Converting p-values to z-values in a variable

[Jane Evans]

Hello, I am trying to find a way to transform a Stata variable of p-values to the corresponding z-value e.g. (0.025 --> 1.96). The only way I know how to do this conversion is through the display invnormal(p) command. However, I don't know how I can save this in a variable since display doesn't save the output anywhere (that I'm aware of). Any ideas?

Thanks!

[William Lisowski]

If p is a variable in your dataset containing p-values, then

generate z = invnormal(p)

will create a variable containing z-values.

[Clyde Schechter]

I don't understand the organization of your workspace for this. Do you have a data set that includes a variable containing some p-values and you want to create a new variable with the z-values? If so

gen z_value = invnormal(p_value)

does that.

Do you have a p-value that is stored in a scalar or local macro and want a corresponding scalar or local macro containing the corresponding z-value?
[/code]
local z_value = invnormal(`p_value') // LOCAL MACRO CASE
scalar z_value = invnormal(p_value) // SCALAR CASE
[/code]

Do you have a p-value that has just been produced by a command that leaves a p-value in r()?

local z_value = invnormal(`r(p)')

will save it in a local macro for you.

Basically, whatever the source of the p-value(s) is, and whatever kind of data structure you wish to store the result in, will be accomplished in ways like this, always relying on an expression using the -invnormal()- function, with an argument designated the source p-value and an appropriate assignment target on the left hand side, depending on how you want to save the result.

Added: Crossed with #2.

[Jane Evans]

Oh wow, I didn't think it was that simple! Thanks!

[Jane Evans]

Oh wow, I didn't think it was that simple! Thanks!