p-value is big while it is not supposed to

[Olivia Li]

Hi everyone, I have a problem considering the p-value in my stata17MP.
The t-value is 1.73 which means the p-value will definitely be smller than 0.5. But the p-value in my stata shows 0.083. (the degree of freedom is big)

Could you please tell me how to adjust the p-value? I suppose there is something wrong in the setting.
Thanks!!

[Felix Bittmann]

Keep in mind that Stata reports two-sided values. If you are used to looking this value up in Tables where the one-sided value is reported, divide by 2, which is then about 0.04. Also see https://www.stata.com/support/faqs/s...-coefficients/

[Andrew Musau]

You can also use the -ttail()- function to calculate the P-value. Note that the shape of the t-distribution depends on the degrees of freedom (DOF). As the DOF tend to infinity, the shape of the distribution approaches a normal distribution. So if say the DOF are 20, 100 and in your case 67,361, you have the following P-values (multiply by 2 taking into account Felix's point):

di 2*ttail(20, 1.73)
di 2*ttail(100, 1.73)
di 2*ttail(67361, 1.73)

Res.:

. di 2*ttail(20, 1.73)
.09903354

.
. di 2*ttail(100, 1.73)
.08671543

.
. di 2*ttail(67361, 1.73)
.08363486

And comparing to the standard normal distribution

 di 2*(1- normal(1.73))

Res.:

.  di 2*(1- normal(1.73))
.08363028

So not much difference at all.

[Olivia Li]

Keep in mind that Stata reports two-sided values. If you are used to looking this value up in Tables where the one-sided value is reported, divide by 2, which is then about 0.04. Also see https://www.stata.com/support/faqs/s...-coefficients/

Thanks ! Helps a lot.

[Olivia Li]

You can also use the -ttail()- function to calculate the P-value. Note that the shape of the t-distribution depends on the degrees of freedom (DOF). As the DOF tend to infinity, the shape of the distribution approaches a normal distribution. So if say the DOF are 20, 100 and in your case 67,361, you have the following P-values (multiply by 2 taking into account Felix's point):

di 2*ttail(20, 1.73)
di 2*ttail(100, 1.73)
di 2*ttail(67361, 1.73)

Res.:

. di 2*ttail(20, 1.73)
.09903354

.
. di 2*ttail(100, 1.73)
.08671543

.
. di 2*ttail(67361, 1.73)
.08363486

And comparing to the standard normal distribution

 di 2*(1- normal(1.73))

Res.:

. di 2*(1- normal(1.73))
.08363028

So not much difference at all.

Thanks for your guide ! Helps a lot!