Sampling The Realities
Sampling The Realities
Below is a question and a response about the realities of sampling for AP Statistics Projects. I fully agree with Floyd and in fact it is what we have been doing for many year. ---------------
---------------
The Response From A Very Experienced Teacher. (Floyd Bullard of The North Carolina School of Science and Mathematics, Feb. 2005)
I'm going to do something very bad. I'm going to recommend on the apstat listserv that your students opt for a convenience sample over a true simple random sample. Because I know that sounds like poor advice to most people, and counter to what we teach our students, I want first to explain why I think it may be preferable, both pedagogically and even in practice, and then I want to justify why it might not be so bad as one might at first think. Okay, first the "why". To take a true simple random sample, one must first have a roster of the whole population. In practice, this is quite rare. In a lot of contexts, such a roster is utterly impossible. There is no such roster of American households or adults, so pollsters often use random phone numbers instead. That is not a true random sample. Biologists do not have a roster of all the bears in the National Parks or all the oak trees in North Carolina, so they have other ways of sampling. Sometimes the sampling technique is such that it requires a special kind of analysis, not part of the AP curriculum. But there are times when the sampling technique is simply assumed to produce samples that roughly approximate simple random samples, and then the "regular" analytical techniques are used. So one reason I think your student might be encouraged to take a convenience sample rather than a random sample is that it mimics what is done in practice, even good Page 1 of 4
Page 4 of 4