Tuesday, December 05, 2006

confidential and anonymous

I am a graduate student in psychology. Maybe I already said that--maybe not. Anyhow, I was in class last night listening to a presentation in my research methods class. We had to do an assignment where we had to develop a measure and then give a brief presentation. One group chose to make their measure about couple satisfaction, and the students in our class made up the bulk of the sample.

Pretty interesting survey: it asked about shared interests, values, household labor, sex, hobbies, etc.

At the end of the survey I noticed that one of their demographic questions was race. I was hestitant to fill it out because I _knew_ that they were probably not going to have a racially diverse sample (I am the only person of color in my class,) and my concern was that if I indicated my race, the survey would no longer be anonymous, and they would have pretty detailed information about my level of satisfaction in my relationship.

I thought about it and decided that I wasn't going to check "white" because I was not white. But I was concerned enough that when I turned this survey in I made a point to say to one of the group members, "Make sure that you get some other 'African-American' respondents because I don't want it to be obvious what survey is mine." She assured me that they would.

Lo and behold, when they are presenting to the class they have gotten no other African-American respondents, so my results are availalable for everyone in class to see. So much for "confidential and anonymous."

Now is the part where I say: I don't think these women are racist dumbasses who did this intentionally. I know both of them: they're friendly people.

However, I think what they did was extremely irresponsible. It was very short sighted of them not to realize that if they had a demographic group of only one person, it is pretty pointless to present the results. First, because it has absolutely no statistical significance. Second, because the results are no longer confidential if the respondent is easily identified by others.

It also makes me upset, because I went through the trouble of bringing my concerns to one of the members, only to have her basically lie to me. It's like, "If you're not going to have a racially diverse sample, don't tell me that you are. That's just rude."

So I debated sending them both an email saying that I thought it was irresponsible, but decided against it because I thought they would immediately become defensive, and I didn't feel like dealing with it.

So instead I write about it on my blog, for who knows who in the universe to find it and read about it.

As always, I welcome your comments.

1 comment:

  1. Anonymous9:23 PM

    I wish I'd seen this blog before your class was over. I'd've recommended a topic for you to discuss: statistical disclosure limitation methodology. I think that a graduate research methods class in psych should include what IRB's & ethicists expect with regard to human research protection, like privacy & confidentiality. I especially would like a word in the ear of your colleague/researcher who pledged confidentiality then presented data in a way where you were individually identifiable. Oh, no! If this were real life instead of grad school, they wouldn't get any more research grants and, if they were in a clinical field, they'd better be all insured/lawyered up.
    And the professor. I'd like to send that bird a heads-up, or maybe a link to the OHRP/NIH/NSF, etc. requirements.
    Sigh. I was a volunteer lay 'non-scientist' member of an IRB that looked at international research. One tried to make sure that cell size was big enough so that personally identifiable data wasn't disclosed ("28, male, single, occupation=mechanic/electrician for music recording company, HIV status=positive, location = Madura, Java, Indonesia, housing=one-bedroom apartment, income less than $10K, amputated finger on left hand from bull-racing competition, diagnosed with depression and schizoaffective disorder" --> "Hey look, Joe has AIDS!").