 10 years 6.7

TABLE 3-2 -- continued

Characteristics Percent


Medical Intensive Care 8.5

Surgical Intensive Care 13.2

Neo-natal 34.0

Surgical 9.4

Medical-Infectious 7.5

Medical-Chronic 5.7

Surgery 2.8

Other 18.9


Days 33.3

Evenings 20.0

Nights 14.3

12-hour days 20.9

12-hour nights 9.5

Other 2.9

29.2%; associate, 32.1%; baccalaureate, 22.6%; and masters, 9.4%. Among the

respondents, 33.3% worked day shift, and 20.0% worked evening shift, 14.3%

worked night shift, 20.9% worked a 12 hour day shift, 9.5% worked a 12 hour

night shift, and 2.9% worked other shifts.


An administrative official of the hospital was approached by the researcher

for the purpose of gaining an introduction to the nursing staff. The first contact

with the nursing staff was with a nursing educator. The education staff was

enthusiastic about participation in the project based on their anticipation of

benefit to the hospital, and a plan was devised to have the researcher attend a

meeting of the nurse managers of each unit to explain the research and arrange

for distribution of questionnaires by the nurse managers on the day following

the meeting. However, the nurse managers expressed reluctance about being

the people who asked the staff nurses to "do extra work". Subsequently, a plan

was devised to mail the questionnaires directly to the homes of the nurses. The

process of developing a plan for distribution of the questionnaires resulted in

added commitment to the research by the Director of Nursing.

Staff nurses were informed of the research and given some information

about the researcher in a memo signed by the Director of Nursing (Appendix A)

and enclosed in pay checks distributed the day before the surveys were mailed.

Complete anonymity was promised, and nurses were told that, while participation

was voluntary, they were encouraged to take the 45 to 60 minutes required to

complete the questionnaire. Individual feedback to participants was promised.

Packets were mailed to all nurses at addresses supplied by the Department

of Nursing. A letter from the researcher was enclosed (Appendix B) explaining

in brief outline the purpose of the research, recognizing that the time required

represented a real investment for participants, encouraging the investment, and

acknowledging that some items would seem repetitious but urging completion of

the full survey. Participants were again guaranteed anonymity, assured that they

would receive individual feedback, and told that results would not include mention

of the hospital by name. A short sheet of instructions (Appendix C) was also

included in the packet, indicating how selection of identification sequence should

be made. Each participant was sent a stamped, pre-addressed envelope for return

of the 50-page questionnaire.

Of the 200 questionnaires mailed, 194 were apparently delivered, and 106

were returned by the designated date. Time elapsed between mailing of the

surveys and designated final date for return was 17 days. At the end of the

second week, a follow-up letter (Appendix D) was sent, thanking participants and

indicating that the researcher was making necessary preparations to provide

individual feedback.

Feedback was provided to recipients in the form of copies of printouts

of scores on dependent variables, along with an explanatory letter indicating in

brief what the scales purport to measure, and mean scores for each scale. Since

names were never provided with the data, feedback was by identification sequence

number and was enclosed with pay checks.

Interviews were held individually with the head nurse of each unit, and

with the Director of Nursing. Each administrator was asked to share her

perception of stressors that were particularly salient for her unit, and each

provided information regarding severity and variability of diagnoses and prognoses,

incidences of emergency and/or death on the unit, average census, and type and

amount of interaction with family and friends of patients.

Three raters independently completed the technological instrument with

each individual unit as referent, and ranked units on manageability, treatability,

and likeability of patients. In addition, data were secured from the Nursing

Education and Director of Nursing offices relative to unit size, average census,

and general demographic factors of the nursing population.


Dependent Variables

Burnout. Burnout, defined by Maslach and Jackson (1981) as "a syndrome

of emotional exhaustion and cynicism that occurs frequently among individuals

who do 'people-work' of some kind" (p. 99), was measured by two instruments,

the Maslach Burnout Inventory (MBI, Appendix E) developed by Maslach, and the

Staff Burnout Scale for Health Professionals (SBSHP, Appendix F), developed by

Jones (1980).

Development of original items for the MBI was based on interview and

questionnaire data, as well as early exploratory research conducted by Maslach

and her associates. The original 47-item form of the scale was tested with a

sample of 605 professionals from a variety of fields. Factor analysis revealed

that four factors accounted for over 75% of variance. The items that loaded

on these four factors were used in a second administration of the scale, again

to professionals including nurses, teachers, social workers, probation officers,

counselors, mental health workers, and agency administrators. Factor analysis

of the results of the second administration revealed a structure very similar to

that which emerged in the first analysis. Subsequently, the results of the two

administrations were combined (N=1025) for assessment of the psychometric

properties of the instrument.

The MBI items are presented as statements about personal feelings and

attitudes, with a job-related orientation. Subjects are asked to respond to each

item on the basis of both frequency (how often?) and intensity (how strong?).

Because research (Gaines & Jermier, 1983; Golembiewski & Munzenrider, 1981)

has shown that frequency is the better predictor measure, only frequency was

measured in this study. The three subscales (the fourth factor was eliminated

due to high correlation with other subscales) of the current MBI attempt to

assess three dimensions of burnout: depersonalization, described as an unfeeling

and impersonal response to clients, measured by a five-item subscale; emotional

exhaustion, described as being emotionally overextended and exhausted by work,

measured by a nine-item subscale; and personal accomplishment, described as

feelings of competence and successful achievement in one's work with people,

measured by an eight-item subscale. Congruent validity for the scale was

supported (Maslach & Jackson, 1981) by findings of significant correlations of

subscale scores with behavioral manifestations of emotional exhaustion and

depersonalization, and with physical fatigue. In addition, job characteristics,

expected to be significantly related to subscale scores, were found to be so.

Discriminant validity of the instrument was demonstrated by the significant but

low correlation with general job satisfaction.

The SBSHP instrument is a 30-item scale, including 20 items designed to

measure burnout and a ten-item lie scale designed to identify attempts by

respondents to "look good," with responses based on a seven-item scale from

"agree very much" to "disagree very much." Although the burnout subscale items

were developed to measure the same dimensions of burnout as those measured

by the MBI (Shinn, 1982), Jones (1980) has interpreted the relatively high item-

with-total SPSHP score correlation (average = .71) as an indication that the

items measured an overall, encompassing construct rather than dimensions of a

construct. Items in the SPSHP attempt to measure psychological,

psychophysiological and behavioral components of strain, while the MBI attempts

to assess only psychological aspects of strain. Jones (1980) offers, as evidence

of validity of the SPSHP, data which demonstrate significant correlations between

SBSHP scores and measures of job-related behavior and affect conceptually related

to burnout, including absenteeism, employee theft and job satisfaction.

The two measures of burnout were used for this study to provide a

comparison of the two measures in a single study, to make replication of the

study in a non-health care setting possible, and to facilitate comparison of the

results with other studies.

Anxiety. Anxiety is defined by Spielberger et al. (1977) as "a palpable

but transitory emotional state or condition characterized by feelings of tension

and apprehension and heightened autonomic nervous system activity" (p. 240).

Research results suggest that anxiety is manifest as both a relatively stable

personality trait (trait anxiety) representing a predisposition to perceive a wide

variety of external stimuli as threatening, and as a transitory state (state anxiety)

that fluctuates over time in intensity of feelings of tension and apprehension.

Both trait anxiety and state anxiety are measured by the State Trait

Anxiety Inventory (STAI) developed by Spielberger and used in over 1900 studies

of anxiety. The instrument has demonstrated both internal consistency (Cronbach

alpha ranges have been .83 to .93) and the expected pattern of test-retest

reliability. That is, trait anxiety test-retest correlations have ranged from .73

to .86, while state anxiety test-retest correlations have ranged from .16 to .54

(Spielberger et al., 1970).

More recently, Spielberger, in collaboration with associates, has developed

the State-Trait Personality Inventory (STPI, Appendix G) which contains six 10-

item subscales: state anxiety, trait anxiety, state anger, trait anger, state

curiosity and trait curiosity. Since correlation of STPI anxiety subscale scores

(Spielberger, no date) with STAI scores for samples of college students and Navy

recruits all exceed .93, it appears that the STPI provides a valid and reliable

measure of state anxiety and of trait anxiety. In addition, it is felt that the

inclusion of the curiosity subscales in the cumulative questionnaire developed for

this study helped provide balance for the overall instrument. Addition of the

anger subscales provided opportunity to examine a possible relationship sometimes

speculated (Maslach & Jackson, 1982) to exist between anger and the

depersonalization subscale of burnout. Consequently, the STPI was used to

measure the dependent variable anxiety for this study.

Absence. Measurement of absence introduces two methodological issues:

means of data collection (self-report or archives) and type of data to be collected

(frequency of absence or amount of time lost from work). Although use of

archival data increases design strength by reducing the possibility of common

method variance, and has been considered a more valid measure of absence

behavior by some researchers (for example, Gupta & Beehr, 1979), others (Hammer

& Landau, 1981) have pointed out that the inconsistencies in classification of

absences as voluntary (usually considered a measure of absence frequency) or

involuntary (usually considered a measure of time lost) introduces potential for

criterion contamination. In addition, use of archival data precludes maintenance

of subject anonymity. Therefore, a self-report measure of absence behavior was

used in this study.