The Building Blocks of Inference

In this lab you will use the following three packages:

library(tidyverse)
library(oilabs)
library(infer)

Examples, revisited

In lecture we’ve gone through several examples of hypothesis tests on the blackboard. Now you’ll get the chance to get your hands on the data and construct those tests.

1. Using the promote data set,
   • Write the null and alternative hypothesis in symbolic notation (in markdown).
   • Compute the observed test statistic.
   • Construct and save the null distribution using infer.
   • Visualize the null distribution.
   • Compute the p-value.
2. Using the millenials data set,
   • Write the null and alternative hypothesis in symbolic notation.
   • Compute the observed test statistic.
   • Construct and save the null distribution.
   • Visualize the null distribution.
   • Compute the p-value.
3. Using the curry data set,
   • Write the null and alternative hypothesis in symbolic notation.
   • Compute the observed test statistic.
   • Construct and save the null distribution.
   • Visualize the null distribution.
   • Compute the p-value.

Synthesizing the process

4. How would you expect the plots of your null distribution to change if reps were increased to 1000? Give it a try.

5. In which of the examples above would you be comfortable using the Normal curve to describe the null distribution?

Survey Inference

6. Return to the survey141 data and select one or two columns to study. Conduct a hypothesis test in the form that you’ve seen in this lab and be explicit about each of the 5 steps that you went through for previous examples. Additionally, please interpret the p-value: what does this mean in the context of our sample of Reedies and the population of Reedies in general?

Note that you may need to dust off your dplyr skills in order work with variables that have only two levels. One approach is to filter the data but the other is to use fct_recode or fct_collapse in the forcats package.