RESEARCH

ORGANISATIONAL CULTURE OF HOSPITALS TO PREDICT TURNOVER INTENTIONS

OF PROFESSIONAL NURSES

Dr E Jacobs
D Com
Human Resource Practitioner, Sol Plaatjie Municipality, Private Bag X5030, Kimberley
Corresponding author: [email protected]

Prof G Roodt
D Admin
Professor, Head of Centre of Work Performance and Departmental Research Unit, Department of Human Resource
Management, University of Johannesburg

Keywords: organisational culture; turnover intentions; professional nurses; general linear modelling; retention strat-
egy

ABSTRACT

The objective of this study was to determine if organisational culture predicts turnover intentions of professional
nurses. A predictive model with organisational culture and various proposed mediating variables, namely knowledge
sharing, organisational commitment, organisational citisenship behaviour and job satisfaction, as well as various
demographic variables was developed to determine turnover intentions through applying General Linear Modelling. A
correlational design with questionnaires was used. A sample of professional nurses (N = 530) in private and provincial
hospitals was obtained. The results indicate that organisational culture has a significantly negative correlation with
turnover intentions. Organisational culture also interacted with job satisfaction, knowledge sharing, and the white
professional nurses’ category to decrease turnover intentions and with Organisational Citisen Behaviours to increase
turnover intentions in a final predictive model. It is therefore recommended that nursing employers seriously embark
on strategies to improve the organisational culture to retain their talent.

OPSOMMING

Die doel van die studie was om te bepaal of organisasiekultuur arbeidsomset-voornemens van professionele
verpleegkundiges voorspel. ‘n Voorspellingsmodel met organisasiekultuur en verskeie voorgestelde tussenkomende
veranderlikes, naamlik kennisdeling, organisasieverbintenis, organisasie-burgerskapsgedrag en werkstevredenheid,
asook verskeie demografiese veranderlikes was ontwikkel deur Algemene Liniêre Modellering. ’n Korrelasie-ontwerp
met behulp van vraelyste is gebruik. ‘n Steekproef van professionele verpleegkundiges (N = 530) in private en
provinsiale hospitale is verkry. Die resultate toon dat organisasiekultuur betekenisvol negatief korrelleer met
arbeidsomsetvoornemens. Organisasiekultuur het ook in interaksie met werkstevredenheid en kennisdeling, asook
die kategorie blanke verpleegkundiges in ‘n finale model arbeidsomsetvoornemens verlaag en saam met organisasie-
burgerskapsgedrag arbeidsomsetvoornemens verhoog. Daar word aanbeveel dat werkgewers oorweging skenk aan
intervensies wat die organisasiekultuur gunstiger maak om personeel te behou.

HEALTH SA GESONDHEID Vol.13 No.1 - 2008 63

BACKGROUND AND PROBLEM STATE- behavior, job satisfaction) as well as various demo-
MENT graphic variables into an equation to determine which
variables independently and/or interactively emerged to
From various reports it is clear that the South African predict turnover intentions. These mentioned variables
nursing profession is in a crisis. Professional nurses were selected on the grounds of their respective rela-
leave the country in search of lucrative work overseas tionship with, or impact on, organisational culture and
and this exodus will have a catastrophic effect on the turnover intentions. This paper focuses on an impor-
delivery of health care over the next decade (Brits, tant part of this objective; it discusses the role of
2003:7; Derby, 2003:12). South Africa is one of the five organisational culture, as well as the interaction with
most prominent suppliers of overseas registrations the mediating and demographic variables, in predicting
(Vujicic, Zurn, Diallo, Adams & Dal Poz, 2004:2). These turnover intentions of professional nurses.
shortages of staff, due to the brain drain and turnover
problems in hospitals, are also creating various other More precisely the empirical objectives of this paper
problems such as enormous pressure on existing can be stated as:
employees (Jewkes, Abrahams & Mvo, 1998:1785; • to determine the relationship between
Newman, Maylor & Chansarkar, 2002:273; Smit, organisational culture and turnover intentions on a
2003:17) resulting in job stress and job dissatisfaction, bivariate level.
as well as medico-legal incidents (Smit, 2003:17). • to determine the amount of variance organisational
Nurse-patient relationships are also at times culture accounted for independently and
characterised by conflict, clinical neglect, verbal and interactively when entering all the selected variables
physical abuse (Jewkes et al. 1998:1781). in a predictive model of turnover intentions on a
multi-variate level.
Financial constraints make it almost impossible to
compete with remuneration offerings from national and LITERATURE OVERVIEW
international competitors to retain talent. Furthermore,
current exchange rates make it very attractive for Organisational Culture
professional nurses to earn foreign money and tax-free
salaries (Kockott, 2003:3). The existence of many job Organisational culture has been defined in many ways
opportunities overseas (Smetherham & Laurance, by various authors and researchers. However, many
2003:5), the influence of globalisation and the belief would agree that organisational culture refers to a set
that a person’s career is enriched with overseas of values, beliefs and behaviour patterns that form the
experience, further contributes to making this problem core identity of organisations; and these help in shaping
almost uncontrollable for hospital employers. the employees’ behaviour (Erwee, Lynch, Millet, Smith
& Roodt, 2001:7; Van der Post, de Coning & Smit,
This raises the question of whether employers should 1997:147).
not rather focus their retention strategies on things they
can control internally to retain their employees. An ap- According to Van der Post et al. (1997:148) the study
proach to build strategies around the needs and work of organisational culture can be approached by
circumstances of professional nurses, rather than only identifying certain dimensions that shape behaviour and
focusing of financial incentives, was adopted in this eventually can be regarded as the culture of the
research. organisation. The advantage of this approach is that
the model provides an organisational analysis that
THE PURPOSE OF THIS STUDY perceives culture as a feature of the organisation that
can be studied and manipulated in isolation (Erwee et
The main purpose of the research was to develop a al. 2001:8). The dimensions are the following: conflict
predictive model of turnover intentions by entering the resolution, culture management, customer orientation,
independent variable (organisational culture) and the disposition towards change, employee participation,
proposed mediating variables (knowledge sharing, goal clarity, human resource orientation, identification
organisational commitment, organisational citisenship with the organisation, locus of authority, management

64 HEALTH SA GESONDHEID Vol.13 No.1 - 2008

style, organisational focus, organisational integration, sors of intention to quit are better understood, the em-
performance orientation, reward orientation and task ployer could possibly institute changes to affect this
structure. Organisational culture, for the purpose of this intention. However, once an employee has left, there is
study, was defined as: ”…a system of shared meaning, little the employer can do except the expense of hire
the prevailing background fabric of prescriptions and and train another employee.
proscriptions for behaviour, the system of beliefs and
values that ultimately shape organisational behaviour The relationship between organisational
(Van der Post et al. 1997:147).” culture and turnover intentions

Turnover Intentions A review of the literature reveals a limitation in research

regarding the relationship between organisational culture
Sager, Griffeth and Hom (1998:255) referred to turnover and turnover intentions. However, it is widely reported,
cognitions as mental decisions intervening between an although not empirically yet determined, that
individual’s attitude regarding the job, and decision to organisational culture in hospitals can contribute
leave or stay. According to Elangovan (2001:162), towards lower turnover (Coile, 2001:224). Coile
intention to leave represents an attitudinal orientation (2001:226) reported that hospitals with lower turnover
or a cognitive manifestation of the behavioural decision clearly indicated a more positive organisational culture
to leave. than the others. Culture within these hospitals had a
strong mission, empowerment of leaders, participative
The theory of planned behaviour (Ajzen, 1991:179), management, patient focus, reward orientation (for
suggested that behavioural intention is a good predictor improvement), visible/accessible leaders and supported
of actual behaviour. It has been successfully education. Tepeci and Bartlett (2002:163) found that
demonstrated in previous studies that behavioural the perceived culture explained 39% of the variance in
intention to leave is consistently correlated with turnover intent to leave in a sample of 182 students in the
(Fox & Fallon, 2003:3). There is considerable support hospitality industry.
for the notion that intention to leave is probably the
most important and immediate antecedent of turnover Petkoon and Roodt (2004:46) were of the opinion that
decisions (Fox & Fallon, 2003:3; Slate & Vogel, South Africa’s success in the global economy will be
1997:402). Shields and Ward (2001:692) reported that positively or negatively affected by the ability of individual
quitting intentions were the strongest predictor of actual companies and industries to adapt their cultures to fit
turnover, with 79% of nurses in a longitudinal study the global challenges. One of these challenges is
reporting an intention to leave and did so within one certainly the retention of professional people in a global
year. Steel and Ovalle (1984:673) reported in a large economy. While culture is not the only determinant of
number of studies between 1965 and 1983, a correlation business success or failure, it can have a competitive
coefficient of 0.50 between quitting intentions and ac- advantage in the retention of top-quality staff (Sadri &
tual turnover. Lees, 2002:11).

Some commentators would postulate that the notion An extensive literature research yielded no results of
of commitment to either job or profession is a more the concept of organisational culture in South African
useful concept than turnover cognitions. However, hospitals. Related concepts, such as lack of staff
research indicates that turnover cognitions have a distinct support and absence of proper incentives were reported
conceptual base and significant explanatory power (Hall, 2004:28). Both these concepts are seen by Van
(Carbery, Garavan, O’Brien & McDonnell, 2003:649). der Post et al. (1997:150) as dimensions of
Lambert, Hogan and Barton (2001:236) suggested that organisational culture, namely the management style
more attention should be given to the direct and indirect and reward orientation. The caring profession in parts
influences of variables on intention to leave as opposed of the South African public health services are
to the actual act of turnover. From the employer’s point characterised by conflict (Jewkes et al. 1998:1781).
of view, intention to leave may be a more important Van der Post et al. (1997:149) saw conflict resolution
variable then the actual act of turnover. If the precur- as an important dimension of organisational culture.

HEALTH SA GESONDHEID Vol.13 No.1 - 2008 65

Furthermore, the severe staff shortages threaten ser- competencies and participating in any organisational
vice capability and the quality of patient care (Newman activity like training or any other programme to enhance
et al. 2002:273; Smetherham & Laurance, 2003:5). knowledge…” (Yang & Wan, 2004:62)”… to ultimately
Once again, Van der Post et al. (1997:149) regarded create value…” (Sveiby in Wagner, Cassimjee & Nel,
customer orientation as an important dimension of 2002:50).
organisational culture. It is therefore clear that although
related aspects of organisational culture were reported Mowday (1999:389) described organisational
in South African hospitals, the concept was not theo- commitment as the attachment that is formed between
retically and empirically investigated and reported. employees and their employing organisation. Roodt
(1997:7) proposed measuring commitment by
Knowledge sharing, organisational distinguishing between different commitment foci. An
commitment, organisational citisenship understanding of the different foci can probably explain
behavior and job satisfaction the dynamics of employee commitment in the work
context. According to Roodt (2004a:85) a golden thread
Knowledge sharing, organisational commitment, running through all the definitions of commitment is the
organisational citisenship behaviours (OCB’s) and job potential of a particular focus to satisfy salient needs.
satisfaction are distinct concepts from organisational Roodt (2004a:85) therefore proposed a motivational
culture. These concepts were included in the study to approach to define and measure commitment and
determine whether they interact with organisational describes it as “…a cognitive predisposition towards a
culture in predicting turnover intentions. To date, most particular focus, insofar this focus has the potential to
of the studies concentrated on job satisfaction and satisfy needs, realise values and achieve goals.”
organisational commitment as antecedents for turnover
intentions and indicated significant relationships Organisational citisenship behaviours (OCB’s) are
(DeConinck & Stilwell, 2004:229). These concepts are employee work behaviours such as helping others,
very important and need further investigation, specifically staying late, or working weekends, performing at levels
with the current scenario of professional nurses in South that exceed enforceable standards, tolerating
Africa. It was also suggested in this study, that two impositions or inconveniences on the job, and being
other organisational processes need investigation. actively involved in company affairs (Podsakoff,
These are knowledge sharing behaviour and Mackenzie, Paine & Bachrach, 2000:516). Organ
organisational citisenship behaviours. These have been (1988:4) defined organisational citisenship behaviour as:
selected as it believed that they will fulfil professional “…individual behaviour that is discretionary, not directly
nurses’ higher order needs as described by the work of or explicitly recognised by the formal reward system,
Maslow (1954). and that in the aggregate promotes the effective
functioning of the organisation”. “Discretionary” means
A short description of each concept is provided as the that the behaviour is not an enforceable requirement of
emphasis of this paper is on organisational culture and the role or the job description, that is, the clearly
turnover intentions. specifiable terms of the person’s employment contract
with the organisation. The behaviour is rather a matter
Knowledge sharing occurs when an individual is of personal choice, so that its omission is not generally
willing to assist as well as learn from others in the understood as punishable.
development of new competencies. To “learn” means
to “digest”, to “absorb”, and to “apply” (Salopek & Dixon, Job satisfaction refers to a worker’s general attitude
2002:63). Ryu, Hee Ho and Han (2003:113) emphasised towards his or her job. Weiss, Dawis, England and
that knowledge sharing is a people-to-people process. Lofquist (1967) cited in Cook, Hepworth, Wall and Warr
For the purpose of this study knowledge sharing was (1981:21) identified various extrinsic factors (for example,
defined as “…a willingness…” (Chua, 2003:117; Hislop, supervision, compensation, company policies and
2003:184) “…on the part of those workers who possess practices) and intrinsic factors (for example, activity,
knowledge to share and communicate it…” (Hislop, variety, responsibility) as potential sources of
2003:184), “…to teach others, to develop new satisfaction or dissatisfaction. The intrinsic factors are

66 HEALTH SA GESONDHEID Vol.13 No.1 - 2008

thought to measure satisfaction with intrinsic The next step was to decide on the actual composition
reinforcement factors. The extrinsic factors are external of the unit of analysis (more or less 1100 professional
to the job. Job satisfaction is therefore, for the purpose nurses work in the selected hospitals). Although the
of this study, defined as “…The degree people achieve target population was determined by the set criteria as
and maintain correspondence with the environment explained; it was decided to make use of a non-
satisfying both extrinsic and intrinsic needs (Weiss et probability sample at each of the selected hospitals. A
al. 1967 cited in Cook et al. 1981:21).” non-probability sample in this sense refers to all
professional nurses who were available to complete the
METHOD questionnaire at a specific point in time (normally a
two day period) in each hospital.
Research design
The respondents were not randomly selected. All who
The research questions were investigated (tested) by met the criteria and were available were included. Such
making use of a correlational design. The an approach is regarded as less scientific, but more
conceptualisation or mode for reasoning to choose a economical in terms of time and money (Welman &
correlational design was first to test hypotheses and Kruger, 1999:62). The limitations compared to
secondly to analyse variables to propose a theoretical systematic or stratified sampling methods are
model. acknowledged. However, Kerlinger (1986) defends the
use of non-probability samples by noting that it may
Sample lack the virtues of random sampling, but are often
necessary and unavoidable. Their weakness can to
The target population was professional nurses working some extent be mitigated by using knowledge,
in hospitals. The requirements were that these nurses expertise, and care in selecting samples as was the
should be regarded as professional and registered as case in this study. It seemed practically and ethically
such. Staf- or assistant nurses were not included in correct to utilise a convenience sample as employers
the target population. Various important challenges and were reluctant to allow time off to complete the
problems had to be addressed to determine the sampling questionnaires due to staff shortages and the complex
frame and unit of analysis. According to figures by the shift system that determine working hours for
South African Nursing Council (in 2004), there were professional nurses, as well as the research procedure
approximately 93 000 registered professional nurses to personally visit hospitals to ensure a high response
in South Africa. (This figure must not be confused with rate of questionnaires.
the estimated 155 400 if staff nurses are also taken in
consideration). Professional nurses are employed in The final sample consists of 530 respondents in the
different sectors (for example, hospitals, municipalities, selected hospitals (more or less 50%). This resulted in
academic institutions and other industries). However, a sample with diverse characteristics regarding race,
hospitals are the most important employers of age, home language, number of dependents, level of
professional nurses (Green, 2003:3). It was therefore seniority, qualifications, tenure in hospital and
decided to only include professional hospital nurses. profession, working in different units and gender (Table
1). According to Huysamen (in Welman & Kruger,
The next step was to select which hospitals to include 1999:64) it is not necessary to draw a sample size
in the sampling frame. It was decided to include five bigger than 500 because it has little effect in decreasing
private hospitals and four provincial hospitals in the Free the standard error of measurement.
State, Northern Cape and North West Provinces of South
Africa in the sampling frame. The motivation for this Table 1 shows that all the different categories of
was based on the assumption that different types of subgroups were well represented. The only demographic
hospitals (private and provincial) in different provinces variable discarded for further analysis was gender as
and surroundings (urban and rural) will probably have nurses were predominantly female (95%).
different organisational cultures that will yield a sample
with diverse characteristics.

HEALTH SA GESONDHEID Vol.13 No.1 - 2008 67

Table 1: Demographics of sample of professional nurses (N=530)

Type of hospital Race* Gender

Private 368 69% African 195 37% Male 26 5%
Government 162 31% White 265 50% Female 504 95%
Coloured 67 13%
Age* Unit Tenure in
hospital
Younger 30 years 88 17% ICU/Casualties 181 34% Not a year 87 16%
completed
30-39 years 188 35% Paed/Maternity 94 18% 1-5 years 173 33%
40-49 years 161 30% Surgery/Theatre 111 21% 6-10 years 160 30%
50 years and older 92 18% General 144 27% 11 years and 110 21%
more
Completed years in Highest Qualification Home language*
unit
Less than 1 115 22% Nursing diploma 220 42% Afrikaans 308 58%
1-5 years 241 45% Post nursing diploma 193 36% Other (English 92 18%
included)
6 years and longer 174 33% Nursing degree or post 117 22% African language 129 24%
qualification
Years completed in Dependents under 18 Dependents
profession above 18
5 years and less 99 19% None 163 31% None 198 37%
6-10 years 97 18% One 105 20% One 158 30%
11-20 years 183 35% Two 164 31% Two 84 16%
20 years and more 151 28% Three or more 98 18% Three or more 90 17%
Rank Marital status
Professional nurse 294 56% Not married 140 26%
Senior professional 65 12% Married 295 56%
nurse
Chief/Unit manager 171 32% Divorced 95 18%
* missing responses recorded

Measuring instruments the dimensions that can influence the nurse’s perception
of the organisation. It was developed in South Africa by
The Organisational Culture Survey (OCS) (97 items) Van der Post et al. (1997:147-168) and a Cronbach
was used to measure fifteen dimensions of Alpha of 0.991 was obtained in a study by Erwee et al.
organisational culture. The OCS is based on a model (2001:9) indicating high reliability. A Cronbach Alpha of
of organisational analysis that perceives culture as a 0,70 or higher is considered acceptable (Cortina,
feature of the organisation that can be studied and 1993:98). It therefore was an appropriate instrument to
manipulated in isolation (Erwee et al. 2001:8).This measure organisational culture. A deviation from the
questionnaire was selected because it seems to contain original questionnaire was followed as each item was

68 HEALTH SA GESONDHEID Vol.13 No.1 - 2008

measured on a seven-point intensity response scale readily share knowledge (for example, that trust exists
anchored at extreme poles. The items were changed -6 questions), why one would not readily share knowl-
from statements to questions. An example is: ”How edge [for example, career would be in danger -3 ques-
aware are employees of the objectives in this tions (-)], and why others do not readily share knowl-
organisation?” (“Not aware at all” 1-low intensity, to edge [for example, colleagues don’t want to do like-
“Always aware” 7-high intensity). By using this wise -3 questions(-)]. A Cronbach Alpha of 0.839 was
questionnaire, the individual work unit’s culture, the obtained for the Knowledge Sharing Questionnaire. The
individual hospital’s culture and the overall organisational questionnaire is attached as it is a new development
culture could be determined. (Annexure B).

Turnover intentions were measured by a questionnaire The Organisational Commitment Questionnaire

developed by Roodt (2004b:1). The questionnaire developed by Roodt (1997:8) was used to measure
consists of 14 items that were measured on a seven- commitment. The questionnaire was suitable for this
point intensity response scale anchored at extreme investigation as it can be regarded as a motivational
poles (e.g. “never” 1-low intensity, to “always” 7-high approach to study commitment. The questionnaire
intensity). Examples of items included in this consists of 38 items, each with a five-point intensity
questionnaire (Roodt, 2004b:1) were: “How often have response scale anchored at extreme poles. The foci of
you recently considered leaving your job?” and “How the questionnaire consist of work, career, occupation
frequently have you been scanning news papers for new and organisation. An example of an item is: “How much
job opportunities?” A Cronbach Alpha of 0.913 was ob- time and energy do you willingly devote to work?” (“no
tained in this study indicating an acceptable reliability. time” 1-low intensity to “all of my time” 5-high intensity).
The turnover literature lacks formally validated scales A Cronbach Alpha of 0.926 was obtained for the
to represent turnover cognitions (Sager et al. 1998:259). Organisational Commitment Questionnaire.
The motivation to develop this questionnaire is that most
instruments in the literature measure turnover intentions Organisational citisenship behaviour was assessed by
on only a relatively small number of items. Various the Van Dyne and LePine (1998:108-120) seven-item
researchers have used only one item (Guimaraes, “helping” scale. The items were originally adapted from
1997:46). The approach to use single-item indicators Organ and Konovsky (1989:160). Four additional items
to measure turnover cognitions is criticised as construct of the original altruism scales from the work of Smith,
validity in unknown (Sager et al. 1998:259). Only a few Organ and Near (1983:653-663) were also included in
studies could be found where more than three items the questionnaire. Voice was measured by a 6-item
per instrument were used (Fox & Fallon, 2003:3). The scale as used by Van Dyne and LePine (1998:108-
questionnaire is attached as it is a new development 120). A deviation from the original helping questionnaire
(Annexure A). was followed as each item was measured on a seven-
point intensity response scale anchored at extreme
A literature study was conducted to compile a poles (“never” 1-low intensity, to “always” 7-high inten-
questionnaire as no suitable measure for knowledge sity). An example of an item for the helping question-
sharing was found. The questionnaire consists of 23 naire was: “How often do you volunteer to do things for
items, each with a five-point intensity response scale this work group?” and for the voice questionnaire: “How
anchored at extreme poles ranging from “to no extent” frequently do you make recommendations concerning
or “disagree” (low intensity) to “to a large extent” or issues that affect your work group?” Self ratings were
“agree” (high intensity). The questionnaire consists of used. A Cronbach Alpha of 0.923 was obtained for the
six sections representing different domains of knowledge Organisational Citisenship Questionnaire.
sharing, namely why knowledge is shared in the
organisation (for example, to get recognition -5 ques- Job satisfaction was assessed by the Minnesota
tions), opportunities for knowledge sharing (for example, Satisfaction Questionnaire ((Weiss et al. 1967 cited in
to attend training courses -3 questions), the contribu- Cook et al. 1981:21). The MSQ20 measures 20 differ-
tion of knowledge sharing to the organisation (for ex- ent items and can be categorised into extrinsic and
ample, competitiveness -3 questions), why others intrinsic satisfaction. The questionnaire measures the

HEALTH SA GESONDHEID Vol.13 No.1 - 2008 69

following satisfaction domains: activity, independence, questionnaire was evaluated by the Statistical
variety, social status, moral values, security, social Consultation Service of the University of Johannesburg.
service, authority, ability utilisation, responsibility, cre-
ativity, achievement (intrinsic), supervision-human re- After completion of the questionnaires, a thorough
lations, supervision-technical, company policies and procedure to clean the data was followed. A
practices, compensation, advancement, working con- questionnaire was discarded if there were missing
ditions, co-workers and recognition (extrinsic). Ques- responses. A number of 570 questionnaires were
tions are measured on a five-point scale and will make distributed. A final number of 530 questionnaires were
use of an intensity scale with 1-low intensity (for ex- usable and 40 rejected. After this, the data was captured
ample, never) and 5-high intensity (for example, always). and processed by STATCON (Statistical Consultation
An example is: “What are the chances for advance- Services at the University of Johannesburg).
ment in your present job? (1-no chance; 5-high chance).
A Cronbach Alpha of 0.886 was obtained for the MSQ. Statistical analysis

Each questionnaire was factor analysed according to The first objective was to determine the relationship
the procedure suggested by Schepers (2004:31) in order between organisational culture and turnover intentions.
to determine the factor structure of the instrument. One This was tested by the Pearson product-moment
of the advantages of this method is to minimise correlation (this hypothesis was further analysed by
artefactors (Schepers, 1992:1). This procedure includes introducing General Linear Modelling).
first and second level factor analysis. Factors were
postulated according to Kaiser’s (1970) criterion The second objective, namely to determine the amount
(eigenvalues-greater-than-unity) and extracted by of variance organisational culture accounted for
means of Principal Axis Factoring. Sub-scores were independently and interactively when entering all the
calculated on the obtained factors and they were selected variables in a predictive model of turnover
subsequently intercorrelated. intentions, was tested by applying General Linear
Modelling. According to Kerlinger (1986) it is unrealistic,
Procedure even wrong, to study and learn only an approach that
is basically bivariate in conception. Multivariate
Permission to do the research was obtained from the methods are like the behavioural reality they try to
Regional Director of the private hospitals and the reflect: complex and difficult to understand. Kerlinger
Director: Knowledge Management of the provincial continues by arguing that from all methods of analysis,
hospitals. A letter of consent was provided by the Ethics multivariate methods are the most powerful and
Research Committee of the Faculty Ethics Committee appropriate for scientific behavioural research. His
of the University of Johannesburg. argument basically rests on the idea that behavioural
problems are almost all multivariate in nature and cannot
Questionnaires were completed in venues outside the be solved with a bivariate (two-variable) approach - that
units, such as training centres, to prevent interruptions. is, an approach that considers only one independent
The purpose of the questionnaire was explained shortly and one dependent variable at a time.
by the researcher and each respondent completed the
questionnaire personally. The main reason for this Broadly stated, the theory driving General Linear
approach was to ensure that all items were completed, Modelling (GLM) seeks to identify those quantities in
time was saved for employees, confidentiality was systems of equations which remain unchanged under
improved and a quick response was promoted. All linear transformations of the variables in the system.
questionnaires were handled anonymously. The General Linear Model differs from the multiple
Questionnaires were only available in English. Although regression model in terms of the number of dependent
many respondents indicated that their home language variables that can be analysed. The General Linear
is Afrikaans or an African language, no problems were Model goes a step beyond the multivariate regression
reported as everyone was proficient in English. A pilot model by allowing for linear transformations or linear
study was conducted in two rounds and the combinations of multiple dependent variables. This

70 HEALTH SA GESONDHEID Vol.13 No.1 - 2008

extension gives the General Linear Model important determined by entering the independent, mediating vari-
advantages over the multivariate regression models, ables and all the demographic variables into the equa-
which are inherently univariate (single dependent tion with interactions. The technique can be seen as
variables) methods. One advantage is that multivariate entering various carefully selected variables into the
tests of significance can be employed when responses equation where only the “strongest will ultimately sur-
on multiple dependent variables are correlated. vive” to predict the outcome variable (turnover intentions
Multivariate tests of significance of independent linear in this study). The GLM follows a process by first en-
combinations of multiple dependent variables can also tering the demographic variables into the equation, then
give insight into which dimensions of the response the independent and mediating variables and finally all
variables are, and are not, related to the predictor the variables simultaneously to determine independent
variables. Independent variables can be continuous (for and interactive effects. From this final model it would
example, dimensions of organisational culture) or be possible to determine the independent and interac-
categorised (for example, demographic variables). tive role of the independent variable (organisational cul-
ture), the demographic variables (categorised variables)
General Linear Modelling is also useful to compile and the mediating variables (knowledge sharing,
predictive models and develop theories (a focus of this organisational commitment, OCB’s and job satisfac-
study). A theory is a group of logical, related state- tion) in determining turnover intentions. Only the re-
ments that are presented as an explanation of a phe- sults applicable to organisational culture will be dis-
nomenon. A theory thus encompasses one or more cussed next.
hypotheses (Welman & Kruger, 1999:11). Analysis of
variance (ANOVA) is used to uncover the main and in- RESULTS
teraction effects of categorical independent variables
(called “factors”) on an interval dependent variable. A The first important result is that a significant negative
“main effect” is the direct effect of an independent vari- relationship (r= -0.521; p<0.01) between organisational
able on the dependent variable. An “interaction effect” culture and turnover intentions was found, meaning that
is the joint effect of two or more independent variables the more positive professional nurses were about the
on the dependent variable. In using General Linear culture of the hospital, the less inclined they were to
Modelling (GLM), the most parsimonious model was turnover intentions. One should remember that turn-

Table 2: Intercorrelation matrix (Pearson Correlations) of the second order factors (new factors) of the
different constructs

B: C:1 C:2
D: F: G:
Organisational Knowledge Knowledge
Org E: OCB Job Turnover
Culture sharing sharing
Commitment satisfaction Intentions
Factor One Factor Two

B 1 0.588(**) 0.404(**) 0.551(**) 0.328(**) 0.688(**) -0.521(**)

C1 0.588(**) 1 0.387(**) 0.445(**) 0.310(**) 0.549(**) -0.418(**)

C2 0.404(**) 0.387(**) 1 0.456(**) 0.448(**) 0.376(**) -0.234(**)

D 0.551(**) 0.445(**) 0.456(**) 1 0.549(**) 0.605(**) -0.402(**)

E 0.328(**) 0.310(**) 0.448(**) 0.549(**) 1 0.448(**) -0.060

F 0.688(**) 0.549(**) 0.376(**) 0.605(**) 0.448(**) 1 -0.516(**)

G -0.521(**) -0.418(**) -0.234(**) -0.402(**) -0.060 -0.516(**) 1

** Correlation is significant at the 0.01 level (2-tailed).

HEALTH SA GESONDHEID Vol.13 No.1 - 2008 71

over intentions are the most important predictor to ac- Secondly, in using General Linear Modeling (GLM), the
tual turnover. A significant positive correlation exists most parsimonious model was chosen including all the
between organisational culture and knowledge sharing, variables with interactions, explaining almost 49% of
organisational commitment, organisational citizenship the variance in turnover intentions. The coefficients are
behaviour and job satisfaction. The correlations between depicted in Table 3. The results of the most
organisational culture and all the other variables appear parsimonious predictive model for turnover intentions
in Table 2. yielded that the interaction between organisational cul-

Table 3: General linear modelling (GLM) for predicting turnover intentions

Coefficients

Unstandardized Standardized
Coefficients Coefficients
T Sig.
Beta
B Std. Error

(Constant) 7.182 0.307 23.429 0.000

Organisational Culture*Job Satisfaction -0.145 0.020 -0.680 -7.122 0.000

Organisational Culture*Organisational Citizen

0.102 0.013 0.632 7.811 0.000
Behaviours

Organisational Culture * Indicator - White -0.106 0.024 -0.155 -4.387 0.000

Organisational Commitment (***) -0.510 0.103 -0.211 -4.942 0.000

Indicator - ICU/Casualties and 50+ years (Age) -0.657 0.225 -0.096 -2.925 0.004

Organisational Culture*Knowledge Sharing 1 -0.056 0.017 -0.256 -3.241 0.001

Indicator - 1-5 years (Current Unit) and Chief

0.429 0.127 0.109 3.367 0.001
Nurse

Indicator - 11+ years (Current Hospital) and

0.729 0.199 0.119 3.672 0.000
None (Dependents > 18)

Indicator - Married/Co-Habitating and None

-0.319 0.115 -0.091 -2.785 0.006
(Dependents > 18)

Indicator - 50+ years (Age) and None

-0.491 0.160 -0.104 -3.078 0.002
(Dependents <18)

Indicator – ICU/Casualties and Degree -0.465 0.156 -0.096 -2.972 0.003

Dependent Variable: Turnover Intentions THE * SIGN DENOTES INTERACTION

(+) Increasing turnover intentions

(-) Decreasing turnover intentions

*** only independent predictor

72 HEALTH SA GESONDHEID Vol.13 No.1 - 2008

ture and job satisfaction (B = -0.145); the interaction workload, and that this in turn leads to more work
between organisational culture and knowledge sharing pressure and higher turnover intentions. This may
(B = -0.056 ) and the interaction between organisational especially be true if they perceive that teamwork is
culture and white professional nurses (B = -0.106) de- lacking or their efforts are not compensated for. As
creased turnover intentions, while the interaction be- stated, although work responsibilities and OCB’s are
tween organisational culture and organisational citizen- conceptually or theoretically distinct, this distinction
ship behaviour (B = 0.102) increased turnover inten- may not prevail in practice. One should also keep in
tions. A significant result was that organisational com- mind that the nursing profession entails helping behavior
mitment emerged as the only independent predictor of as a logical task characteristic of the profession. They
turnover intentions in the final model (positive B coeffi- may therefore not realise that they are involved in
cients increase turnover intentions and negative B co- citizenship behavior. One also may speculate that self
efficients decreases turnover intentions). ratings on OCB’s might have influenced this result in
the sense that performance was evaluated and not
DISCUSSION OCB’s.

The result of this study indicated that a statistically Turnover intentions of nurses can be actively managed
significant negative relationship exists between through the management of contextual variables such
organisational culture and turnover intentions. This as organisational culture and commitment. Since
finding also supports the view that organisational culture limited availability of professional nurses is a world-wide
contributes to organisational success and outcomes phenomenon, it is highly recommended that turnover
(Martins & Martins, 2002:58) and also lower turnover prevention strategies should not only focus on
intentions (Coile, 2001:226). It seems that organisational compensation factors, but on well-planned interventions
culture contributes significantly in decreasing turnover by hospital and nursing employers in the short- and
intentions, emerging with knowledge sharing, job long term. The predictive model developed for turnover
satisfaction and white professional nurses (as race intentions can be regarded as an important tool for
category) as predictors of turnover intentions of health and nursing administrators to plan talent retention
professional nurses. This result contributed to a greater strategies. Since this model focused primarily on
understanding of the importance of organisational culture internal components, possible strategies can be derived
in hospitals and amongst professional nurses. Since from this model to prevent turnover intentions. The
the importance of knowledge sharing was evident in practical value of this model is that it focused on
the results obtained from this study, the responsibility controllable dimensions of organisational culture.
of managers and human resource departments to Employers can plan interventions specifically, even if
create opportunities for learning and sharing, such as some results can only be achieved in the long term.
training and development interventions to meet growth
and development needs, are extremely important. RECOMMENDATIONS

As mentioned earlier, although the emphasis of this Organisational culture is a complex concept and
paper is not on organisational commitment, it emerged simplistic models to explain organisational behaviour
as the only independent predictor of turnover intentions and effectiveness will not improve the understanding
(the only variable “strong enough” to predict turnover thereof. Erwee et al. (2001:11) suggestion for a more
intentions on its own). This result supports previous complex model of organisation culture, constructed from
research that organisational commitment predicts different systems or subsystems, which interact on
turnover intentions. (DeConinck & Stilwell, 2004:229). different levels, is supported. It is recommended that
more turnover models should be developed, especially
A surprising result is that the interaction between with manageable variables. More emphasis should be
organisational culture and organisational citizenship placed on organisational cultures of hospitals to ensure
behavior are related to increasing turnover intentions. A that professional nurses see it as an employer of choice.
possible explanation might be that professional nurses It is also recommended that nursing employers per-
feel that being involved in OCB’s increases their ceive the organisational culture as manageable, not just

HEALTH SA GESONDHEID Vol.13 No.1 - 2008 73

something that it the natural outcome of “something and their use. London: Academic Press.
not controllable”. CORTINA, JM 1993: What is coefficient Cronbach alpha? An
examination of the theory and applications. Journal of Applied
The turnover model developed in this study furthermore Psychology, 78(1):98-104.
only focused on internal dimensions and excluded DECONINCK, JB & STILWELL, CD 2004: Incorporating organizational
external forces such as job opportunities, which justice, role states, pay satisfaction and supervisor satisfaction
theoretically were described as important in turnover in a model of turnover intentions. Journal of Business Research,
cognitions. It is recommended to determine the effect 57(3): 225-231.
of job opportunities available on the model. DERBY, R 2003: Netcare gets a R1,4-bn boost from the UK.
Citizen, 13 September.
CONCLUSION ELANGOVAN, AR 2001: Causal ordering of stress, satisfaction
and commitment, and intention to quit: A structural equations analy-
It is surprising that organisational culture has not sis. Leadership and Organization Development Journal,
previously been viewed as an important antecedent of 22(4):159-165.
organisational outcomes in both South African hospitals ERWEE, R; LYNCH, B; MILLET, B; SMITH, D & ROODT, G 2001:
and in the nursing profession alike. This study has Cross cultural equivalence of the organisational culture survey in
contributed to a greater understanding of the importance Australia. SA Journal of Industrial Psychology, 27(3):7-12.
of organisational culture in hospitals and amongst FOX, SR & FALLON, BJ 2003: Modeling the effect of work/life
professional nurses. balance on job satisfaction and turnover intentions. Symposia
paper presented at the 5th Australian Industrial and Organisational
In conclusion it is clear that organisational culture, in Psychology Conference, Melbourne, Australia.
interaction with the selected variables of knowledge GREEN, J 2003: Plan to make grass greener for SA nurses. Star,
sharing, job satisfaction and OCB’s, as well as 6 May.
organisational commitment as independent predictor GUIMARAES, T 1997: Assessing employee turnover intentions
and various demographic variables, interactively predict before/after TQM. International Journal of Quality and
turnover intentions. This answers the objectives of this Reliability Management, 14(1):46-63.
study, namely that a relationship between organisational HALL, EJ 2004: Nursing attrition and the work environment in
culture and turnover intentions of professional nurses South African health facilities. Curationis, 27(4):28-36.
exists, and that organisational culture accounted HISLOP, D 2003: Linking human resource management and
interactively with knowledge sharing, job satisfaction, knowledge management via commitment: A review and research
OCB’s and white professional nurses to predict turnover agenda. Employee Relations, 25(2):182-202.
intentions. JEWKES, R; ABRAHAMS, N & MVO, Z 1998: Why do nurses
abuse patients? Reflections from South African obstetric services.
REFERENCES Social Science & Medicine, 47(11):1781-1795.
KAISER, HF 1970: A second-generation Little Jiffy.
AJZEN, I 1991: The theory of planned behaviour. Organizational Psychometrika, 35(4):401-415.
Behavior and Human Decision Processes, 50(2):179-211. KERLINGER, FN 1986: Foundations of Behavioural Research.
BRITS, E 2003: Te veel verpleërs land-uit, sê RGN. Burger, 2 July. Orlando Florida: Harcourt Brace Jovanovich College Publishers.
CARBERY, R; GARAVAN, TN; O’BRIEN, F & MCDONNELL, J 2003: KOCKOTT, F 2003: Government developing strategies to retain
Predicting hotel managers’ turnover cognitions. Journal of health professionals. Business Day, 21 January.
Managerial Psychology, 18(7):649-679. LAMBERT, EG; HOGAN, NL & BARTON, SM 2001: The impact of
CHUA, A 2003: Knowledge sharing: A game people play. Aslib job satisfaction on turnover intent: A test of a structural
Proceedings: New information perspectives, 55(3):117- measurement model using a national sample of workers. Social
129. Science Journal, 38(2):233-250.
COILE, RC 2001: Magnet hospitals use culture, not wages, to MARTINS, E & MARTINS, N 2002: An organisational culture model
solve nursing shortage. Journal of Healthcare Management, to promote creativity and innovation. SA Journal of Industrial
46(4):224-228. Psychology, 28(4):58-65.
COOK, JD; HEPWORTH, SJ; WALL, TD & WARR, PB 1981: The MASLOW, A 1954: Motivation and Personality. New York. Harper
experience of work. A compendium and review of 249 measures & Row.

74 HEALTH SA GESONDHEID Vol.13 No.1 - 2008

MOWDAY, RT 1999: Reflections on the study and relevance of participation in correctional decision making and its impact on
organizational commitment. Human Resource Management employee stress and thoughts about quitting. Journal of Criminal
Review, 8(4):387-401. Justice, 25(5):397-408.
NEWMAN, K; MAYLOR, U & CHANSARKAR, B 2002: “The nurse SMETHERHAM, J & LAURANCE, J 2003: UK poaching of SA nurses
satisfaction, service quality and nurse retention chain”: Implications carries on despite ban. Cape Times, 13 May.
for management of recruitment and retention. Journal of SMIT, M 2003: Die lampie brand al flouer. Die Burger, 30 January.
Management in Medicine, 16(4):271-291. SMITH, CA; ORGAN, DW & NEAR JP 1983: Organizational
ORGAN, DW 1988: Organizational citizenship behaviour: The good Citizenship Behavior: Its nature and antecedents. Journal of
soldier syndrome. Lexington, MA: Lexington Books. Applied Psychology, 68(4):653-663.
ORGAN, DW & KONOVSKY, M 1989: Cognitive versus affective STEEL, R & OVALLE, N 1984: A review and meta-analysis of
determinants of organizational citizenship behavior. Journal of research on the relationship between behavioural intentions and
Applied Psychology, 74(1):57-164. employee turnover. Journal of Applied Psychology, 69(4):673-
PETKOON, L & ROODT, G 2004: The discriminant validity of the 686.
culture assessment instrument: A comparison of company sub- TEPECI, M & BARTLETT, ALB 2002: The hospitality industry culture
cultures. SA Journal of Industrial Psychology, 30(1):46-54. profile: A measure of individual values, organizational culture, and
PODSAKOFF, PM; MACKENZIE, SB; PAINE, JB & BACHRACH, DG person-organization fit as predictors of job satisfaction and
2000: Organizational citizenship behaviours: A critical review of behavioral intentions. International Journal of Hospitality
the theoretical and empirical literature and suggestions for further Management, 21(2):151-170.
research. Journal of Management, 26(3):513-563. VAN DER POST, WZ; DE CONING, TJ & SMIT, EV 1997: An
ROODT, G 2004a: Concept redundancy and contamination in instrument to measure organisational culture. South African
employee commitment research: Current problems and future Journal of Business Management, 28(4):147-168.
directions. SA Journal of Industrial Psychology, 30(1):82-90. VAN DYNE, L & LEPINE, JA 1998: Helping and voice extra-role
ROODT, G 2004b: Turnover Intentions. Unpublished Document. behaviors: Evidence of construct and predictive validity. Academy
University of Johannesburg: Johannesburg. of Management Journal, 41(1):108-120.
ROODT, G 1997: Theoretical and empirical linkages between work- VUJICIC, M; ZURN, P; DIALLO, K; ADAMS, O & DAL POZ, MR 2004:
related commitment foci. SA Journal of Industrial Psychology, The role of wages in the migration of health care professionals
23(2):6-13. form developing countries. Human Resources for Health. Available
RYU, S; HEE HO, S & HAN, I 2003: Knowledge sharing behavior of from http://www.human-resources-health.com/content/2/1/3 (Ac-
physicians in hospitals. Expert Systems with applications, cessed 19 October 2004).
25(1):113-122. WAGNER, C; CASSIMJEE, N & NEL, H 2002: Some key principles in
SADRI, G & LEES, B 2002: A cultural competitive advantage. Human implementing knowledge management: the views of employees in
Resource Management International Digest, 10(3):10-12. a small software company. SA Journal of Industrial
SAGER, JK; GRIFFETH, RW & HOM, PW 1998: A Comparison of Psychology, 28(2):49-54.
Structural Models Representing Turnover Cognitions. Journal of WELMAN, JC & KRUGER, SJ 1999: Research methodology for the
Vocational Behaviour, 53(2):254-273. business and administrative sciences. Halfway House: Thomson
SALOPEK, JJ & DIXON, NM 2002: Common knowledge: How International.
companies thrive by sharing what they know. Training & YANG, JT & WAN, CS 2004: Advancing organizational
Development, 54(4):63–64. effectiveness and knowledge management implementation.
SCHEPERS, JM 2004: Overcoming the effects of differential Management Accounting Research, 15(1):53-76.
skewness of test items in scale construction, SA Journal of
Industrial Psychology, 30(4):27-43.
SCHEPERS, JM 1992: Toetskonstruksie: Teorie en praktyk.
Johannesburg: RAU Drukpers.
SHIELDS, MA & WARD, M 2001: Improving nurse retention in the
National Health Service in England: the impact of job satisfaction
on intentions to quit. Journal of Health Economics, 20(5):677-
701.
SLATE, RN & VOGEL, RE 1997: Participative management and
correctional personnel: a study of the perceived atmosphere for

HEALTH SA GESONDHEID Vol.13 No.1 - 2008 75

ANNEXURE A: TURNOVER INTENTIONS QUESTIONNAIRE

The following questionnaire measures your intentions to quit.

1. How often have you considered

leaving your current job? Never 1----2----3----4----5----6----7 Most of the times
2. How frequently do you scan news
papers for job opportunities? Never 1----2----3----4----5----6----7 Most of the times
3. To what extent is your current job
not addressing your important To no extent 1----2----3----4----5----6----7 To a large extent
personal needs?
4. How often are opportunities to
achieve your most important goals Never 1----2----3----4----5----6----7 Always
at work jeopardised?
5. How often are your most important
personal values at work Never 1----2----3----4----5----6----7 Always
compromised?
6. How frequently are you day-
dreaming about a different job that Never 1----2----3----4----5----6----7 Always
will suit your personal needs?
7. What is the probability that you will
leave your job, if you get another Low 1----2----3----4----5----6----7 High
suitable offer?
8. How frequently do you look forward
to another day at work? Always 1----2----3----4----5----6----7 Never
9. How often do you think about
starting your own business? Never 1----2----3----4----5----6----7 Most of the times
10. How often do only family
responsibilities preventing you from Never 1----2----3----4----5----6----7 Always
quitting?
11. How often do only vested personal
interest (pension fund, Never 1----2----3----4----5----6----7 Always
unemployment fund, etc.) prevent
you from quitting?
12. How frequently are you emotionally
agitated when arriving home after Never 1----2----3----4----5----6----7 Always
work?
13. How often is your current job
affecting on your personal well- Never 1----2----3----4----5----6----7 Always
being?
14. How often do the troubles
associated with relocating, prevent Never 1----2----3----4----5----6----7 Always
you from quitting?

76 HEALTH SA GESONDHEID Vol.13 No.1 - 2008

ANNEXURE B: KNOWLEDGE SHARING QUESTIONNAIRE
The following questionnaire measures your perceptions about knowledge sharing in your organisation.
To what extent do you share knowledge in this organisation …

1. to get recognition? To no extent 1 2 3 4 5 To a large extent

2. to be rewarded? To no extent 1 2 3 4 5 To a large extent

3. to satisfy your self-fulfillment

needs? To no extent 1 2 3 4 5 To a large extent

4. to support management
strategic objectives? To no extent 1 2 3 4 5 To a large extent

5. to enhance your career? To no extent 1 2 3 4 5 To a large extent

How often do you have the opportunity …

6. to attend training courses? Never 1 2 3 4 5 Most of the times

7. to share your knowledge with

colleagues? Never 1 2 3 4 5 Most of the times

8. to attend informal gatherings

where knowledge is shared? Never 1 2 3 4 5 Most of the times

How much do you agree that knowledge sharing contributes to…

9. the success of this

organisation? Disagree 1 2 3 4 5 Agree

10. the competitiveness of this

organisation? Disagree 1 2 3 4 5 Agree

11. the innovativeness of this

organisation? Disagree 1 2 3 4 5 Agree

To what extent do you experience that others share knowledge due to the following reasons…

12. trust that exists in the

organisation? To no extent 1 2 3 4 5 To a large extent

13. the likelihood that colleagues

will do likewise? To no extent 1 2 3 4 5 To a large extent

14. it is highly valued by

management? To no extent 1 2 3 4 5 To a large extent

15. the organisational culture

facilitates a learning To no extent 1 2 3 4 5 To a large extent
environment?
16. people who share knowledge
are regarded as experts? To no extent 1 2 3 4 5 To a large extent

17. it contributes to positive

performance appraisals? To no extent 1 2 3 4 5 To a large extent

HEALTH SA GESONDHEID Vol.13 No.1 - 2008 77

To what extent you do not readily share knowledge due to the following reasons …

18. you are afraid your career

would be in danger if you To no extent 1 2 3 4 5 To a large extent
make mistakes?(-)
19. not enough trust exists in this
organisation? (-) To no extent 1 2 3 4 5 To a large extent

20. others don’t want to do

likewise? (-) To no extent 1 2 3 4 5 To a large extent

To what extent do you experience that others do not readily share knowledge due to the following reasons

21. they are afraid their careers

would be in danger if they To no extent 1 2 3 4 5 To a large extent
make mistakes? (-)
22. not enough trust exist in this
organisation? (-) To no extent 1 2 3 4 5 To a large extent

23. colleagues don’t want to do

likewise? (-) To no extent 1 2 3 4 5 To a large extent

78 HEALTH SA GESONDHEID Vol.13 No.1 - 2008