Here are the answers to exam 2. 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 is worth 25% of your grade for the class.

**1. **One nominal variable, earthworm or no earthworm in the first box the bird pecks at; theoretical null expectation of 35% (7/20) worms; sample size less than 1000: **exact test of goodness-of-fit**

**2. **Two nominal variables, habitat type and crab type (juvenile male, juvenile female, adult male, adult female); total sample size greater than 1000: **chi-square or G test of independence.**. Some of you considered crab type to be two nominal variables, juvenile vs. adult and male vs. female, which actually could make sense; because the question was confusing, everyone got full credit, that's why there's an "X" not a check mark.

**3. **Three nominal variables, drug vs. placebo, snoring vs. not snoring, hospital: **Cochran-Mantel-Haenszel test.**

**4. ****Alpha: pick 0.05 because it is commonly used in most areas of biology
Power: pick a fairly high number, like 0.80 or 0.90, pretty arbitrary
Effect size: Calculate based on the potential extra winnings from going X mph faster vs. the cost of five apples per day
Standard deviation of running speed: get from previous literature or a pilot study**.

**5. **One measurement variable, snake-predator distance; one nominal variable, identity of predator: **one-way anova or Welch's anova**.

**6. ** **The most obvious confounding variable is the order in which predators are presented; the snake might be getting tired of rattling (or extra-nervous) by the time you get to the badger. You could fix this by randomizing the order in which the predators are presented, or using a different snake for every predator-snake encounter.**

**7. ****In 50 experiments testing null hypotheses that are all really true, you will expect about 2.5 experiments to give you significant (P<0.05) results. If you are gullible, you will conclude that no-touch reiki (which really is a thing; Google it) works on a few diseases. If you are smart, you will conclude that a few significant P-values are false positives and require a much smaller P-value to convince you that no-touch reiki works. **

**8. **One ranked variable, order in which the turtles hatch; one nominal variable, eaten by bird, fox, raccoon, or not eaten: **Kruskal-Wallis test**.

**9. ****Use the median when data are highly skewed**.

**10. **One nominal variable, species of firefly; one measurement variable, flashing rate; experiment is unbalanced and heteroscedastic: **Welch's anova**.

**11. ****For your big experiment, you should use as many rabbits as you can afford, with a small number of samples per rabbit**.

**12. **Two nominal variables, soil type and germination (if you plant 490 mustard seeds and count 320 mustard plants, you know that 170 seeds didn't grow; for each soil type, you have 490 observations of grew/didn't grow); total sample size greater than 1000: **chi-square or G test of independence**

**13. **One nominal variable, field; one measurement variable, phosphate:** one-way anova or Welch's anova.**

**14. **Two nominal variables, E-strain vs. Z-strain, sweet corn vs. yellow dent corn; total sample size greater than 1000: **chi-square or G-test of independence**

**15. ****It's wrong to look at the P-values of the anova while you're trying different transformations, because you will be tempted to pick a transformation that gives you a significant P-value, and that increases the chance of a false positive.**

**16. ****Standard deviations shows how spread out the individual observations of purring rate are.
Standard error shows how close each estimate of the mean is likely to be to the true mean.**

**17. ****You should do a Tukey-Kramer test, so you can see which pairs of cat types have significantly different purring rates.**

**18. **One nominal variable, green vs. blue; theoretical null hypothesis of 50% in each side; total sample size greater than 1000: **chi-square or G-test of goodness of fit**