Вы находитесь на странице: 1из 1

Research update

Validity and reliability Whats it all about?


Part 2 Reliability in quantitative studies

This is one of a
series of short
papers on aspects
of research by
Alison Twycross
and Linda Shields

In the last update (November 2004) we considered

may not weigh the baby in exactly the same way, or may

validity in quantitative studies: whether a tool meas-

read the scales differently (Alsop-Shields and Alexander

ures what it sets out to measure. Reliability refers to

1997).

the consistency, stability and repeatability of results.

Measures of reliability are summarised in the table

Testing for reliability ensures that consistent results

below and, as with consideration of the validity of a study,

would be obtained in identical situations on different

you would look for evidence that the researcher has con-

occasions. It is possible for a measure to be reliable

sidered reliability of the measurement tools used in the


methodology section of a research report PN

but not valid.


The example often cited to explain the difference
between validity and reliability is that of a clock that is
five minutes fast. The clock is reliable because it is always

REFERENCES

five minutes fast. However, it is not valid because it does


not conform to a recognised standard such as Greenwich
Mean Time.
To give a clinical example, the weight of an infant on
one set of scales may not be the same on a different set
of scales the scales themselves are reliable, but set to
a different standard. Or, if one nurse weighs the baby
on the same set of scales as another nurse, the resulting weights may not be reliable as two different people

Alsop-Shields L Alexander H (1997) A study of errors that can


occur when weighing infants. Journal of Advanced Nursing. 25,
587-594.
Carter DE, Porter S (2000) Validity and reliability. In Cormack
D (ed) The Research Process in Nursing. 4th edition. Oxford,
Blackwell Science, 29-42.
Knapp TR (1998) Quantitative Nursing Research. Sage,
Thousand Oaks.
Peat J (2002) Health Services Research: A Handbook of
Quantitative Methods. London, Sage.

Alison Twycross RGN, RMN,

Table 1. Measures to assess the reliability of data collection tools

RSCN, MSc, CertEd (HE), DMS,

(adapted from Knapp 1998, Carter and Porter 2000, Peat 2002)

Senior Lecturer in Childrens


Nursing/Senior Nurse (Research
Development), Glasgow

Type of reliability

Characteristics

Stability

administrations

Caledonian University/Yorkhill

Hospital
Linda Shields PhD, FRCNA,
Professor of Nursing, University

An instrument is thought to be stable if the same results are obtained on repeated


Assessed using the test-retest technique that is by administering the same
research tool to the same people on two or more occasions

Homogeneity

of Limerick, Ireland

A reliability co-efficient provides an indication of how reliable the tool is

A measure of the internal consistency of the scales

Concerned with the extent that the items in a tool measure the variable being
investigated

Assessed using the split half reliability technique

Cronbach's alpha (a correlation co-efficient) is calculated to establish how reliable


the tool is

Equivalence

Refers to the level of agreement among researchers using the same data
collection tool the inter-rater reliability

The ratings of the two or more researchers are compared by calculating a

TAXI

correlation co-efficient

36

Paediatric Nursing

vol 16 no 10 December 2004

Вам также может понравиться