On 24 Feb 2004 07:53:02 -0800, [EMAIL PROTECTED] (burt) wrote: > How would you test the following hypothesis: > A fast-food restaurant claims that 75% of their revenue is from the > "drive-thru". Suppose you have 50 days of receipts from the > restaurant. Each days' receipt shows the total revenue and the > "drive-thru" revenue for that day. I'm in a quandary as to how one > would conduct this test. The two revenues are obviously dependent > under the null hyp; thus it's not just a simple test of comparing two > means from independent samples.(Even if the samples were independent, > I'm not sure how to do it-because of the 75% factor in the hyp. This > is a moot point ,though, since the samples are clearly dependent to > start with)
For the point assertion, I would take the two subtotals for the 50 days, and ask, "Is this one 3 times that one?" Is that a test, or an assertion of the-state-of-the-world? Do you want to know whether it is true every week? - then take weekly totals. That gives you seven weeks. The claim about "75%" might be shown to depend on other contingencies or chance; or might not. Taking 50 days separately is not very necessary, since it is not (I think) primary to the claim. However, it could be an additional point of convincing evidence, if the assertion were true (almost always) on that daily basis. For a statistical assertion about the proportion, I would probably use the weekly ones -- not nearly the accidental changes owing to days, or the day-to-day autocorrelation in whatever is going on. -- Rich Ulrich, [EMAIL PROTECTED] http://www.pitt.edu/~wpilib/index.html "Taxes are the price we pay for civilization." . . ================================================================= Instructions for joining and leaving this list, remarks about the problem of INAPPROPRIATE MESSAGES, and archives are available at: . http://jse.stat.ncsu.edu/ . =================================================================
