8/12/2007

One of the problems with survey data

Obese people underestimate the amount of sugar they eat, making studies into the condition based on self-reporting very unreliable, UK researchers say. . . . . In a study of hundreds of volunteers, researchers compared what people said they ate with data from urine tests.


The problem with this is that it produces a systematic bias in the survey data. The high end of the survey results are biased downward. The problem is probably even worse than might be claimed here if those being studied in this case understood why their urine samples were being taken. The reason is that the people may have been relatively more accurate in their answers if they thought that they were being checked for accuracy. As the article points out, claims that sugar intake are unrelated to health problems or obesity could simply be due to this bias in the data.

Labels: