Ethical Guidelines for Research Online
College of Computing
Georgia Institute of Technology
No set of rigid rules can ever capture the subtlety of ethical situations
that arise. The following 'rules of thumb' are intended as a
Seeing where your research
goes beyond these guides should provide helpful insight into
areas that need special attention.
- You may freely quote and analyze online information without consent if:
- It is officially, publicly archived.
- No password is required for archive access.
- No site policy prohibits it.
- The topic is not highly sensitive.
- For everything else not covered by 1, you typically need consent.
- The process of requesting consent must not disrupt normal group activity.
- Note that in real-time chatrooms:
The process of requesting consent is
often perceived as disruptive by participants.
- When participants are made aware that their chat is being recorded,
they often react with anger. (Try
connecting to a chatroom and typing "I am recording this for research
purposes." You will usually be greeted with anger and often booted out.)
- It is sometimes permissible
to do research which displeases subjects. However, subjects' perceptions
are a factor for IRBs to weigh carefully in considering such a study.
- When recording material that would
otherwise be ephemeral,
consideration should be given to whether the act of recording potentially
creates risks for subjects.
- If a study is likely to record illegal or socially undesirable activity,
a certificate of confidentiality should be obtained.
- Records of sexual activity may be vulnerable to subpoena in divorce
- Consent may be obtained electronically if:
- Subjects are 18-years-of age or older.
- The online consent form steps people through each sub-element,
one at a time.
- The risks to subjects are low.
- Otherwise, consent must be obtained with a signature on
paper--returned to the researchers via surface mail or fax.
- To do research involving minors and others not allowed to consent
for themselves, consent is obtained from the parent or legal
guardian and assent from the individual.
- Parental consent may be obtained on paper (sent to the researcher
via paper mail or fax) or by telephone if the research is low risk.
- Parental consent should be obtained in a face to face to interview with the parent/guardian if the research is not low risk.
- Before the start of a study, the researchers and IRB must decide
whether subjects' identities will be disguised, and if so to what degree.
- Note that pseudonyms function similarly to real names, and should
be treated in the same way one treats real names
- They are often de facto traceable to real names.
- People care about the reputation of their pseudonyms.
- In other words, they function for most purposes just
like real names, and should be treated
the same way.
- Levels of disguising:
- No disguise
The Internet is rather like a playground for amateur artists. In many
cases, people deserve credit for their creative work.
- Pseudonyms and real names may be used with permission of the individual.
- The individual's claim to copyright over the work is respected.
- The real author of the work is verified. (Some work may be
copied or highly derivative, meaning credit belongs to the original author.)
- The researcher should omit details
that would be harmful to the subject if revealed.
Suppose you are studying artwork posted online, and discover
that one of the most prolific artists rarely leaves his house and
has few friends, devoting all his time to online art. You probably should
omit that from the written account, even if it
significantly contributes to your reader's
understanding of the work being studied.
- Light disguise
- The group is named.
- Pseudonyms and some other identifiying details
(place names, organizational and institutional names, etc.) are changed.
- Verbatim quotes may be used, even if they could be used to
identify an individual.
- Group members themselves may be able to guess who is being discussed.
- An outsider could probably figure out who is who with a little investigation.
- Details that are harmful to individuals should be omitted.
- Example: the author of these guidelines, Amy Bruckman, used
light disguise in her study of MOOSE Crossing, the educational online
environment for kids she created. While
sensitive details (for example a particular
child's difficulties in school) would have enriched the written accounts
of learning on the site, they were omitted.
- Moderate disguise
- A compromise position is chosen,
incorporating some features of light disguise and some of complete disguise,
as appropriate to the situation.
- Details of the specific compromise are reviewed by the researcher's IRB.
- Complete disguise
- The group is not named.
- Pseudonyms and other identifying details are changed.
- Verbatim quotes are not used if a search mechanism could link those
quotes to the person in question.
- Some false details may be deliberately introduced.
- For example, if you are studying a support site for a chronic disease,
you might change the disease in published accounts.
- A subject might or might not recognize themself.
- No one else would recognize the subject.
- Someone deliberately seeking to find a subject's identity would
be unable to do so.
- Details that would be harmful to the subject if disclosed may be revealed.
This document assumes that you are already familiar with
the basic requirements of human subjects research.
These guidelines represent a strict interpretation of what is ethical.
Many people will find looser standards acceptable.
Even if you prefer a much looser standard, I hope you will find
this document useful: you can compare your plans
to this strict interpretation to find places in your plan that
need special attention. So instead of thinking, "beyond this you
may not proceed," many people may think instead "beyond this,
proceed with caution."
These guidelines do not address the steps one needs to take to assure
that research is accurate.
Use of online media introduces new risks of error, for example
indiscriminate use of
low-quality or erroneous information found online, errors
of sampling method, etc. Taking reasonable steps to prevent error
is also an ethical obligation of researchers, but is beyond the
scope of this document.
It's important to remember that all "Internet research" takes place
in an embedded social context. To understand Internet-based
phenomena, you need to understand that broader context. Consequently,
most "online research" really also should have an offline component,
and many ethical issues become identical to those for traditional
Research Integrity Versus Ethical Obligations
A frequent mistake made by many Internet researchers is, when faced
with a tradeoff between needs of subjects and integrity of the
research, to give priority to the integrity of the research. On
further reflection, it should be obvious that this reasoning is
faulty--the rights of subjects come first. For example, when faced
with situations where getting consent from potential study
participants in a computer-mediated communication forum is
logistically difficult or potentially disruptive to the environment,
some researchers have concluded that consent is not required. In
fact, consent is still required, and substantially disrupting the
environment is not acceptable. In such a situation, the investigator
must fundamentally rethink the research plan or even abandon it, not
lessen their ethical obligations. Many errors in research ethics stem
from a researcher's sincere dedication to the quality of results.
While producting quality research results is in itself an ethical
imperative, it always takes back seat to the needs of subjects. Be
careful not to make this common mistake.
Consent from Whom?
Where consent is required,
you need to get consent from each individual you wish to record,
quote, or study in any way. In cases where you have a vulnerable
or under-age population, you need permission of the group's leader/
"gatekeeper" before you may solicit
participation. For regular groups,
you do NOT need permission of the group leader to ask for voluntary
participation from individuals, unless that is explicitly required
by a posted site policy.
An archive is "official" if its presence is announced in the welcome
message sent to new group members, or the archive address is
regularly posted to the list or noted in a group FAQ. An archive
is public if it is accessible without a password.
The point of this language is try to articulate how we as researchers
can determine when members of a list have a reasonable expectation
that their communication is private, and when there is no such
Research is "low risk" if it is highly unlikely that material harm
will come to subjects as a result of their participation. For example,
a study of a hobby newsgroup is probably low risk. A study of a group
for people with a serious illness is probably not low risk. Consult
your IRB for guidance on whether your planned study is low risk.
Q: Suppose I want to critique hate speech online. Do the rules change
at all? You're not tell me I have to protect the rights of those people,
A: The rules don't change. You can respond to hate speech or other
undesirable behavior online as a netizen or as a journalist, and there
are few restrictions on your ethical conduct--email their site
manager, publish letters decrying their behavior, do whatever you can.
But as soon as you put on your researcher hat, you owe them the same
treatment you do to any other subject. There are no exceptions in
the Belmont Report for doing research on evil-doers.
Research on Chatrooms
Q: Doesn't this make it difficult to do research on chatrooms?
A: Yes, it does. One suggested technique is to create your own
chatroom just for the purposes of your study. (That's what the author
did for the IRC
Francais project, a study of the potential of synchronous CMC to
support second-language acquisition.) When you
create your own chatroom, you can greet each person joining the channel/room
with a message informing them about the study. Depending on the nature
of the study, a channel logon message may be sufficient, or you may
need also to get informed consent from participants.
Changing Risk Levels
Q: My research started off as "low risk" but something unanticipated
that is more controversial happened. What do I do?
A: Consult your IRB for guidance. In some cases, you may need to
withold some information from your publications. For example, in
reports on one educational environment the author created, it would
shed light on important issues to comment on some children's personal
problems. However, because it was impossible to disguise their
identity sufficiently, this information was omitted from publications.
While this does diminish the quality of the scholarship, there was
really no alternative.
Please do write your own version of these guidelines. These are just
the opinion of one practitioner. Many other interpretations are
possible. I'll be happy to add a link to any guidinelines that link
back to these--email your URL to firstname.lastname@example.org. My goal is to create a set of contrasting interpretations
so individuals and IRBs can find a set they are most comfortable with.
You may add thoughts to the
CoWeb (collaborative website) page we've
setup on this topic. Thanks to Mark Guzdial's Collaborative Software Lab
for offering this resource.
People who have provided invaluable feedback on this document
and the ideas behind it include:
This document does not represent the views of those groups--this
is my own interpretation. However, I am indebted to both groups
for their wisdom, insight, and patience.
- Members of the online-research-ethics mailing list
- Members of the Association of Internet Researchers (AoIR) ethics
To subscribe to the online-research-ethics list, email
and place in our message body "subscribe