You have two opportunities to master randomization-based hypothesis testing. Email me your submission (pdf format, please). I will grade it and give you feedback within 3 business days. The last submission that I will consider for mastery must be submitted to me by noon on Friday, December 22. If you want to only do one submission/attempt, you can submit it as late as Friday, December 22 at noon, though I do not recommend this approach. I highly recommend submitting your first attempt by Monday, December 18th.
Select one question from the following list. Answer the research question using randomization-based hypothesis testing. Be sure to include your hypotheses, the test statistic from the original data set, a visualization of the histogram and the test statistic, the p-value, the decision (whether or not to reject the null in favor of the alternative), and a conclusion in the context of the research question. Use a significance level of .1.
There are two types of dusky mountain salamanders: white side and rough butt. Male and female salamanders of both types were collected. Scientists then paired salamanders and observed whether the pair mated. The following code creates a new variable that splits the data into two categories: pairings that involved a white side female and a rough butt male (TRUE) and those that don’t (FALSE).
library(glmm)
data(salamander)
attach(salamander)
salamander$CrossWR <- factor(Cross=="W/R")Are white side females and rough butt males mate less likely to mate than other pairings? That is, calculate the proportion of CrossW/R salamanders that mated. Then calculate the proportion of the remaining 3 types of crosses. Use this difference as your test statistic.
library(alr3)
data(blowdown)In 1999, a humongous storm created intense winds that blew down trees in the Boundary Waters Canoe Area. Determine whether the median diameter of blown-down trees differed from the median diameter or trees not blown down.
library(glmm)
data(murder)A bunch of people were asked how many homicide victims they know. Do the medians differ for black and white respondents?
library(glmm)
data(murder)A bunch of people were asked how many homicide victims they know. Categorize the respondents into two categories: those who know a single homicide victim and those who know multiple homicide victims. Does the proportion of white respondents who know a single homicide victim differ from the proportion of black respondents who know a single homicide victim? I added a new variable in the murder data-set (constructed in the following code). TRUE means the subject knows just one murder victim. FALSE means the subject knows multiple murder victims.
attach(murder)
murder$justone <- factor(y == 1)library(alr3)
data(cathedral)Does the median height of Gothic cathedrals exceed the median height of Romanesque cathedrals?
library(resampledata)
data(BookPrices)Does the median book price for Math & Science classes exceed the median book price for Social Sciences courses?
library(resampledata)
data(Cuckoos)Does the median number of eggs laid by wrens exceed that of robins?