BMJ Publishing Group

Statistics Notes: Cronbach's Alpha

Author(s): J. Martin Bland and Douglas G. Altman
Reviewed work(s):
Source: BMJ: British Medical Journal, Vol. 314, No. 7080 (Feb. 22, 1997), p. 572
Published by: BMJ Publishing Group
Stable URL: http://www.jstor.org/stable/25173851 .
Accessed: 27/11/2011 07:41

Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at .
http://www.jstor.org/page/info/about/policies/terms.jsp

JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of
content in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new forms
of scholarship. For more information about JSTOR, please contact support@jstor.org.

Digitization of the British Medical Journal and its forerunners (1840-1996) was completed by the U.S. National
Library of Medicine (NLM) in partnership with The Wellcome Trust and the Joint Information Systems
Committee (JISC) in the UK. This content is also freely available on PubMed Central.

BMJ Publishing Group is collaborating with JSTOR to digitize, preserve and extend access to BMJ: British
Medical Journal.

http://www.jstor.org
General practice

2 RawlesJ. Magnitude of benefit from early thrombolytic treatment in acute
myocardial infarction: new evidence from Grampian region early
anistreplase trial (GREAT).BMJ 1996;312:212-6.
3 British Heart Foundation Working Group. Guidelines for the early man
agement of patients with myocardial infarction. BMJ 1994;308:767-71.
4. Rawles J. On behalf of the GREAT group. Halving of mortality at one
year by domiciliary thrombolysis, inGrampian region early anistreplase
trial (GREAT)./ Am Coll Cardiol 1994;23:1-5.
5 Rawles J, Adams JN. Implications for the Grampian region early
anistreplase trial (GREAT) for pre-hospital coronary care in Scotland.
Health Bulletin 1994;52:106-17.
6 Barradell LB, Goa KL. Alteplase: a pharmacoeconomic evaluation of its
use in the management of myocardial infarction. PharmacoEconomics
1995;8:428-59.
(Accepted 12 December 1996)

Statistics notes
Cronbach's alpha
J Martin Bland, Douglas G Altaian

This is the 28th Many of interest in medicine, such as are all then s^= Lsf and a = 0. Thus a
quantities anxiety independent,
in a series of or
degree of
handicap,
are
impossible
to measure will be 1 if the items are all the same and 0 if none is
occasional notes explicitly. Instead, we ask a series of questions
and related to another.
on medical combine the answers into a
single numerical value. For the mini-HAQ the standard devia
example,
statistics Often this is done by simply adding a score from each tions of each item and the total score are shown in the
answer. For the is a measure of table. We have ls? = 11.16, 4= = 10.
example, mini-HAQ 77.44, and k
for with cervical we
Department of impairment developed patients these into the equation, have
Putting
Public Health myelopathy.1 This has 10 items (table 1) recording the
Sciences, St 10 11.16\ = AftK
degree of difficulty experienced in carrying out daily a =? x (,
V
11?-) 0.95
George's Hospital
Medical School, activities. Each item is scored from 1 (no difficulty) to 4 9 77.447
London SW17 ORE (can't do). The scores on the 10 items are summed to
which indicates a high degree of consistency.
J Martin Bland, give the mini-HAQ score.
For scales which are used as research tools to com
professor ofmedical When items are used to form a scale need to
statistics they a may be less than in the clinical
pare groups, situation,
have internal consistency. The items should all measure
ICRF Medical when the value of the scale for an individual is of inter
the same so be correlated with one
Statistics Group, thing, they should est For a values to 0.8 are
comparing groups, of 0.7
Centre for Statistics another. A useful coefficient for assessing internal con
inMedicine, as For the clinical
is Cronbach's The formula is: regarded satisfactory. application,
Institute of Health sistency alpha.2
much values of a are needed. The minimum is
higher
Sciences, PO Box k / I^\ =
777, Oxford 0.90, and a 0.95, as here, is desirable.
OX3 7LF k-l V s2T
'
In a recent
example, McKinley
et al devised a
ques
Douglas G Altman, where k is the number tionnaire to measure satisfaction with calls
head of items, s] is the variance of the patient
?th item and of the total score made out of hours.3 This
s? is the variance formed by general practitioners
Correspondence to:
all the items. If the items are not included separate scores, which
Professor Bland by summing simply eight they interpreted
added tomake the score, but first multiplied by weight as
measuring
constructs such as satisfaction with com

BMJ 1997,314:572 ing coefficients, we multiply the item by its coefficient munication and management, satisfaction with doc
before the variance s2. Clearly, we must have tor's attitude, etc. a for each score, ranging
calculating They quoted
at least two items?that is k>l, or a will be undefined. from 0.61 to 0.88. conclude that the question
They
The coefficient works because the variance of the naire has internal as five of the
satisfactory validity,
sum of a group of variables is the sum of scores had a > 0.7. In this issue Bosma et al report
independent eight
their variances. If the variables are correlated, similar from 0.67 to 0.84, for assessments of
positively values,
the variance of the sum will be increased. If the items three characteristics of the work environment4
the score are all identical and so perfectly a direct
making up Cronbach's has The
alpha interpretation.
correlated, all the s2 will be and s^ = k2 s2, so that
equal items in our test are some of the many
only possible
= l/k and a = 1.On the other hand, if the items
Xs2/sy items which could be used to make the total score. If we
were to choose two random of k o? these
samples pos
sible items, we would have two different scores each
Table 1 Mini-HAQscale in249 severely impairedsubjects
made up of k items. The correlation between
expected
Mean score
Item SD of score s, scores is a.
these
Stand 2.96 1.04
Get out of
bed_^57_1.11
Cut meat 2.91 1.12 1 Casey ATH, Crockard HA, Bland JM, Stevens J, Moskovich R, Ransford
AO. Development of a functional scoring system for rheumatoid arthritis
Hold cup 2.41
1.06
patients with cervical myelopathy Ann Rheum Dis (in press).
Walk_2^_1.04 2 Cronbach LJ. Coefficient alpha and the internal structure of tests.
Climb stairs 3.06 1.04 Psychometrika 1951;16:297-333.
3 McKinley RK, Manku-Scott T, Hastings AM, French DP, Baker R.
Wash 325~ ?.01
Reliability and validity of a new measure of patient satisfaction with out of
Use hours primary medical care in the United Kingdom: development of a
toilet_^59_1.09
Open a jar_2JS6_1.02 patient questionnaire. BMJ 1997;314:193-8.
Enter/leave car 1.03
2.80 4 Bosma H, Marmot MG, Hemingway H, Nicholson AC, Brunner E, Stans
= 8.80 field SA. Low job control and risk of coronary heart disease inWhitehall
Mini-HAQ 28.06 sT II (prospective cohort) study.BMJ 1997;314:558-65.

572 BMJ VOLUME 314 22 FEBRUARY 1997