I loathe how statistical instruction privileges obtaining a magical p-value by reference to an area underneath the standard normal curve, only to botch what the actual z-value is corresponding to the magical p-value. This simple function converts the p-value you want (typically .05, thanks to R.A. Fisher) to the z-value it actually is for the kind of claims we typically make in inferential statistics. If we're going to do inference the wrong way, let's at least get the z-value right.

p_z(x, ts = TRUE)

Arguments

x

a numeric vector (one or multiple) between 0 or 1

ts

a logical, defaults to TRUE. If TRUE, returns two-sided critical z-value. If FALSE, the function returns a one-sized critical z-value.

Value

This function takes a numeric vector, corresponding to the p-value you want, and returns a numeric vector coinciding with the z-value you want under the standard normal distribution. For example, the z-value corresponding with the magic number of .05 (the conventional cutoff for assessing statistical significance) is not 1.96, it's something like 1.959964 (rounding to the default six decimal points).

Details

p_z() takes a p-value of interest and converts it, with precision, to the z-value it actually is. The function takes a vector and returns a vector. The function assumes you're doing something akin to calculating a confidence interval or testing a regression coefficient against a null hypothesis of zero. This means the default output is a two-sided critical z-value. We're taught to use two-sided z-values when we're agnostic about the direction of the effect or statistic of interest, which is, to be frank, hilarious given how most research is typically done.

Examples

library(stevemisc) p_z(.05)
#> [1] 1.959964
p_z(c(.001, .01, .05, .1))
#> [1] 3.290527 2.575829 1.959964 1.644854
p_z(.05, ts=FALSE)
#> [1] 1.644854
p_z(c(.001, .01, .05, .1), ts=FALSE)
#> [1] 3.090232 2.326348 1.644854 1.281552