My thanks to Tripod for making this page available.
(The ads are Tripod's, not mine!)
The current counter number is:
A publication by Gunther Eysenbach and James E. Till, entitled
"Ethical issues in qualitative research on internet communities"
has been published in BMJ 2001(10 Nov); 323(7321): 1103-1105
In it, this commentary is cited (see reference 10).
RESEARCH ETHICS: INTERNET-BASED RESEARCH
PART 1: ON-LINE SURVEY RESEARCH
Jim Till, Ph.D.
Joint Centre for Bioethics, University of Toronto
c/o Ontario Cancer Institute
610 University Avenue, Room 9-416
Toronto, Ontario M5G 2M9
Canada Send me email at: firstname.lastname@example.org
Internet-based mailing lists and newsgoups frequently receive requests
for participation in on-line surveys. (A discussion about such requests,
circulated to members of the Breast-Cancer mailing list in July, 1997,
had as the subject line: "Researchers - A Plague of Angels?"!).
The purpose of this article is to provide three examples ('case studies')
of requests of this kind, and to comment on some ethical implications of
From an ethical perspective, Internet-based mailing lists and newsgroups
might regarded as having some of the characteristics of 'virtual
collectivities'. The Council for International Organizations of Medical
Sciences (CIOMS) defines 'collectivities' as "population groups with
social structures, common customs, and an acknowledged leadership".
Collectivities include nations, cultural groups, small indigenous
communities and some neighbourhood groups. The CIOMS definition is also
explicitly intended to include families. In contrast, 'groups' are
defined solely in "demographic or statistical terms, with neither leaders
It could be argued that Internet-based mailing lists or newsgroups should
*not* be regarded as 'collectivities' in the sense of the CIOMS
definition, because the extent to which they exhibit either formal
governance structures, or acknowledged leadership, may be open to
question. For the purpose of this article, it will be assumed that
current views about the ethics of research involving 'collectivities' may
provide a useful context within which to consider the ethics of research
involving Internet-based mailing lists or newsgroups.
The three major federal agencies which support research in Canada have
prepared a draft 'Tri-Council's Code of Conduct' for research. One
section (Section 13) of the Tri-Council's Code of Conduct that addresses
research with 'collectivities'. Two aspects of the guidelines
contained in Section 13 will be highlighted here:
"Article 13.2: Except where it is self-evidently not the case, the
researcher, regardless of the nature of the research to be undertaken,
must consider the group to be researched a 'collectivity' as defined
above. Therefore, if such a researcher does not wish to follow the
Code of Conduct respecting 'collectivities', it is incumbent upon her
or him to explain how the group to be researched is not a collectivity
as it has been defined above".
"Article 13.6: When undertaking research with a family, community or
other collectivity, the researcher may not begin until permission has
been obtained from the appropriate authorities for that collectivity.
Furthermore, obtaining group and individual consent before beginning
any research involving a collectivity, or its members, must be
regarded as a minimum requirement".
A key refinement of the definition of a 'collectivity' is included in
a revised version of the Tri-Council's Code:
"They [collectivities] are ordinarily groups in which there is mutual
recognition of membership both by those in the group and those outside
This revised definition does not include a requirement for formal
governance structures, nor for acknowledged leadership.
Below, I outline three (real, not hypothetical) 'case studies' that I
think are relevant to Internet-based research, and offer some preliminary
comments and questions about them. These comments and questions will be
based on current widely-accepted guidelines about research ethics, as
described by CIOMS, and by the Tri-Council's draft report.
The first three 'case studies in this series are:
1. Surveys - Case #1: A survey questionnaire, with a subject heading
'Query', was posted to the Breast-Cancer mailing list (at
BREAST-CANCER@morgan.ucs.mun.ca) on March 17, 1997.
The author of the message began by identifying himself as a student, at a
School for Math and Science, who is required to do a senior thesis. He
then outlined his plans for the thesis:
"My thesis this year consists of a comparison of
radiation chemotherapy versus combination chemotherapy using other
chemicals. An appendices of alternative medicine and sources to get
more information will be included, but what follows is the part that
I need your help on. Although I am seeking to become a doctor, I
want to add a more "human" aspect to my thesis, so I am asking you to
submit your answers to the following questions, but please be advised
that some of the questions are quite personal".
This introduction was followed by a questionnaire. The first question
"1) Can you describe how you felt when you or someone close to you
This question was followed by more questions, asking about "philosophies
of living life", about plans "to cope with the aspect of death and hope
for a cure", about (if a caregiver) how to "cope with the feeling of hope
and despair", about what advice to give to "someone who has just been
diagnosed", and about sources of "your best support".
The last question (of a total of 7) was:
"7) What is an impression that you wish to leave on the world?"
The message concluded with a paragraph that included an apology for
writing "such a depressing e-mail", and the comment:
"there is no way that a person can be empathic to the plight of someone
who has had a problem in their life, unless you have experienced it,
and even then the experience is different. I have never experienced
this pain, hope, or tragedy, but I want to , and to make those that
read my thesis feel it all through their body and to the depths of
what we call a soul".
Comments about Case #1:
Subsequent information sought by mailing list members, or supplied by the
original author of this questionnaire, revealed that the author was not an
impostor, that he was a student at a secondary (high) school, and that he
was considered to be one of the top students at his school. This student
stated plans to study medicine, as a result of direct experience with
cancer (deaths of a grandmother, grandfather, and close personal female
Although it appeared that the proposed research project had been approved
by the student's faculty advisor, it apparently had not been subjected to
reviews of scientific merit and ethical acceptability of a kind that
normally would be expected of 'real' research projects (such as research
projects involving human subjects, designed and carried out by faculty
members at major universities).
Subsequent comments sent by other subscribers revealed a range of opinions
about this survey, from ones that were supportive of the student's plans,
to ones that criticized such use of the Breast-Cancer mailing list. (This
list has, as major purposes, the provision of information, support and
advocacy about breast cancer).
Should one conclude that it is inappropriate for the standards of research
ethics to be less rigorous at the secondary school level (if the intent of
the research is considered to be benevolent) than the standards for
research ethics at the university level? In other words, is it
inappropriate for the standards of research ethics to be dependent on the
particular context within which the research is to be done?
2. Surveys - Case #2: A survey questionnaire, with a subject heading
'Survey: Circumcision and the Waterworks' was posted to the newsgroup
sci.med.prostate.bph on April 11, 1997.
The purpose of this survey was (at least in part) to explore possible
relationships between circumcision and prostate-related problems.
The 2nd and 3rd paragraphs of the message read as follows:
"This questionnaire is journalistic in intent, is not sponsored by any
academic or other body and is not a part of any scientific research
programme. However the results will be analyzed according to strict
scientific standards, and anonymous data can be provided to reputable
medical research institutions".
"All forms will be stored and entered anonymously. Email headers, names
or addresses will not be preserved with the saved forms".
The survey begins with a question that asks whether or not the respondent
has been circumcised, and, if so, at what age. It includes several
questions about urinary characteristics or problems. The final (12th)
"12. Do you feel that being circumcised/uncircumcised has affected
your waterworks in any way? If so please give details ....
Comments about Case #2:
Note that it is stated very clearly that the questionnaire is journalistic
in intent, that it is not sponsored by any academic or other body, and is
not part of any scientific research program (however, it is also stated
that "anonymous data can be provided to reputable medical research
If this survey happened, in fact, to be part of a scientific research
program supported by a reputable medical research program, the research
protocol of which it formed a part would normally be expected to have been
reviewed, in advance, by an appropriately-constituted Research Ethics
Board (usually called an IRB in the U.S.A.). The membership of such a
Board would normally include non-researchers (such as ethicists or
lawyers) as well as researchers.
Ideally, such a research protocol should also have been peer-reviewed in
advance in relation to its scientific merit. At the very least, a
research questionnaire should have been pretested, in a preliminary
attempt to evaluate its clarity and feasibility, reliability, and
Does the inclusion of the 2nd paragraph (see above) in the 'Waterworks'
questionnaire imply that this questionnaire does not need to meet the same
ethical standards that are usually applied to questionnaires designed for
use in a scientific research program?
Should one conclude that it is inappropriate for the standards of
journalistic ethics to be less rigorous than the standards for research
What if it turns out that the responses to the survey are, in fact,
subsequently used as part of a research study (e.g. as preliminary data to
help guide the development of a 'final version' of the questionnaire,
designed for use in a scientific research program)?
Should such use of the results be avoided, because, if such results are
used in research, then the original 'journalistic survey' becomes, in
effect, research disguised as 'journalism'? If research is disguised as
'journalism', then might it be claimed that some deception was involved in
the process used in the development of the 'final version' of the
3. Surveys - Case #3: A short message was posted by a college student
(email address provided, but no author's name) on April 11, 1997.
The message was cross-posted to several "alt-support" newsgroups,
including those for cancer, grief, single-parents, arthritis,
multiple-sclerosis, anxiety-panic disorder, and depression.
The purpose of the short message was to ask people to complete an on-line
questionnaire about on-line support groups. The questionnaire itself was
not included in the message. Instead, the location of a web page
containing the on-line questionnaire was provided, and people were asked
to: "Please complete my on-line questionnaire if you have a moment".
This short message elicited quite a few responses sent to the newsgroups,
most of which ranged from tolerant to very negative. There was concern
about cross-posting to newsgroups involving people who feel particularly
sensitive, fragile or vulnerable. Some people felt that their privacy had
been invaded, and that they were being "used as human guineapigs".
The alternative (minority) view was that such newsgroups are public, that
the main problem is people with commercial interests who approach such
groups. It was suggested that a researcher who posts a short message,
which simply provides the location of a web page containing an on-line
questionnaire, should be regarded with tolerance.
Comments about Case #3: In circumstances where concern is expressed about
a particular survey research project, it seems likely that responses to an
on-line questionnaire may be quite biased, and thus be of questionable
scientific validity. For example, selection bias is always a problem in
survey research, in that the people who choose to respond to a survey may
not be at all representative of the larger group from whom they have been
self-selected. Appropriate scientific sampling of the membership of an
Internet newsgroup or mailing list poses very difficult methodological
(and ethical) problems.
For example, are there some newsgroups or mailing lists that should be
regarded a 'semi-private' zones in cyberspace, ones that should not be
intruded upon by researchers whose only goal is to do survey research?
Indeed, should on-line research only be done within the context of
"participant-observer" research, where the researchers are known (and
trusted) members of the newsgroup or mailing list, and are regular
participants in a variety of discussions?
If one of the goals of the "participant-observers" is to undertake on-line
survey research, should they then be expected to prepare formal research
protocols which should be shared with (and developed with the help of) the
members of the newsgroup or mailing list? Should such protocols, when in
final form, always also be evaluated in relation to their scientific and
ethical acceptability, as would normally be expected for all (university
level) research protocols?
Case #1 raises the question: Is it inappropriate for the standards of
research ethics to be influenced by the particular context within which
the research is to be done? My own opinion is that there are both
empirical and conceptual justifications for such an influence.
Empirically, all surveys (or polls) are not the same (consider the
examples of scientific surveys, political surveys, and marketing surveys).
Conceptually, if standards of research ethics ignore the particular
context within which the research is to be done, then they may sacrifice
pragmatic flexibility in an (in my opinion, vain) effort to achieve
theoretical ethical neatness.
Case # 2 raises the question: If research is disguised as 'journalism',
then might it be claimed that some deception was involved? In my opinion,
an answer is "probably yes". However, if the original purpose of the
survey was indeed 'journalistic', not 'scientific', but very interesting
results emerged, would it be ethically-unacceptable for these results to
then be used as a basis for 'real' scientific research? I think not - but
the contextual details would matter (see Case #1).
Case #3 raises the question: Should on-line research only be done within
the context of "participant-observer" research, where the researchers are
known (and trusted) members of the newsgroup or mailing list, and are
regular participants in a variety of discussions? My own answer would be:
"probably yes, but the context matters" (see above). For example, within
the context of mailing lists involving vulnerable groups, some participants
may regard the mailing list as a kind of sanctuary, and thus may regard
an intrusion by a researcher into this sanctuary as a violation of the
participant's right to privacy.
Some Concluding Opinions and Suggestions:
I think that on-line survey research poses some significant ethical
and methodological problems. For example, its scientific validity is
likely to be questionable, at best. If such research is undertaken
at all, perhaps it should only be regarded as a form of 'pilot study',
useful in the initial design and pre-testing of a survey questionnaire?
For example, in private correspondence, a colleague (Barry, see
Acknowledgment) suggested that it would be quite feasible for someone
to establish a number of fake email addresses, and then to fill out a
number of on-line questionnaires with fake data. How would a researcher
know which email adresses were fake and which were not, and which
responses were valid and which were not?
From the perspective of potential participants in the research, there
should be awareness that the 'investigator' may be an impostor, or may
have ulterior motives, or may be poorly-qualified to undertake the
research, for either scientific or ethical reasons.
From the perspective of researchers and ethicists, there should be
awareness that some newgroups or mailing lists may fit the definition of a
'collectivity' better than that of a 'group'. Investigators (and IRBs)
should be aware of the ethical issues raised by research with
'collectivities'. Mailing lists or newsgroups should be selected with
care, approached with sensitivity, and treated with respect, with careful
attention to the purposes and the 'culture' of that particular newsgroup
or mailing list.
The members of mailing lists or newsgroups, if treated with inappropriate
insensitivity, are not helpless. For example, they could take steps
designed to invalidate a survey (see Barry's proposed means of doing
so, outlined above).
An attached Appendix includes some questions that might be answered,
in advance, by researchers who propose to carry out an on-line survey
(and which, if not answered in advance, might be asked by potential
participants in the research, before they decide whether or not to
The ethics of on-line survey research obviously involves balancing a
number of sometimes conflicting rights and duties, including those
related to privacy, freedom of inquiry, and participation in a
'collectivity'. Of these, the question that most needs further
attention from research ethicists may be the CIOMS definition of a
'collectivity', as "population groups with social structures, common
customs, and an acknowledged leadership".
If the concept of a 'collectivity' is to be flexible enough to be
applicable to a wide range of contexts, including populations groups
of the kind brought together by Internet-based mailing lists or
newsgroups, then perhaps this definition needs to be modified,
and/or more attention needs to be paid to the different ways in which
'social structures', 'common customs' and 'acknowledged leadership'
may themselves be defined. Perhaps a better definition is: "They
[collectivities] are ordinarily groups in which there is mutual
recognition of membership both by those in the group and those outside
Internet-based mailing lists do, of course, have 'acknowledged
leadership' in the person of the 'list-owner(s)'. Permission
probably always should be sought from the principal list-owner about
contacting members of that mailing list in relation to a proposed
survey research project.
However, my own opinion is that list-owners should not be expected to
speak for the mailing list membership as a whole. Instead, I suggest
that survey researchers should be asked to consider the questions
posed in the attached Appendix, and to provide answers to these questions.
Individual list members can then make their own informed decisions
about whether or not they wish to participate in the research.
The author gratefully acknowledges very helpful discussions with Barry
(a pseudonym), who has asked me: "Why can't researchers do it the 'hard
way' as they used to ... and leave us alone on the Breast-Cancer list?".
REFERENCES AND NOTES
 Council for International Organizations of Medical Sciences (CIOMS),
International Ethical Guidelines for Biomedical Research Involving Human
Subjects, 1993 [revised in 2002].
Click here for information about CIOMS
 Tri-Council's Code of Conduct (initial draft, no longer available).
A revised version of the Code was published in July, 1997 (also no longer
A 'final' version of the revised code was published on December 22, 1999.
An updated version is available via:
(Much material about "collectivities" was omitted from the 1999 version).
 Research With Collectivities, Section 13 of Tri-Council's Draft
Code of Conduct (no longer available).
A key refinement of the definition of a 'collectivity' was included in
Appendix E of the revised version (July 1997, no longer available):
"They [collectivities] are ordinarily groups in which there is mutual
recognition of membership both by those in the group and those outside
of it". If it is accepted that a subscription to a mailing list like
the Breast-Cancer Discussion Group involves "mutual recognition of
membership", then such a mailing list does fit this definition of a
If it is accepted that such a mailing list fits this definition of a
'collectivity', then the main issue, from a research ethics perspective,
is whether or not messages posted to the mailing lists may be regarded
as information that is in the public domain. I'll argue (click here)
that messages sent to mailing lists are the intellectual property of
their authors, and are not in the public domain (unless the authors
specifically state that they wish the messages to be in the public
 'Ishmael' (Jeremy Brown), 'Query'.
Click here for a link to Case #3
(Via the Deja News archives - see newsgroup: alt.support.cancer)
APPENDIX: Examples of questions that might be asked of survey
researchers (and might best be answered by them in an introductory
"information statement" distributed to potential participants in a
1. Has the principal list-owner been informed about the proposed survey?
2. Who is the researcher, what are her/his research credentials, and
what is the researcher's host institution (if a student, who is the
student's research advisor, and what is her/his email address)?
3. What is the purpose and justification for the proposed survey (why
is this an important research question that needs to be answered)?
4. What opportunity will potential participants have to comment on the
research question, and on other aspects of the proposed study,
before it is initiated?
5. Who, or what organization, is providing financial support for the
6. Has the research protocol already been subjected to an ethics review?
Has it been subjected to any kind of scientific peer-review? If the
answer to either of these questions is "no", then why not?
7. How have potential participants in the research been identified (or,
how will they be identified)? How have they been (or will they be)
8. What steps have been taken (and will be taken) to protect the privacy
of the research participants, and of the information that they will
9. Will comments or messages authored by the research participants be
quoted verbatim in any reports of the research? If so, will these
comments or messages be regarded as the intellectual property of
the research participants? Will their authorship of the comments be
acknowledged? Will the permission of the authors to be quoted be
sought? Will the authors be offered an opportunity to edit or revise
their comments before they are quoted in print?
10. What feedback will be provided to participants about the results of
the study? What are the plans for publication of the results?
Last revised November 18, 1997
(Minor modifications to the reference list were made on 8 July 2001
& on 31 Dec. 2003. A link to BMJ 2001(10 Nov); 323(7321): 1103-1105
was added on November 9, 2001).
Copyright 1997-2005 by Jim Till
Send me an email message, to: email@example.com
The use of these real 'case studies' itself poses ethical issues.
Click here to see Part 2 of this commentaryDisclaimerReturn to top of page