Null Distribution of SD not Appropriately Calculated

[davidaarmstrong]

The problem

Using the infer package, when trying to calculate the null distribution of the standard deviation, the wrong values are produced. The problem is that the bootstrap() function that is called by generate() does not impose the null hypothesis condition when the hypothesis is about the standard deviation (like it does for the mean and median).

Reproducible example

In this example, the histogram should be centred at 5 - the null hypothesized value for sigma. Instead, it's centred around the empirical sd of the raw x variable used as input. For a slightly longer discussion, see this question that I answered on SO.

library(tidyverse)
library(infer)
set.seed(207)
dat <- tibble(x = rnorm(100, mean = 2, sd = 2))
sd(data$x)
#> Error in data$x: object of type 'closure' is not subsettable
dat %>%
  specify(response = x) %>%
  hypothesise(null = "point", sigma = 5) %>%
  generate(reps = 1000, type = "bootstrap") %>%
  calculate("sd") %>%
  visualize()

^{Created on 2024-01-26 with reprex v2.0.2}

[simonpcouch]

Thanks for the issue!

Duplicate of #127 and #464. You may find the discussion in #127 eliciting. :)

[github-actions]

This issue has been automatically locked. If you believe you have found a related problem, please file a new issue (with a reprex: https://reprex.tidyverse.org) and link to this issue.

[simonpcouch]

Woops, wrong reference. That commit should read #532.