Exercise

Percentile method - bootstrap CI for slope

Alternatively, a CI for the slope can be created using the percentiles of the distribution of the bootstrapped slope statistics. Recall that a CI is created in such a way that, over a lifetime of analysis, the coverage rate of a CI is (1-alpha)*100%. If you always set alpha = 0.05, then the 95% confidence intervals will capture the parameter of interest (over your lifetime) 95% of the time. Typically, out of the 5% of the time when the interval misses the parameter, sometimes the interval is too high (2.5% of the time) and sometimes the interval is too low (2.5% of the time).

The bootstrapped estimates of slope, boot_slope, are loaded in your workspace.

Instructions

100 XP
  • Set alpha to be 0.05 (although for your own work, feel free to use a different confidence level).
  • Calculate the relevant percentiles needed to create the confidence interval.
    • The lower percentile cutoff is at half alpha.
    • The upper percentile cutoff is at one minus half alpha.
  • Create the confidence interval of stat using quantile() and the percentile cutoffs. Your interval ends should be named lower and upper.