Here are the answers to practice exam 4. The answer is in **bold**; my explanation, which you don't need to write on the exam, is in regular type.

**Where two correct answers are shown for these practice questions (such as chi-squared or G-test of independence), you must only write down one on the actual exam.**

- One measurement variable, insulin sensitivity index; one nominal variable, genotype (individual is not a nominal variable because only one measurement per individual);
**Fisher's one-way anova or Welch's one-way anova** **There may be many studies that were unpublished because they didn't show an effect.**- Two nominal variables, position of treat, caught or not; total sample size less than 1000:
**Fisher's exact test** - One nominal variable, cell type; one measurement variable, fluorescence:
**Fisher's one-way anova, Welch's one-way anova, Student's two-sample t-test or Welch's two-sample t-test** - One nominal variable, maggot species; one measurement variable, tetracycline dose (measurement because 11 different amounts); if there's a cause-and-effect relationship, tetracycline dose affects percentage of
*Phaenicia regina*:**simple logistic regression** - One nominal variable, fertilized vs. unfertilized; two measurement variables, number of walnuts, diameter of tree:
**ancova** - Two measurement variables, number of apples eaten, number of doctor visits:a
**correlation/linear regression** - One measurement variable, amount of mRNA for MPI; one nominal variable, normal vs. cancer:
**Fisher's one-way anova, Welch's one-way anova, Student's two-sample t-test or Welch's two-sample t-test** - One measurement variable, distance from black walnut tree; one nominal variable, damaged vs. undamaged leaf; if there's a cause-and-effect relationship, distance affects leaf damage:
**simple logistic regression** - Two measurement variables, length of ulna and size of flexion/extension moment arm:
**linear regression/correlation** - One nominal variable, red vs. green; theoretical expectation of 1:1 ratio if null is true; sample size less than 1000:
**exact test of goodness-of-fit** - One measurement variable, arsenic concentration; one nominal variable, Halomodadaceae or not; if there's a relationship, arsenic concentration affects the proportion of Halomondadaceae:
**simple logistic regression** - One measurement variable, corn borers per quadrat; one nominal variable, crop type:
**Fisher's one-way anova or Welch's one-way anova** - One measurement variable, time until first gull arrives; one nominal variable, food type:
**Fisher's one-way anova, Welch's one-way anova, Student's two-sample t-test or Welch's two-sample t-test** - One measurement variable, sodium level; two nominal variables, cell line, gramicidin or no gramicidin; each cell line in combination with gramicidin or no gramicidin; multiple measurements per combination:
**two-way anova with replication** - One measurement variable, projection length; three nominal variables, challenged vs. unexposed, embryo, pigment cell:
**nested anova**. - One nominal variable, right vs. left, theoretical expectation of 50:50 ratio, total sample size greater than 1000:
**chi-square test of goodness-of-fit**or**G-test of goodness-of-fit**. - Two nominal variables, genotype and cancer type, total sample size greater than 1000:
**chi-square test of independence**or**G-test of independence**. - Three measurement variables, vigilant vs. not vigilant, edge vs. interior, 11 different flocks:
**Cochran-Mantel-Haenszel test** - Two measurement variables, shear modulus and range-of-motion; one nominal variable, injury type; goal is to compare different regression lines of shear modulus vs. range of motion:
**ancova** - One measurement variable, running speed; two nominal variables, food type and mouse identity (since there are multiple measurements per mouse); each mouse has only one food type:
**nested anova** - One measurement variable, maze time; two nominal variables, mouse identity and blindfold/noseplug/normal; each mouse in combination with each condition, run 5 times:
**two-way anova with replication** - Two measurement variables, time (since you sample from 11 months), taste:
**linear regression/correlation** - One measurement variable, latitude; one nominal variable, right vs. left-handed; if there's a relationship, latitude affects handedness, handedness doesn't affect latitude:
**simple logistic regression** - One measurement variable, weight of pollen; two nominal variables, orchard type, hive identity (since you have multiple measurements per hive):
**nested anova** - Three measurement variables, age, weight, and height; one nominal variable, defective vs. normal sperm:
**multiple logistic regression**. - Two nominal variables, embryonic vs. adult stem cells, cartilage vs. undifferentiated; total sample size greater than 1000:
**chi-square test of independence**or**G-test of independence**. Note that because you start with 750 undifferentiated cells and count how many cells differentiated into cartilage cells, you know (by subtraction) how many cells didn't differentiate into cartilage. - Two measurement variables, foot length, time on board:
**linear regression/correlation** - One measurement variable, speed; two nominal variables, time of testing and identity of snail (since there are multiple measurements per snail); one measurement per snail/time combination:
**Two-way anova without replication**. - One measurement variable, number of goldenrod plants; one nominal variable, presence or absence of praying mantis; if there's a cause-and-effect relationship, number of goldenrods affects presence of praying mantis:
**simple logistic regression**