Many of the same ethical issues need to be considered for on-line research as with any other type of research, but there are a few differences, an expert says.
It's working out the details of these differences that can be challenging.
For instance, privacy issues, security, and qualifying participants can be trickier with Internet studies, says Robert Kraut, PhD, a professor at the Human Computer Interaction Institute, Carnegie Mellon University in Pittsburgh, PA. Kraut was the lead author of an article published in American Psychologist, in the February/March 2004 issue, that offers recommendations to researchers and IRBs on how to deal with on-line research.
At the same time, Internet research creates many new opportunities, and IRBs sometimes are too cautious, Kraut says.
"I'm on our institution's IRB, and I got on it because I thought they were erring too much on the side of caution," Kraut says. "They didn't know enough about what life is like on-line."
Since Kraut joined the IRB and has educated members about Internet research, their review approach has improved, he notes.
Some of the reasons why researchers are attracted to Internet research, include the following:
- the Internet decreases the cost of recruiting survey subjects and others;1
- social behavioral scientists can observe a wide variety of communication archives;1
- other archived data about human behavior can be observed on-line;1
- computers provide automation and experimental controls that otherwise are difficult or unavailable;1
- the Internet itself is a social phenomenon that raises research questions.1
However, IRBs and researchers need to be sensitive to how Internet research may impact ethical debate and resolutions.
Here are Kraut's suggestions of ethical areas to consider in on-line research:
• Recruitment criteria: One of the major issues with on-line research is that it's difficult to exclude children from participation, Kraut says.
"You can put a statement on the study saying a participant has to be 18 years of age to enter, but there's no guarantee that any kid who's interested wouldn't still participate in your research anyway," he says. "And you have no way of excluding kids."
So if the research involves greater than minimal risk and it's likely to appeal to children, then it probably shouldn't be conducted on-line, Kraut says.
Here's one example: One researcher has done experiments on-line, looking at the issue of ostracism. Participants play a small game in groups of three. After they have played for a little while, two of the people in the game ignore the third person, who is the real research participant, Kraut describes.
"There's evidence that ostracism makes you feel bad, makes you feel excluded, and if that's greater than minimal risk, then that might not be research you want to do if you think you can't exclude underage participants," he says.
In the example above, the IRB debated the ethics of the research and finally decided that it was of minimal risk, and so they allowed the investigator to conduct it on-line, Kraut adds.
• Knowing when harm has occurred: In a laboratory, researchers can observe how participants react, and they will typically know whether someone is upset by a psychological or social experiment, Kraut says.
"On-line you can't see that," he says. "So you might want to build in probes that give you more insight because you're missing this other thing that's happening to them."
For instance, the Internet survey might include intermediate questionnaires that ask about the person's mode, and those can be used to terminate a study for a particular participant, Kraut suggests.
"The flip side is when somebody comes into the laboratory, the experience of being there places them in a quite compelling social situation," Kraut says. "It's hard for someone to extract themselves without feeling stupid or feeling they've violated a set of norms or promises they've made."
For this reason, participants are more likely to sit through events they might find distasteful or potentially harmful because it's hard for them to exit the laboratory setting, Kraut explains.
So IRBs have to balance the risk with the flip side in on-line research.
"It's harder to tell whether someone is being adversely affected by the experience when they're on-line, but there's less necessity for researchers to do that because the research participant has more control over his or her behavior," Kraut says. "There's a trade-off there."
• Privacy poses the biggest risk: "Probably the biggest risk of on-line research is not the psychological risk of participation itself, but the risk that the data that's collected is going to be exposed and cause harm outside of the research context," Kraut says.
However, that risk exists with any research pro-ject in which data is put on a computer, he notes.
"Although many people worry about the ability of outsiders to get the data when it's on-line, it's a risk that's common to any network computer application system," Kraut says. "If you use data analysis software, it's the same risk of privacy violations whether it's on-line or off-line and collected and stored on a computer."
The answer is the same: "Separate personally-identifying information from information that's harmful, if exposed outside of the research context, and try to do that as soon as possible when you collect the data on computer," Kraut explains.
Although many Internet surveys are conducted anonymously, not all are, he notes.
For example, for a gay men's health study, one researcher is conducting the survey on-line, and as part of the inducement is giving participants a payment, Kraut says.
"So you need some way to crack their identities so they don't receive multiple rewards for participating in the research," he says.
The key is to separate the identifiers from the potentially revealing health and behaviorial information, Kraut adds.
• Is Internet behavior public or private? "A number of people have started to do research that looks at the many on-line groups and communities," Kraut says. "One issue that's unique to on-line research is how to figure out the extent to which the behavior you are observing is public behavior where people have the expectation that strangers are able to see what it is that they do."
For example, suppose a participant is in a user group that talks about testicular cancer.
"This is an intimate topic," Kraut says. "You're revealing intimate things about yourself, and you're doing it in a public forum that anybody in the world can see."
In another example, users may belong to a registration-only usernet in which the information shared is not entirely private, but can only be viewed by others who have gone through the effort of registering with the Internet site, he says.
This example might involve a cancer support group on-line that requires registration, and an investigator also registers to be able to view the messages that are exchanged, Kraut says.
"That's a borderline case of whether it's public behavior that's exempt from the IRB regulations or whether it's considered private behavior that requires informed consent," he explains.
Unlike the real world, where it's very clear that observing people playing with their kids in a park is public, and observing them play with their children in a living room is private, the Internet provides various shades of gray on the public versus private continuum, he notes.
"People are more likely to think they are behaving privately when they have to log in when posting these messages," Kraut says.
"IRBs have to be sensitive to that issue of where in this continuum of private versus public a particular circumstance is," he says. "You have to evaluate that based on each circumstance."
• Consider harm to the group: "The other thing that's quite distinct, and we're still talking about the study of on-line groups, is to think about the group and it's potential harm," Kraut says.
"The IRB regulations focus on potential harm to research participants, but people who are doing research in on-line settings also have to think about the potential harm to bystanders who are not participants," Kraut explains.
"You need to take group-level harms into account, and not just the people about whom you are collecting data," he adds. "You need to consider the bystanders, as well."
Reference:
- Kraut R, Olson J, Banaji M, et al. Psychological research online. Am Psych. 2004;59(2):105-117.
You have reached your article limit for the month. Subscribe now to access this article plus other member-only content.
- Award-winning Medical Content
- Latest Advances & Development in Medicine
- Unbiased Content