To The Editor:
The article "Stats class questions social norms" (The Dartmouth, May 25) contained an assertion that "the statistic is inaccurate because of the possible sample bias of those who responded to the online survey." My office, the Office of Evaluation and Research, conducted that survey and presented the results to the Health Educators who then crafted the social norms messages.
I have two points I'd like to make. One is that there is extensive research on the impact of survey non-response that the article never examines. There is, in fact, a branch of this research that specifically examines the validity of alcohol and drug use surveys. This research (some of which is summarized in a recent paper of mine published in the Spring 2004 issue of the journal "New Directions for Institutional Research") found that carefully-controlled experimental studies of survey responses indicate that respondents are answering truthfully. Most of these experiments ask people to respond to self-use questions on surveys, and then ask the respondents to take a blood or urine test to detect alcohol or drug use. Participants do not know that the testing will occur before they answer the survey questions. I believe, as do most of my professional colleagues, that while survey research is not perfect, it does have value if conducted properly. Decades of research on survey methodology has helped us to refine these methods.
My second point is that if anyone from The Dartmouth had contacted me, I would have been glad to present the other side of the story. But nobody did. I was contacted by Kailin Kroetz, the student who is quoted in your article, and provided her with answers to questions she had about the survey. If she had asked me at the time about sampling bias and validity, I could have pointed her to the above information as well as detailed the many other steps we take to ensure the best possible survey results. But she did not.
In closing I will also point out that the article cited from the Harvard School of Public Health that purports to examine Social Norms campaigns has also been severely criticized (including by people cited previously in your own paper), yet the article claims that there is a question as to why Dartmouth uses the social norms strategy. Many of these questions survey validity, sample bias, the reasoning behind the Social Norms strategy-can be answered. The article could have carried various possible answers to these questions if they had been asked before the article was printed.
Wouldn't it have been more interesting an article, and a better learning experience for students, if the headline had been "Stats class thoroughly examines the complex issues of social norms and survey research" instead of "Stats class questions social norms"?
