Dave's ESL Cafe's Student Discussion Forums

[shvetsov2005]

Hi, there.
I am going though a test.

Question in test:

A quality engineer has asked you to use a tool to create a visual and statistical means to determine if there is a correlation or relationship between temperature and yield. What tool would be most useful?

Answers:
1. Correlation
2. Changes over time
3. Percent defect
4. Capability

I've chosen number 2. The test shows the correct answer is 1. Can you explain why?

Alex.

[CP]

The answer, correlation, must refer to the statistical test for correlation between two variables. The statistic, r, can range from -1.00 through 0 to +1.00 to express, respectively, a perfect inverse corrlation, no correlation at all, and a perfect positive correlation. In most real-world applications, of course, the correlation is less than perfect. The strength of the correlation, however, is generally greater as the number approaches 1.00.

Now, besides using just the statistic, the researcher can actually plot out the data on a graph, where each point represents the individual score on two dimensions at once. Here, we are looking at temperature and yield. Let's suppose that the warmer the climate, the greater the yield of corn. So we'll put yield on the Y axis and temperature on the X axis. So for each field or country or township or farm that is being measured, we can put a point where that entity's yield and temperature fall, and once we have, say, 50 or more points, we will have a nice scatterplot whose shape will tell us immediately whether there is any relationship between the two variables, whether it is positive or negative, and whether it is strong (more linear) or weak (more amorphous).

If warmer climates produce greater yield of corn, then the general shape of the scatterplot will be oval with the left end lower and the right end higher on the graph. The stronger the correlation, the thinner the oval.

Changes over time won't help, since we are looking at two variables that do not include time. In fact, if the relationship is true, it shouldn't matter when the data were collected. Percent defect makes no sense for our problem, since we are not looking at any sort of defect. And capability is even less sensible -- what does that even mean?

Hope this helps.
_________________
You live a new life for every new language you speak. -Czech proverb