Hi. In this module, I'm going to come back and talk about the other major concern we have when it comes to human subjects protections in social science research, confidentiality. When we talked about subject protections in social science research, we identified a few major areas. One that I just talked about in the previous module was consent. Have the subjects provided informed consent? Now I'm going to talk about, again, confidentiality. How will the identity of the participants be protected? How will the information that they have provided, how will that be maintained confidential and how will their privacy be preserved? I'll also talk about a few other issues including deception and the rare cases when it may be acceptable in certain kinds of studies. And I'll talk about the limited, although still present, risks of psychological harm that may arise in very specific types of studies in social science research. Now, confidentiality and maintaining confidentiality is one of the really big topics in social science research. So how do we make sure that our research subject's identities are not leaked, that their information is not leaked? There are many situations where subjects could suffer harm if their responses were made public and they were identified and that information became accessible to others. In some cases, there are situations, especially in countries where they have authoritarian governments. But even in countries that are democracies, subjects, certain people, could become at risk even if they were identified as having participated in a particular study even if none of the information about the contents of their responses was leaked. Think about a ethnography or a study that focused on a stigmatized behavior in North America where perhaps all of the information provided by the respondents was kept secret, but the very fact that somebody had participated in the study, identified them as a member of a group, an identity that they had hoped to keep hidden. Now, again, as I said, this is most clear for research on illegal or stigmatized behavior, but there are situations where we have to think about this for even more mundane topics, especially when we think about details like people's incomes. It's not acceptable to release information about people's incomes or, indeed, a lot of other details about people. I'll talk about some of the specific reasons why even mundane details can be problematic and need to be kept secure, in just a moment. So, how do we maintain confidentiality? Normally, studies will maintain information about the subject's identity offline, possibly on paper, separately from the data to be analyzed. So, as much as possible, a dataset that is used for analysis should be anonymous and not contain things like names and so forth. Those should be stored separately. Computers on which data are analyzed should be secured with the advice of experienced IT personnel and taking advantage of their expertise, encrypting hard drives, perhaps disconnecting them from the internet, etc. External drives and USB drives and so forth on which data are transported need to be encrypted using widely available tools. Again, computers may be disconnected physically from the internet, and may be placed in a secured enclave. They may also have their USB ports disabled to prevent the transfer of files. And most commonly, users of sensitive data may be required to sign agreements where they understand that they are at legal jeopardy if they misuse the data, that they may be sued or even in the case of misusing, say, government data in the United States, census data from the United States, they may be at risk of jail because they're committing a felony if they misuse census data. Now, one issue that we're really worried about in the last 10 or 20 years is re-identification. This is the possibility that a anonymous dataset or what seems to be an anonymous data set – all the names have been scrubbed, all the addresses have been scrubbed, there's no phone numbers, no social security numbers, etc. – but where people in the study could be re-identified by linkage to other perhaps commercially available databases. So, combinations of characteristics may be used to identify respondents in other databases, and then when their identity is known, go back and look at the original dataset to reveal sensitive information about them. So, preventing re-identification is an especially important topic when we think about publicly releasing data on living people. We have an emerging standard as we talked about in previous lectures of releasing surveys and so forth so that people can download them and analyze them. But we have to be worried about the prospect that somebody could use a combination of characteristics to identify somebody in some other dataset and then match the sensitive information they provided in the survey to their real name and perhaps leak it. So, what we are most concerned about are things like what seem to be innocuous details, at least innocuous if they're by themselves: age, education, income, occupation, region. You put all those things together and, even though any one of them doesn't seem to identify a person individually, it might be possible to look somebody up in a commercial database. For example, if you looked for a somebody who's 49, they have a Ph.D., they have a particular income, their occupation is professor, and then they live in Clearwater Bay, you may be very easy to identify me as somebody who participated in a survey. So, this is what we worry about with re-identification. It's taking apparently innocuous details but with enough of them, three, four, five, six apparently innocuous details, looking people up to get their names, their contact information, from a commercially available database or some public database and then matching the sensitive information they've revealed in a survey to their name and then releasing it. Now, I want to talk a little bit about some other considerations. Certain types of studies, in fact, although they're rare, we do have to accept the prospect that they could at least raise the risk of psychological harm for participants. So, we worry about this, in particular, if we're conducting surveys or interviews where we're trying to get participants to talk about or recall traumatic experiences. Sometimes, we know that the recall of especially traumatizing experiences may stress people. So think about this in the context of important topics, like experiences of sexual abuse, rape, they being victimized in crime. For some people, simply being asked to recall these sorts of traumatizing experiences may send them into a spiral of recollecting a terrible experience. And then in those situations, we have to think about how do we either try to minimize the chances that people suffer this kind of harm or how do we help them if, in fact, we do trigger some sort of a recall that is, again, traumatizing. So, another situation is that, especially in psychology where people conduct experiments, sometimes as part of the experiment, people may be in, they may be asked to do things that will induce stress. This may be deliberate. So, psychological studies looking at the relationship between stress and health might deliberately stress people out by asking them to perform public speaking. That's stressful for a lot of people. Standing in front of a camera may cause people's blood pressure to increase. Now, for most people, that's a manageable risk, but for a certain number of people, the stress may have unintended or unforeseen consequences. We have to have a plan in place to deal with that. There are also situations where we have to think about field experiments with control treatment designs that might have unintended consequences. So, especially we talked about in previous lectures, sometimes people carry out field tests or field experiments involving policy changes, perhaps the introduction of particular dietary changes or dietary supplements or changes to the educational curriculum and so forth. Now, it's always possible that these could have unintended consequences on the residents of the communities that are subjected to these policies. So IRBs will look at these sorts of risks in great detail. And if the study appears to have these sorts of elements, they will need to have a plan to mitigate the risks or address any adverse consequences that may arise as a result of participation in the study.
