You are on page 1of 2

Downloaded from bmj.

com on 24 August 2007

Standard deviations and standard errors


Douglas G Altman and J Martin Bland

BMJ 2005;331;903-
doi:10.1136/bmj.331.7521.903

Updated information and services can be found at:


http://bmj.com/cgi/content/full/331/7521/903

These include:
References This article cites 4 articles, 3 of which can be accessed free at:
http://bmj.com/cgi/content/full/331/7521/903#BIBL

1 online articles that cite this article can be accessed at:


http://bmj.com/cgi/content/full/331/7521/903#otherarticles
Rapid responses 3 rapid responses have been posted to this article, which you can access for
free at:
http://bmj.com/cgi/content/full/331/7521/903#responses

You can respond to this article at:


http://bmj.com/cgi/eletter-submit/331/7521/903
Email alerting Receive free email alerts when new articles cite this article - sign up in the
service box at the top left of the article

Topic collections Articles on similar topics can be found in the following collections

Other Statistics and Research Methods: descriptions (600 articles)

Notes

To order reprints follow the "Request Permissions" link in the navigation box
To subscribe to BMJ go to:
http://resources.bmj.com/bmj/subscribers
Downloaded from bmj.com on 24 August 2007
Education and debate

6 The Cardiac Arrhythmia Suppression Trial I. Preliminary Report. Effect of 16 Beauchamp TL, Childress JF. Respect for autonomy. Principles of biomedical
encainide and flecainide on mortality in a randomized trial of arrhythmia ethics. 4th ed. New York: Oxford University Press, 1994:120-88.
suppression after myocardial infarction. N Engl J Med 1989;321:406-12. 17 Roberts LW. Evidence-based ethics and informed consent in mental
7 Dickert N, Grady C. What’s the price of a research subject? Approaches to illness research. Arch Gen Psychiatry 2000;57:540-2.
payment for research participation. N Engl J Med 1999;341:198-203. 18 Bayer R, Oppenheimer GM. Toward a more democratic medicine: sharing the
8 Grady C. Money for research participation: does it jeopardize informed burden of ignorance. AIDS Doctors: voices from the epidemic. New York: Oxford
consent? Am J Bioethics 2001;1:40-4. University Press, 2000:156-69.
9 Macklin R. “Due” and “undue” inducements: On paying money to 19 Coulter A, Rozansky D. Full engagement in health. BMJ 2004;329:1197-8.
research subjects. IRB: a review of human subjects research 1981;3:1-6. 20 Joffe S, Manocchia M, Weeks JC, Cleary PD. What do patients value in
10 McGee G. Subject to payment? JAMA 1997;278:199-200.
their hospital care? An empirical perspective on autonomy centred
11 McNeil P. Paying people to participate in research. Bioethics
bioethics. J Med Ethics 2003;29:103-8.
1997;11:390-6.
21 Heesen C, Kasper J, Segal J, Kopke S, Muhlhauser I. Decisional role pref-
12 Wilkenson M, Moore A. Inducement in research. Bioethics 1997;11:
373-89. erences, risk knowledge and information interests in patients with multi-
13 Halpern SD, Karlawish JHT, Casarett D, Berlin JA, Asch DA. Empirical ple sclerosis. Mult Scler 2004;10:643-50.
assessment of whether moderate payments are undue or unjust induce- 22 Azoulay E, Pochard F, Chevret S, et al. Half the family members of inten-
ments for participation in clinical trials. Arch Intern Med 2004;164:801-3. sive care unit patients do not want to share in the decision-making proc-
14 Bentley JP, Thacker PG. The influence of risk and monetary payment on ess: a study in 78 French intensive care units. Crit Care Med
the research participation decision making process. J Med Ethics 2004;32:1832-8.
2004;30:293-8. 23 Dunn LB, Gordon NE. Improving informed consent and enhancing
15 Viens AM. Socio-economic status and inducement to participate. Am J recruitment for research by understanding economic behavior. JAMA
Bioethics 2001;1. 2005;293:609-12.

Statistics Notes
Standard deviations and standard errors
Douglas G Altman, J Martin Bland

The terms “standard error” and “standard deviation” example. By contrast the standard deviation will not Cancer Research
UK/NHS Centre
are often confused.1 The contrast between these two tend to change as we increase the size of our sample. for Statistics in
terms reflects the important distinction between data So, if we want to say how widely scattered some Medicine, Wolfson
description and inference, one that all researchers measurements are, we use the standard deviation. If we College, Oxford
OX2 6UD
should appreciate. want to indicate the uncertainty around the estimate of
Douglas G Altman
The standard deviation (often SD) is a measure of the mean measurement, we quote the standard error of professor of statistics
variability. When we calculate the standard deviation of a the mean. The standard error is most useful as a means in medicine

sample, we are using it as an estimate of the variability of of calculating a confidence interval. For a large sample, Department of
a 95% confidence interval is obtained as the values Health Sciences,
the population from which the sample was drawn. For University of York,
data with a normal distribution,2 about 95% of individu- 1.96×SE either side of the mean. We will discuss confi- York YO10 5DD
als will have values within 2 standard deviations of the dence intervals in more detail in a subsequent Statistics J Martin Bland
mean, the other 5% being equally scattered above and Note. The standard error is also used to calculate P val- professor of health
statistics
below these limits. Contrary to popular misconception, ues in many circumstances.
The principle of a sampling distribution applies to Correspondence to:
the standard deviation is a valid measure of variability Prof Altman
other quantities that we may estimate from a sample, doug.altman@
regardless of the distribution. About 95% of observa-
such as a proportion or regression coefficient, and to cancer.org.uk
tions of any distribution usually fall within the 2 standard
contrasts between two samples, such as a risk ratio or
deviation limits, though those outside may all be at one BMJ 2005;331:903
the difference between two means or proportions. All
end. We may choose a different summary statistic, how-
such quantities have uncertainty due to sampling vari-
ever, when data have a skewed distribution.3
ation, and for all such estimates a standard error can be
When we calculate the sample mean we are usually
calculated to indicate the degree of uncertainty.
interested not in the mean of this particular sample, but In many publications a ± sign is used to join the
in the mean for individuals of this type—in statistical standard deviation (SD) or standard error (SE) to an
terms, of the population from which the sample comes. observed mean—for example, 69.4±9.3 kg. That
We usually collect data in order to generalise from them notation gives no indication whether the second figure
and so use the sample mean as an estimate of the mean is the standard deviation or the standard error (or
for the whole population. Now the sample mean will indeed something else). A review of 88 articles
vary from sample to sample; the way this variation published in 2002 found that 12 (14%) failed to
occurs is described by the “sampling distribution” of the identify which measure of dispersion was reported
mean. We can estimate how much sample means will (and three failed to report any measure of variability).4
vary from the standard deviation of this sampling distri- The policy of the BMJ and many other journals is to
bution, which we call the standard error (SE) of the esti- remove ± signs and request authors to indicate clearly
mate of the mean. As the standard error is a type of whether the standard deviation or standard error is
standard deviation, confusion is understandable. being quoted. All journals should follow this practice.
Another way of considering the standard error is as a
Competing interests: None declared.
measure of the precision of the sample mean.
The standard error of the sample mean depends 1 Nagele P. Misuse of standard error of the mean (SEM) when reporting
on both the standard deviation and the sample size, by variability of a sample. A critical evaluation of four anaesthesia journals.
Br J Anaesthesiol 2003;90:514-6.
the simple relation SE = SD/√(sample size). The stand- 2 Altman DG, Bland JM. The normal distribution. BMJ 1995;310:298.
ard error falls as the sample size increases, as the extent 3 Altman DG, Bland JM. Quartiles, quintiles, centiles, and other quantiles.
BMJ 1994;309:996.
of chance variation is reduced—this idea underlies the 4 Olsen CH. Review of the use of statistics in Infection and Immunity. Infect
sample size calculation for a controlled trial, for Immun 2003;71:6689-92.

BMJ VOLUME 331 15 OCTOBER 2005 bmj.com 903

You might also like