Buy Walking VideosFaulty Sample Size Skews Fattest & Fittest Colleges
They interviewed over 10,000 students at 650 schools. That’s only an average of 15 students per college. When are magazine editors going to take a statistics class and learn about statistically valid percentages?
According to CollegeBoard.com, there are approximately 3800 colleges in the United States. With an average enrollment of approximately 2,800 students, there are over 10 million students in school right now. Do you really think the answers from only 10,000 students is statistically valid? That’s only 0.09% of the student population.
Magazines want to attract your eye with promises. They want you to think that they know the secret to becoming thin and strong.
“Which colleges and universities are helping their students keep their weight under control?”
Can you really believe what they have to say, when they can’t even grasp the basics of statistics? When are colleges going to start requiring advanced mathematics for their journalism degrees? They sure need it.
Previous: PostSecret: Sick Halloween
Next: The Biggest Loser: The Biggest Winner
November 5th, 2005 at 10:42 pm
The fact that it’s 0.09% of the total population doesn’t affect the statistical significance; significance depends on lots of other issues, such as variability within the population, how the sample groups were chosen, etc. I don’t have a clue whether it’s a significant study, but the fact that “they only sampled 10,000 of 10 million students” certainly doesn’t tell you the answer one way or the other.
November 14th, 2005 at 8:04 pm
When the sample size is so small, it creates a large percentage of error. One or two percentage points is normal, but 0.09% of the population makes the percentage of error so large that the study is irrelevant.