From Kevin Rees: >the >AP exam questions give you FAR more information than >you need in order to determine whether or not the student >understands the situation of the experiment. >Question #5 gave sample statistics for each sample and >for the differences, about 95% of my students correctly >choose to ignore the statistics for the differences as two >different ovens are "assumed" to be independent (yes they >could be connected to the same power source, be next to >each other, etc. yet without more in the question it seems >more appropriate to treat them as independent). >I am assuming that when the test is graded they will be >looking for two-independent sample analysis rather than >dependent samples analysis and will grade accordingly.
Kevin-- What I found most interesting about question #5 was the decrease in % of defective chips over time in both ovens. Time is ovbiously important, and it raises the question of whether or not the "population" of % defective chips is normally distributed (a normal quantile plot of Oven A's output is not particularly linear). My initial reaction to the problem was to lean toward doing matched-pairs on the hourly differences to eliminate the issue of time, but I didn't think about the problem of independence. (I haven't met with my students yet, so I don't know what they did). How big a problem is that? Shows how much I don't know.