U10D1-68 - READ AND FOLLOW ALL DIRECTIONS BELOW AND READ ATTACHMENTS...VERY IMPORTANT TO HOMEWORK...Credibility, Dependability, and Transferability

O R I G I N A L P A P E R

The strengths and weaknesses of research designs involving quantitative measures

Wendy Walker MSc Health Studies; Post Graduate Diploma in Adult Education, BSc(Hons) Nursing Studies, Diploma in Professional Studies in Nursing

Senior Research Fellow/Senior Lecturer

Faculty of Health and Sciences, Staffordshire University

Abstract This paper presents a critical review of the strengths and weaknesses of research designs involving quantitative measures and, in particular,

experimental research. The review evolved during the planning stage of a PhD

project that sought to determine the effects of witnessed resuscitation on

bereaved relatives. The discussion is therefore supported throughout by reference

to bereavement research. Three levels of quantitative research are presented:

descriptive, correlational and experimental. The findings suggest that

experimental research is subject to a number of methodological limitations that

may jeopardise internal and external validity of the research results and,

consequently, limit their applicability for practice. Nurses are therefore

encouraged to carefully consider the virtues of experimental designs, in their

quest for evidence-based practice and in the planning of future research.

Key words evidence-based practice, experimental research, positivism, bereavement, witnessed resuscitation

Introduction

In today’s political climate of demand for cost and clinical effectiveness

(NHSE, 1996; Department of Health, 1997), nurses are increasingly

expected to engage in evidence-based practice (EPB). The widespread

movement to promote EBP represents a concerted effort to progress away

from procedures based on tradition, ritual and routine, to a clinical

service informed at every level by evidence that is scientifically derived

(Ford and Walsh, 1994; Hicks, 1997; Hicks and Hennessy, 1997). Con-

sequently, ‘the perceived importance and profile of research has never

571

Journal of Research in Nursing

© 2005

SAGE PUBLICATIONS

London, Thousand Oaks and

New Delhi

VOL 10(5) 571–582

ISSN 1744-9871

jr n

been higher and the expectation that research will deliver solutions to

problems has never been greater’ (Balcombe, 1996: 1206). Notable

health service research and development (R&D) initiatives include:

funding for research to inform policy and practice (Salvage, 1998), strat-

egies for the dissemination and integration of research such as the

Cochrane Collaboration (Lefebvre, 1994) and the establishment of a

National Institute of Clinical Excellence as a means of giving ‘new coher-

ence and prominence to information about clinical and cost effectiveness’

(Department of Health, 1997: 3). According to Mulhall et al. (1998),

emphasis on effectiveness and efficiency within the health service research

and development programme, including discussions on the development

of guidelines, give the highest regard to research designs that are less sus-

ceptible to bias. The quantitative approach to research is therefore likely

to be embraced by those who support this point of view; being founded

on the belief that the social world lends itself to objective forms of mea-

surement (Cowman, 1993) and characterised by a set of orderly and dis-

ciplined procedures (Polit and Hungler, 1999).

Defining quantitative research

Quantitative research is depicted as the traditional scientific approach to

research that has its underpinnings in the philosophical paradigm for

human inquiry known as positivism (Polit and Hungler, 1999). Research

driven by the positivist tradition is a ‘systematic and methodological

process’ (Koch and Harrington, 1998: 884) that places considerable value

on ‘rationality, objectivity, prediction and control’ (Streubert and Carpen-

ter, 1999: 7). A distinguishing feature is the collection of numerical data

(Jack and Clarke, 1998) that, in turn, can be subjected to statistical analy-

sis (Carter, 2000a). Advocates of the quantitative approach are therefore

described as objective scientists (Duffy, 1986) committed to the discovery

of quantifiable information (Carr, 1994). Parahoo (1997) identifies three

levels of quantitative research: descriptive, correlational and causal; causal

referring to experiment as a research design.

Descriptive research Descriptive research provides an account of the characteristics of indi-

viduals, groups or situations (Jack and Clarke, 1998) that may form the

first stage of more complex designs (Clifford, 1997; Carter, 2000b). The

overall aim is to ‘discover new meaning, describe what exists, determine

the frequency with which something occurs and categorize information’

(Burns and Grove, 1999: 24). This is illustrated by Fraser and Atkins

(1990) who carried out a telephone survey to identify survivors’ recollec-

Journal of Research in Nursing 10(5)

572

tions of emergency nurse activities following the sudden death of a loved

one. Using descriptive statistics, the authors reported on the frequency

and percentage of nurse activities that either ‘helped’ or ‘would have

helped’ survivors cope in this crisis situation. Recommendations for

future research included expansion of the study to determine whether

there is a relationship between the demographics of victims and survivors’

recollections of helpful and unhelpful nurse activities.

Correlational research In correlational research, the investigator deliberately seeks to examine

links (or relationships) between variables without introducing an inter-

vention. The purpose is often to generate hypotheses that can be tested in

experimental research (Parahoo, 1997; Burns and Grove, 1999). Evidence

of this is seen in the study by Tye (1993). Data collected by means of a

structured questionnaire were used to examine the relationship between

age, length of professional experience and death education and their effect

on qualified nurses’ perceptions of the needs of suddenly bereaved family

members. One of the implications arising from this research was the need

for education and training relevant to the needs of the suddenly bereaved.

This lends itself to the application of an experimental design to determine

its effects on staff perceptions.

Experimental research Experimental research provides the framework for establishing a relation-

ship between cause and effect (Roe, 1994, Mulhall, 1994). In experi-

ments, the researcher as an active agent (Polit et al., 2001) uses deductive

reasoning to prove or falsify hypotheses (Proctor, 1998). This involves

manipulating an independent variable (cause) and observing the outcome

on a dependent variable (effect) whilst attempting to hold extraneous

variables constant (Newell, 1994). Similarity of subjects is ensured by

‘matching’ cases with respect to an infinite number of characteristics and

allocating one from each pair to a control and experimental group on the

basis of randomisation (Clifford, 1997; Hicks, 1998; Polit et al., 2001).

Only observable facts are relevant and the techniques of inferential stat-

istics produce precise numerical results (Hicks and Hennessy, 1997). A

variety of experimental designs have been developed, ranging from the

relatively simple before–after design, to the more complex multivariate

factorial designs (Burns and Grove, 1999). According to Polit and

Hungler (1999), random sampling, manipulation and control are the

characteristics of ‘true’ experimental research. An approach to research

that embraces these three virtues is the highly valued randomised con-

trolled trial (RCT).

Walker The strengths and weaknesses of research designs involving quantitative measures

573

Robinson et al. (1998) carried out an RCT to determine the psycholog-

ical effect (dependent variable) of witnessed resuscitation (independent

variable) on bereaved relatives. Using manipulation, relatives of patients

who required resuscitation were given the option to remain with the

patient (experiment group) or were not given this choice (control

group). The unit of randomisation was the patient undergoing resuscita-

tion. Eligible resuscitations were defined and a chaperone was provided to

give emotional support and technical information, adding control to the

research design. Outcome measures included psychiatric and psychologi-

cal morbidity at one and six months after the resuscitation event.

Strengths of experimental research

Experimental research is regarded by many as the optimum quantitative

methodology for obtaining reliable information about treatment or inter-

vention effect (McMahon, 1994; Mulhall, 1994; Sibbald and Roland,

1998; Donnan, 2000; Richardson, 2000; Polit et al., 2001). Moreover,

the randomised controlled trial holds a superior status over other research

methods as the ‘gold standard’ of evidence on which to base decisions

about healthcare (Knipschild, 1993; Black, 1996; May, 1997). From a

review of this literature, it is readily apparent that the power and strength

of experimental research is related to control. This involves strict applica-

tion of standardised procedures to reduce systematic bias and eliminate

erroneous conclusions (Hicks, 1998; Burns and Grove, 1999). Control

may be exerted in several ways, including: random sampling,

inclusion/exclusion criteria, use of a comparison group, subject matching

across groups, manipulation of the independent variable, single, double

or treble blinding procedures, the use of precise measuring tools and the

application of standardised statistical tests in the final analysis of data. The

dividend of this control is the researcher’s ability to state with confidence

that the outcome produced can only be attributed to the effects of the

experiment (Duffy, 1985). This is in stark contrast to descriptive and cor-

relational research, where less rigid approaches to explore and describe

phenomena (as it exits) limits the extent to which firm conclusions can

be drawn. In other words, the scientific and statistical rigour of experi-

mental research maximises internal validity and increases the probability

of generalising the findings beyond the study sample. Despite such excep-

tional advantages, the literature also points to several methodological

limitations that may jeopardise internal and external validity of the

research results and, consequently, limit their applicability for practice.

Journal of Research in Nursing 10(5)

574

Threats to internal and external validity

Sampling The sampling technique of experimental research relies on the develop-

ment of explicit criteria prior to initiation of the study (Duffy, 1985). To

qualify as a true experiment, the researcher is obliged to select a sample

from the study population and allocate subjects to the various study

groups on the basis of randomisation (Duffy, 1985; McMahon, 1994;

Clifford, 1997; Polit et al., 2001). Problems may arise, however, in ran-

domised controlled trials when potential participants are not prepared to

opt for treatment on a random basis (Brewin and Bradley, 1989; Black,

1996; Silverman and Altman, 1996; Torgerson and Sibbald, 1998).

Failure to achieve randomisation may limit the extent to which the study

sample is representative of the parent population and, with it, generalis-

ability of the study findings (Torgerson and Sibbald, 1998). Supporting

this argument is the study by Stroebe and Stroebe (1989), who identified

a range of what they call ‘accepter and refuser characteristics’ that could

limit generalisations concerning health and recovery patterns among

bereaved people. A further limitation is that the researcher may be unable

to match patients with respect to certain physical, psychological and social

traits (Hicks, 1998), all of which may have an influence on the outcomes

of the research.

Recruitment Successful random sampling also depends on a sufficiently large sample

(Thompson, 1999). Difficulty in recruiting subjects to participate in a

clinical trial is, however, remarkably common (McMahon, 1994). An

important issue identified by Wilson and Rose (1998) is the role of ‘gate-

keepers’ in the process of recruitment. For example, once an intervention

or treatment becomes widespread, or in the absence of any preliminary

evidence to suggest the intervention or treatment may be beneficial, it

may be difficult to recruit participants or clinicians who are prepared to

test alternatives (Black, 1996; Fairhurst and Dowrick, 1996; Getliffe,

1998; Sibbald and Roland, 1998). The latter could certainly apply to the

phenomenon of witnessed resuscitation. Limited empirical evidence is

available to guide decision-making in practice, and support for this inter-

vention amongst healthcare professionals is not yet universal (RCN,

2002). Recruitment difficulties may be overcome by carrying out the

study in multiple geographical locations (Burns and Grove, 1999),

although Getliffe (1998) warns of their complexity and calls for stringent

monitoring by the project management team. Power analysis can also be

used to calculate and guide sample size. However, its use relies on having

some estimation of the degree of change expected in the dependent

Walker The strengths and weaknesses of research designs involving quantitative measures

575

variable and is therefore limited to studies where research on the subject

already exists (McMahon, 1994).

Mortality, maturation and history Despite application of systematic and protocolised procedures, experimenta-

tion with humans is subject to a number of external influences that may

dilute the study results. Once enrolled on a study, subjects may fail to

comply with treatment or follow-up arrangements (Donnan, 2000), particu-

larly if the study is a longitudinal research design (Watson, 1998). Mortality

or attrition raises serious doubts about observed effects due to differences in

the characteristics between the subjects who ‘drop out’ and those who

remain in the study (Nieswiadomy, 1998). Changes within subjects rather

than as a response to a treatment or intervention (maturation) may also have

an intense effect on the study results (Haughey, 1994a; Oldham, 1994;

Nieswiadomy, 1998). For example, a researcher may credit reduced symp-

toms of depression among bereaved relatives to the intervention of witness-

ing resuscitation, whereas changes in mental health may have resulted from

personal adaptation to bereavement and loss, in spite of the intervention.

Staying with this example, relatives may be exposed to a media campaign

during the course of a study that portrays the stages of grief and gives insight

into therapeutic interventions that seek to enhance mental health. When an

event other than the experimental treatment or intervention influences the

dependent variable, the study is said to be threatened by history (Haughey,

1994a; Oldham, 1994; Nieswiadomy, 1998).

Hawthorne effect A further limitation of experimental research is that subjects may change

their behaviour or respond in a specific manner simply because of aware-

ness of being observed (Haughey, 1994a; Clifford, 1997). It is for this

reason that the researcher may adopt what is known as a ‘double-blind’

technique in which neither the participants nor the experimenter knows

which subjects receive the active treatment or intervention (Oldham,

1994). This procedure is viewed as a major strength in RCTs to avoid

experimenter and subject bias (Hicks, 1998). Its use however, is restricted

to studies where treatment or intervention can be disguised, for instance

in clinical drug trials (Polit et al., 2001). Consider, for example, research

to evaluate the effects of witnessed resuscitation. The relative as a

‘witness’, i.e. ‘a person present; one who sees or hears what happens’

(Oxford Current English Dictionary, 1990) cannot be blinded to the inter-

vention. It also follows that those performing the intervention cannot be

blinded for the same reasons. Clearly, this illustrates how efforts to

strengthen external validity in experimental research are not always pos-

sible due to practical, professional or ethical reasons.

Journal of Research in Nursing 10(5)

576

Reductionism Experimental research relies heavily on the control and removal of conta-

minating variables within the internal structure of the study and is, there-

fore, by design, considered to be reliable (Duffy, 1985). However, in

striving to attain consistent internal validity, the research location may

become so false that threats to external validity increase (Haughey,

1994a). Evidence of this is seen in the study by Baker et al. (2000) who

applied a total of 9 inclusion and 11 exclusion criteria in an attempt to

meet the RCT’s demand for homogenous groups. However, the authors

concluded that their results on family satisfaction with end-of-life care

were not generalisable to the experience of families who did not meet

their study criteria for enrolment. It is also argued that the inherent reduc-

tionism of experimental research is incongruent with the humanistic

philosophy espoused for nursing practice (Munhall, 1982; Playle, 1995;

Clark, 1998; Leininger, 1998) and that certain aspects of patient care are

not amenable to manipulation, easily expressed or measurable (Corner,

1991; Parahoo, 1997; Mulhall et al., 1998; Polit et al., 2001). A case in

point is the effects of widowhood on physical and psychological health

(Stroebe and Stroebe, 1987). Clearly, we cannot manipulate widowhood.

‘People lose their spouses by a process that is neither random nor subject

to research control’ (Polit et al., 2001: 178).

Measurement effects For results to be meaningful, it is essential that reliable and validated

outcome measures are used (Getliffe, 1998). In relation to bereavement

research, Tomita and Kitamura (2002) identify 16 measurement devices

available for the assessment of grief. However, their ability to differentiate

between normal and pathological grief is questioned. Furthermore, while

some instruments assess grief in general, others seek to measure grief in

specific situations such as reactions to loss induced by sudden death (as

may be the case for relatives witnessing a resuscitation event) or after mis-

carriage or foetal death. This therefore suggests that reliability and validity

of research results also depend on careful assessment of the circumstances

in which outcome measures will be applied.

Drawing upon the work of Cook and Campbell (1979), Burns and

Grove (1999) also stress the importance of statistical conclusion validity

in order to prevent distortions of the truth. According to Haughey

(1994a), this type of validity can be threatened, for example if the crite-

rion of statistical tests is violated. However, despite the importance placed

on statistical significance, this may not represent the clinical or practical

importance of the research results (LeFort, 1993; Hollis, 1994; Hicks,

1998). For example, research findings may indicate overall satisfaction

with end-of-life care, but this may mean very little in terms of the real

Walker The strengths and weaknesses of research designs involving quantitative measures

577

needs of dying patients and their families (Hanson et al., 1997). Con-

versely, Richardson (2000) argues that many RCTs will produce only

moderate differences in outcome, but this difference may be of clinical

importance either to the patient or the degree to which study results lead

to clinical changes in behaviour.

Ethical issues

Experimental research is subject to a number of ethical considerations,

particularly when a double-blind trial is being developed (Clifford, 1997).

Apart from the element of deception that is involved, participants allocated

to the control group may be considerably disadvantaged (Hicks, 1998)

especially when the outcome of treatment or intervention is uncertain or

believed to be inferior to existing treatment regimes (Sibbald and Roland,

1998). On the other hand, where uncertainty exists about the effectiveness

of current treatment or interventions, Hicks (1998: 22) argues that ‘it is

self-evidently unethical not to subject these to rigorous RCT scrutiny, since

without this, potentially harmful procedures may persist with all the negat-

ive consequences this may have for patient well-being’. There are also

ethical problems associated with leaving some people untreated (Abbott

and Sapsford, 1998). Presume, for example, that a researcher was inter-

ested in evaluating the effectiveness of a witnessed resuscitation protocol. It

would certainly be unethical to withhold interventions for relatives in the

control group who asked to remain with their loved one during the resus-

citation process. This therefore requires measures to ensure that this group

receives the intervention, irrespective of randomisation. Similarly, steps

need to be taken to ensure that subjects in the control group receive the

‘normal’ or routine intervention (Haughey, 1994b). An ensuing dilemma,

however, is that some participants may have received a less-than-beneficial

treatment, when at the end of an experiment the outcome of one treatment

or intervention is found to be significantly more effective (McMahon,

1994). Newell (1992) also has misgivings about patient participation in a

trial that is insufficient in size to detect a better treatment, particularly as

experimentation often involves some element of risk, including adverse

effects that may only manifest in the longer term (Black, 1996; Hicks,

1998). When considering the effect that witnessing resuscitation may have

on the grieving process, Offord (1998) and Fulbrook (1998) suggest that

it is not beyond the realms of possibility for relatives to suffer from

nervous shock or psychological injury and concerns for the medico-legal

implications of this practice have been voiced (RCN, 2002). The ethics of

experimental research therefore demands careful assessment of the risks

and benefits that might be incurred and that this information is clearly

articulated to prospective participants during the process of obtaining

Journal of Research in Nursing 10(5)

578

informed consent. A general guideline is that ‘the degree of risk to be

taken by those participating in the research should never exceed the poten-

tial humanitarian benefits of the knowledge to be gained’ (Polit and

Hungler, 1999: 135).

Conclusion

Central to current healthcare provision is the call for nursing practice to

be evidence-based, with the underlying assumption that patient care will

be enhanced (Hunt, 1996; Hicks, 1998). Taking into account the facts

and arguments presented in this paper, it seems reasonable to suggest that

research designs involving quantitative measures can make a valuable con-

tribution to the evidence-base required. Both descriptive and correlational

designs have a key role to play in the development of new knowledge,

generating questions and hypotheses that could form the basis of further

research. However, at a time of increased demand for treatment and inter-

ventions that produce positive outcomes on health, experimental research

and in particular the RCT is seen as the hallmark of scientific enquiry, in

that it proffers the possibility of predicting cause-and-effect relations

beyond reasonable doubt. Yet despite the application of rigorous proce-

dures, including measures to control systematic error and bias, the use of

experimental research is subject to a number of methodological and

ethical concerns. Nurses are therefore encouraged to carefully consider

the virtues of experimental designs in their quest for evidence-based prac-

tice and in the planning of future research.

Walker The strengths and weaknesses of research designs involving quantitative measures

579

Key points

• Central to current healthcare provision is the call for nursing prac-

tice to be evidence-based

• Quantitative research is depicted as the traditional scientific

approach to research

• Research designs for quantitative studies include descriptive, cor-

relational and experimental

• The randomised controlled trial is seen as the ‘gold standard’ of

evidence on which to base decisions about healthcare

• Despite exceptional advantages of experimental research, nurses

are encouraged to carefully consider its virtues

References Abbott, P., Sapsford, R. (1998) Research Methods for Nurses and the Caring Professions. (2nd

edn). Buckingham: Open University Press.

Baker, R., Wu, A.W., Teno, J.M. et al. (2000) Family satisfaction with end-of-life

care in seriously ill hospitalized adults. Journal of American Geriatric Society 4488:5, 61–69.

Balcombe, K.P. (1996) Effects of the purchaser/provider split on research

implementation. British Journal of Nursing 55:19, 1206–1209.

Black, N. (1996) Why we need observational studies to evaluate the effectiveness of

health care. British Medical Journal 331122: 1215–1218.

Brewin, C.R., Bradley, C. (1989) Patient preferences and randomised clinical trials.

British Medical Journal 229999: 313–315.

Burns, N., Grove, S.K. (1999) Understanding Nursing Research. (2nd edn). Philadelphia:

W.B. Saunders Company.

Carr, L.T. (1994) The strengths and weaknesses of quantitative and qualitative

research: what method for nursing? Journal of Advanced Nursing 2200: 716–721.

Carter, D.E. (2000a) Quantitative research. In: Cormack, D. (ed.) The Research Process in

Nursing (4th ed.). Oxford: Blackwell Science Ltd.

Carter, D.E. (2000b) Descriptive research. In: Cormack, D. (ed.) The Research Process in

Nursing (4th ed.). Oxford: Blackwell Science Ltd.

Clark, A.M. (1998) The qualitative–quantitative debate: moving from positivism and

confrontation to post-positivism and reconciliation. Journal of Advanced Nursing 2277:

1242–1249.

Clifford, C. (1997) Nursing and Health Care Research: A Skills-Based Introduction (2nd edn).

London: Prentice Hall.

Cook, T.D., Campbell, D.T. (1979) Quasi-experimentation: Design and Analysis Issues for Field

Settings. Chicago: Rand McNally.

Corner, J. (1991) In search of more complete answers to research questions.

Quantitative versus qualitative research methods: is there a way forward? Journal of

Advanced Nursing 1166: 718–727.

Cowman, S. (1993) Triangulation: a means of reconciliation in nursing research.

Journal of Advanced Nursing 1188: 788–792.

Department of Health (1997) The New NHS: Modern, Dependable. London: The Stationery Office.

Donnan, P.T. (2000) Experimental research. In: Cormack, D. (ed.) The Research Process

in Nursing (4th ed.). Oxford: Blackwell Science Ltd.

Duffy, M.E. (1985) Designing nursing research: the qualitative–quantitative debate.

Journal of Advanced Nursing 1100: 225–232.

Duffy, M.E. (1986) Quantitative and qualitative research: antagonistic or

complimentary? Nursing and Health Care 88:6, 356–357.

Fairhurst, K., Dowrick, C. (1996) Problems with recruitment in a randomised

controlled trial of counselling in general practice: causes and implications. Journal of

Health Services Research Policy 11:2, 77–80.

Ford, P., Walsh, M. (1994) New Rituals for Old: Nursing Through the Looking Glass. Oxford:

Butterworth-Heinemann Ltd.

Fraser, S., Atkins, J. (1990) Survivors’ recollections of helpful and unhelpful

emergency nurse activities surrounding sudden death of a loved one. Journal of

Emergency Nursing 1166:1, 13–16.

Fulbrook, S. (1998) Legal implications of relatives witnessing resuscitation. British

Journal of Theatre Nursing 77:10, 33–35.

Getliffe, K. (1998) Developing a protocol for a randomised controlled trial: factors to

consider. Nurse Researcher 66:1, 5–17.

Journal of Research in Nursing 10(5)

580

Hanson, L.C., Danis, M., Garrett, J. (1997) What is wrong with end-of-life care?

Opinions of bereaved family members. Journal of American Geriatrics Society 4455:11,

1339–1344.

Haughey, B.P. (1994a) Evaluating quantitative research designs: part 2. Critical Care

Nurse December: 169–72.

Haughey, B.P. (1994b) Evaluating quantitative research designs: part 1. Critical Care

Nurse October: 100–102.

Hawkins, J.M. (1990) Oxford Current English Dictionary. Oxford: Oxford University Press.

Hicks, C. (1997) The dilemma of incorporating research into clinical practice. British

Journal of Nursing 66:9, 511–515.

Hicks, C. (1998) The randomised controlled trial: a critique. Nurse Researcher 66:1, 19–32.

Hicks, C., Hennessy, D. (1997) Mixed messages in nursing research: their

contribution to the persisting hiatus between evidence and practice. Journal of Advanced

Nursing 2255: 595–601.

Hollis, S. (1994) Statistical analysis: not just ‘p < 0.05’. Nurse Researcher 11:4, 48–67.

Hunt, J.M. (1996) Barriers to research utilization (Guest editorial). Journal of Advanced

Nursing 2233: 423–425.

Jack, B., Clarke, A.M. (1998) The value of quantitative research in nursing. Professional

Nurse 1133:1, 753–756.

Knipschild, P. (1993) Trials and errors. British Medical Journal 330066: 1706–1707.

Koch, T., Harrington, A. (1998) Reconceptualizing rigour: the case for reflexivity.

Journal of Advanced Nursing 2288:4, 882–890.

Lefebvre, C. (1994) The Cochrane collaboration: the role of the UK Cochrane Centre

in identifying the evidence. Health Libraries Review 1111: 235–242.

LeFort, S.M. (1993) The statistical versus clinical significance debate. Journal of Nursing

Scholarship 2255:1, 57–62.

Leininger, M.M. (1998) Nature, rationale and importance of qualitative research

methods in nursing. In: Leininger, M.M. (ed.) Qualitative Research Methods in Nursing.

USA: Greyden Press.

McMahon, R. (1994) Trial and error: an experiment in practice. In: Buckeldee, J.,

McMahon, R. (eds) The Research Experience in Nursing. London: Chapman & Hall.

May, A. (1997) Tried and tested remedies. Health Service Journal 110077:5540, 22.

Mulhall, A. (1994) The experimental approach and randomized controlled trials. In:

Hardey, M., Mulhall, A. Nursing Research: Theory and Practice. London: Chapman & Hall.

Mulhall, A., Alexander, C., le May, A. (1998) Appraising the evidence for practice:

what do nurses need? Journal of Clinical Effectiveness 33:2, 54–58.

Munhall, P.L. (1982) Nursing philosophy and nursing research: in apposition or

opposition? Nursing Research 3311:176–177, 181.

National Health Service Executive (1996) Promoting Clinical Effectiveness: A Framework for

Action in and Through the NHS. Leeds: NHSE.

Newell, D.J. (1992) Randomised controlled trials in health care research. In: Daly, J.,

McDonald, I., Willis, E. (eds) Researching Health Care: Designs, Dilemmas, Disciplines.

Tavistock: Routledge.

Newell, R. (1994) Defining variables and hypotheses. Nurse Researcher 11:4, 37–47.

Nieswiadomy, R.M. (1998) Foundations of Nursing Research. (3rd ed). Stamford: Appleton

& Lange.

Offord, R.J. (1998) Should relatives of patients with cardiac arrest be invited to be present

during cardiopulmonary resuscitation? Intensive and Critical Care Nursing 1144: 288–293.

Oldham, J. (1994) Experimental and quasi-experimental research designs. Nurse

Researcher 11:4, 26–36.

Walker The strengths and weaknesses of research designs involving quantitative measures

581

Parahoo, K. (1997) Nursing Research: Principles, Process, Issues. London: Macmillan.

Playle, J.F. (1995) Humanism and positivism in nursing: contradictions and

conflicts. Journal of Advanced Nursing 2222: 979–984.

Polit, D.F., Hungler, B.P. (1999) Nursing Research: Principles and Methods (6th edn).

Philadelphia: J.B. Lippincott.

Polit, D.F., Beck, C.T., Hungler, B.P. (2001) Essentials of Nursing Research: Methods, Appraisal,

and Utilisation (5th edn). Philadelphia: Lippincott.

Proctor, S. (1998) Linking philosophy and method in the research process: the case

for realism. Nurse Researcher 55:4, 73–89.

Richardson, J. (2000) The use of randomized control trials in complementary

therapies: exploring the issues. Journal of Advanced Nursing 3322:2, 398–406.

Robinson, S.M., Mackenzie-Ross, S., Campbell Hewson, G.L. et al. (1998)

Psychological effect of witnessed resuscitation on bereaved relatives. The Lancet 335522:

614–617.

Roe, B. (1994) Is there a place for the experiment in nursing research? Nurse Researcher

11:4, 4–12.

Royal College of Nursing (2002) Witnessing Resuscitation: Guidance for Nursing Staff. London:

RCN.

Salvage, J. (1998) Evidence-based practice: a mixture of motives. NT Research 33:6,

406–418.

Sibbald, B., Roland, M. (1998) Why are randomised controlled trials important?

British Medical Journal 331166: 201.

Silverman, W.A., Altman, D.G. (1996) Patients’ preferences and randomised trials.

Lancet 334477: 171–174.

Streubert, H.J., Carpenter, D.R. (1999) Qualitative Research in Nursing: Advancing the

Humanistic Imperative (2nd edn). Philadelphia: Lippincott.

Stroebe, M., Stroebe, W. (1989) Who participates in bereavement research? A review

and empirical study. Omega 2200:1, 1–29.

Stroebe, W., Stroebe, M.S. (1987) Bereavement and Health: the Physical and Psychological

Consequences of Partner Loss. London: Cambridge University Press.

Thompson, C. (1999) If you could just provide me with a sample: examining sampling

in qualitative and quantitative research papers. Evidence-Based Nursing 22:3, 68–70.

Tomita, T., Kitamura, T. (2002) Clinical research measures of grief: a

reconsideration. Comprehensive Psychiatry 4433:2, 95–102.

Torgerson, D.J., Sibbald, B. (1998) What is a patient preference trial? British Medical

Journal 331166: 360.

Tye, C. (1993) Qualified nurses’ perceptions of the needs of suddenly bereaved

family members in the accident and emergency department. Journal of Advanced Nursing

1188: 948–956.

Watson, R. (1998) Longitudinal quantitative research designs. Nurse Researcher 55:4,

41–53.

Wilson, K., Rose, K. (1998) Patient recruitment and retention strategies in

randomised controlled trials Nurse Researcher 66:1, 35–46.

Correspondence should be addressed to:

Wendy Walker,

Faculty of Health and Sciences,

Staffordshire University,

Blackheath Lane,

Stafford

Journal of Research in Nursing 10(5)

582