Here are the answers to exam 1. For some of the questions, I have provided explanatory material in regular type, and the answer in bold; all you need to write down is the answer. If you don't understand why your answer was wrong, you may e-mail me, talk to me before or after class, or set up a time to talk to me in my office. The exam was worth 15 points, so each question was worth 0.75 points.
1. Two nominal variables, Lignextra vs. control, snoring vs. not snoring; total sample size is greater than 1000 (600+600); chi-squared test of independence or G-test of independence.
2. Because the null hypothesis is true, the probability of getting a significant result is equal to the significance level, or alpha. Because we're using a significance level of 0.05 in this class, all you needed to write for full credit was 0.05 or 5%.
3. Order that turtle pokes its head up: ranked
walking speed: measurement
eaten vs. safe: nominal .
4. One nominal variable, dimly lit vs. dark; total sample size is less than 1000; exact test of goodness-of-fit.
5. Two nominal variables, HTPAP genotype, cancer vs. no cancer; total sample size greater than 1000 (635+725): chi-squared test of independence or G-test of independence..
6. Three nominal variables, kind of cat, adopted vs. not adopted, which shelter; Cochran-Mantel-Haenszel test.
7. Total number of salamanders: measurement
soil pH: measurement
number of dead logs: measurement
white oak vs. non-white oak leaf: nominal
human activity scale: measurement
amount of light: ranked .
Whether you consider the quadrats to be a nominal variable is a gray area for this kind of experiment; we'll talk about it later in the semester. You didn't get points off for omitting or including it.
8. 0.003 is the probability of getting a difference in mean milk production between grass-fed and hay-fed goats of 0.7 liters per day, or more, by chance if the null hypothesis is true.
9. Behavior: ranked
Sugar amount: measurement
10. Because 42.3% of the area is cars, The null hypothesis is that 42.3% of the poops will be on cars.. There is one nominal variable, car vs. asphalt, and the total sample size is less than 1000 (61+57), so the test is exact test of goodness-of-fit.
11. Two nominal variables, Gpi allele, beach; total sample size greater than 1000 (743+89+581+7): chi-square test of independence or G-test of independence.
12. Number of fireflies, measurement (because you don't observe the fireflies that aren't flashing, it's not the nominal variable flashing vs. non-flashing)
sand particle size: measurement
percent of area that is bare sand: measurement (because it's the percentage of the area, you're not counting individual grains and sorting them into sand vs. something else).
streetlight present or absent: nominal.
Again, whether you consider the locations to be a nominal variable is optional.
13. Two nominal variables, near vs. far from cedar tree, rust spot vs. no rust spot; total sample size less than 1000 (100+100); Fisher's exact test of independence.
14. One nominal variable, raccoon vs. feces vs. cheese; total sample size less than 1000 (32+19+12): exact test of goodness-of-fit .
The null hypothesis is that one-third of the dogs will roll in the dead raccoon, one-third will roll in the feces, and one-third will roll in the cheese.
15. Mannose concentration: measurement
oxygen content: measurement
16. An exact test is better when the sample size is small, because the P-value is more accurate than for a chi-square or G-test. We use the chi-square or G-test when the sample size is large, because the calculations for an exact test are difficult even for a computer, and because all three tests give about the same P-value when sample sizes are large. .
17. Two nominal variables, species of slug, water content (because just two values, 15 and 30%); total sample size less than 1000 (50+50); Fisher's exact test.
18. Two nominal variables, kind of corn (sweet vs. yellow dent), strain of borer (E or Z); total sample size greater than 1000 (800+940); chi-square test of independence or G-test of independence.
19. alpha: 0.05 (also known as significance level)
beta: 0.20 or power: 0.80 . You got points off for putting down both beta and power, because power equals one minus beta, you only need to pick one.
Effect size: 15 percent fewer deaths (the effect size for a test of independence is a percentage or proportion, not the number of deaths)
20. Make alpha larger; make beta larger (or power smaller); make effect size larger..
Return to the Biological Data Analysis syllabus