Critical values are used for construction of parametric CIs.
For a two-sided 95% CI, they are the x-values that correspond to the locations on the x-axis, beyond which 5% of the total probability density lies.
In other words, the shaded regions in the tails of the figure below add up to an area of 5%.
In yet other words, the shaded area in the center has an area of 95%.
Critical values for the Standard Normal distribution are also called critical Z values.
A 95% CI constructed from a Standard Normal distribution has critical Z values of \(\pm 1.96\):
How do I know?
I can use the qnorm()
function to find the 2.5% and
97.5% quantiles of the standard normal:
qnorm(c(0.025, 0.975))
## [1] -1.959964 1.959964
Note that to calculate the quantiles I used endpoints of 0.025 and 0.975 instead of 0.05 and 0.95. Why?
Recall the t-distribution with it’s parameter: the degrees of freedom.
Q2 (1 pt.): Consult the help entry for
qt()
and calculate the critical values of a 95% CI for df =
10. Show the R-code you used to perform the calculation.
Q3 (2 pts.): How many degrees of freedom are required for the 2.5% lower critical value of a t-distribution to match the 2.5% lower critical z-value (from the standard normal) to within one decimal place (i.e. -2.0)? Show the R-code you used to perform the calculation.
Q4 (1 pt.): How many degrees of freedom are required for the 2.5% lower critical value of a t-distribution to match the 2.5% lower critical z-value (from the standard normal) to within two decimal places (i.e. -1.96)? Show the R-code you used to perform the calculation.
Recall the general procedure for constructing a CI? Check out the last section of slide deck 5 if you need a refresher.
Suppose you know that the sample standard deviation for a group of 50 measurements is 3.14. The mean value is 10.0.
Q5 (2 pts.): What are the critical t-values for a 95% CI on the mean?
Q6 (3 pts.): Construct the interval. Show the R-code you used to perform the calculation.
Upload your group’s answers to the questions to Moodle