Web Surveys 101

With the media up in arms about the new "sexting" survey report, some researchers are trying to shed some light on the alarming results. David Finkelhor, director of the University of New Hampshire's Crimes Against Children Research Center, told the Wall Street Journal, "These kinds of samples select internet cowboys and cowgirls. These are more likely to be the kind of people who engage in this kind of activity." Finkelhor estimates the respondents in the sample are 2-4 times more likely to send nude photos of themselves than than the average teen. The point? The survey research was skewed by the survey method and survey sample.

Stop and think for a moment about the survey methodology, can you identify any possible biases in the collection method? The first one that pops into my head is the argument that the sample wasn't representative. Not all teens have texting capabilities or are technologically savvy. A teen who is not technologically advanced is not the type of person participating in online panels.

The Wall Street Journal article points out other issues with the online survey, but I think the main point here is to really think about your survey method. Is it going to represent everyone in the population or just a sub-group?