Information in practice
Ethical issues in qualitative research on internet
communities
Gunther Eysenbach, James E Till
The internet is the most comprehensive electronic
archive of written material representing our world and
peoples’ opinions, concerns, and desires. Physicians
who surf the internet for the first time are often
stunned by what they learn on websites set up by lay
people or patient self support communities. Material
on these venues can be a rich source for researchers
interested in understanding the experiences and views
of people and patients. Qualitative analysis of material
published and communicated on the internet can
serve to systematise and codify needs, values, concerns,
and preferences of consumers and professionals
relevant to health and health care. While the internet
makes people’s interactions uniquely accessible for
researchers and erases boundaries of time and
distance, such research raises new issues in research
ethics, particularly concerning informed consent and
privacy of research subjects, as the borders between
public and private spaces are sometimes blurred.
Internet communities
Internet communities provide a way for a group of peers
to communicate with each other. They include
discussion boards on websites, mailing lists, chat rooms,
or newsgroups. Examples of health related mailing lists
can easily be found by inserting a key word such as can-
cer in the search box at the Topica website (www.
topica.com/). One example is the Breast Cancer Mailing
List, based at Memorial University in Newfoundland
(www.bclist.org), which provides a forum for those seek-
ing peer support and information, with an emphasis on
information on treatment and disease, practical infor-
mation (such as relevant online resources), personal
experiences, and emotional support.
Qualitative research on the internet
Qualitative research seeks to acknowledge the
existence of and study the interplay of multiple views
and voices including, importantly, lay voices.
1
Inter-
net postings are accessible for qualitative research of
these voices for example, to determine information
needs and preferences of consumers or to investigate
how health related information can best be converted
into knowledge and disseminated widely.
2
Three different types of internet based research
methods can be distinguished.
3
One is passive analysis,
such as studies of information patterns on websites or
interactions on discussion groups without the
researchers actually involving themselves. Examples
include the study of helping mechanisms and content
of online self help groups for colorectal cancer,
4
breast
cancer,
5
Alzheimer’s disease,
6
and eating disorders.
7
The second type of online research is through active
analysis, in which researchers participate in
communications for example, to determine the accu-
racy of responses to healthcare questions on the
Usenet.
8
In the third type researchers identify
themselves as such and gather information in the form
of online semistructured interviews, online focus
groups, or internet based surveys or use the internet to
recruit participants for traditional research.
Internet communities’ members do not
expect to be research subjects
There is increasing evidence that researchers posting
or lurking on such communities may be perceived as
intruders and may damage the communities. King
quotes a group member who, realising that the
community had been monitored by a researcher,
Summary points
Internet communities (such as mailing lists, chat
rooms, newsgroups, or discussion boards on
websites) are rich sources of qualitative data for
health researchers
Qualitative analysis of internet postings may help
to systematise and codify needs, values, and
preferences of consumers and professionals
relevant to health and health care
Internet based research raises several ethical
questions, especially pertaining to privacy and
informed consent
Researchers and institutional review boards must
primarily consider whether research is intrusive
and has potential for harm, whether the venue is
perceived as private or public space, how
confidentiality can be protected, and whether and
how informed consent should be obtained
Unit for
Cybermedicine and
eHealth,
Department of
Clinical Social
Medicine,
University of
Heidelberg,
Bergheimer Str 58,
69115 Heidelberg,
Germany
Gunther Eysenbach
senior researcher
Joint Center for
Bioethics, University
of Toronto, Toronto,
Ontario M5G 2M9,
Canada
JamesETill
professor emeritus
Correspondence to:
G Eysenbach
ey@yi.com
BMJ 2001;323:1103 5
1103BMJ VOLUME 323 10 NOVEMBER 2001 bmj.com

retreated from a mailing list with the remark: When I
joined this, I thought it would be a support group, not
a fishbowl for a bunch of guinea pigs. I certainly don’t
feel at this point that it is a safe environment, as a sup-
port group is supposed to be, and I will not open
myself up to be dissected by students or scientists.
9
One subscriber to the Breast Cancer Mailing List
responded to one of the frequent requests from
researchers interested in obtaining insights into patients’
personal experiences with breast cancer: Why can’t
researchers do it the ‘hard way’ as they used to . . . and
leave us alone on the Breast-Cancer list?
10
Sharf
reported that among 14 people from a mailing list con-
tacted to obtain informed consent, one woman was
somewhat hostile, assuming that [the researcher] had
behaved voyeuristically, taking advantages of people in
distress and that the idea of using the conversations as
data had not occurred to many members.
11
We searched the Dejanews archive (http://
groups.google.com/) to find comments of newsgroup
participants responding to researchers’ requests. Enter-
ing the search terms research survey project health
identified 85 messages from researchers. By following
the message threads, we could find newsgroup
participants’ reactions, which were often negative.
Among the concerns expressed was that often the
researcher isn’t familiar with newsgroup culture
problems akin to those occasionally experienced by
anthropologists when they try to interpret the behaviour
of cultures they really don’t understand. And on the rare
occasions when I’ve seen someone who’s part of that
culture attempt to study the group he/she is part of, it
resulted in a rippling sense of resentment and betrayal
among those who find such things underhanded.
Public spaces or private rooms?
Informed consent, privacy, and confidentiality are basic
ethical tenets of scientific research on people.
12 14
To
determine whether informed consent is required, you
first have to decide whether postings on an internet
community are private or public communications.
This distinction is important because informed
consent is required when behaviour of research
participants occurs in a private context where an indi-
vidual can reasonably expect that no observation or
reporting is taking place.
15
On the other hand,
researchers may conduct research in public places or
use publicly available information about individuals
(such as naturalistic observations in public places and
analysis of public records or archival research) without
obtaining consent,
15
and research involving observa-
tion of participants in, for example, political rallies,
demonstrations or public meetings should not require
Research Ethics Board review since it can be expected
that the participants are seeking public visibility.
16
Although publication on the internet may have
parallels to publishing a letter in a newspaper or saying
something in a public meeting, there are important
psychological differences, and people participating in
an online discussion group cannot always be assumed
to be seeking public visibility. On the internet the
dichotomy of private and public sometimes may not be
appropriate, and communities may lie in between.
Several measures can be used to estimate the
perceived level of privacy. Firstly, if a subscription or
some form of registration is required to gain access to a
discussion group then most of the subscribers are likely
to regard the group as a private place in cyberspace.
17
Secondly, the number of (real or assumed) users of
a community determines how public the space is
perceived to be: a posting to a mailing list with 10 sub-
scribers is different from a posting to a mailing list with
100 or 1000 subscribers. However, as messages sent to
mailing lists are sometimes also stored in web accessi-
ble archives, the actual number of people accessing
messages may be greater than assumed and may be
impossible to determine.
Thirdly, and perhaps most importantly, the percep-
tion of privacy depends on an individual group’s norms
and codes, target audience, and aim, often laid down in
the frequently asked questions or information files of
an internet community. For example, SickKids is a
discussion list for children who are ill. The information
file about the mailing list states that adults will NOT be
permitted to participate on this list as its purpose is to
provide kids with their own personal place to share. It
seems clear that children who send messages to this list
are unlikely to be seeking public visibility. Similarly, a
virtual self support group of sexual abuse survivors was
reported to have a group policy explicitly discouraging
interested professionals who were not sexual abuse sur-
vivors from joining the group,
9
yet a decision was made
to analyse postings without obtaining prior or retrospec-
tive consent from the group members.
18
Can informed consent be waived?
If it is thought that a community may be perceived to be
private, the next question is whether informed consent
for passively analysing the postings is needed or whether
this requirement can be waived. In clinical studies
non-intrusive research such as retrospective use of exist-
ing medical records may be conducted ethically without
the express consent of the individual subjects if the
material is anonymised at the earliest possible stage, if
there is no inconvenience or hazard to the subjects, and
if the institutional review board has reviewed and agreed
the research protocol.
19
Similar considerations may be
applied to passive analysis of messages on mailing lists.
When considering potential hazards to group partici-
pants or the community as a whole, privacy issues are
especially important, and it should be considered
whether publication of the results (especially when men-
tioning the group name) may negatively affect group
members or harm the community as a whole. Much will
depend on what data will be collected and how they will
be reported, how vulnerable the community or sensitive
the topic is, and the degree to which the researcher
interacts with group members.
How can informed consent be obtained?
If a researcher and the institutional review board feel
that obtaining informed consent is necessary, how
should it be obtained? For mailing lists, there are basic-
ally two possibilities. The first is to send an email to the
mailing list describing the research prospectively and
giving participants the opportunity to withdraw from
the list. The second is to ask retrospectively each person
whose postings have been or will be used, giving them
the possibility to withdraw themselves from the analysis.
Information in practice
1104 BMJ VOLUME 323 10 NOVEMBER 2001 bmj.com