P-value is not according to t-statistics

[Farheen Akram]

Respected researchers
i ran the panel regression using the "xtscc" command and the probability is behaving strangely. like the t is 1.96 and p>|t| is 0.098. i think it should be equal or less than 0.05.
anyone please help me in this regard.

[Nick Cox]

Depends on how many degrees of freedom, you have. Some mental arithmetic leads to

. di 2 *  ttail(6, 1.96)
.09769515

so that I guess you have 6 df.

Here is a check

. di 2 *  ttail(6, 0.43)
.68221048

So, looks like 6 df.

Hence a secondary question is whether your dataset is large enough to do this well. But then the results are telling you that any way.

[William Lisowski]

Adding to Nick's answer, I wonder if the question perhaps reflects confusion between the two-tailed p-value, which is reported as 0.098, and the one-tailed p-value, which would be 0.049. That is, "P>|t|" is reporting the probability that a random variable following a t distribution with 6 degrees of freedom is either >1.96 or <-1.96; each of which has probability 0.049 so together they have probability 0.98.

[Farheen Akram]

Depends on how many degrees of freedom, you have. Some mental arithmetic leads to

. di 2 * ttail(6, 1.96)
.09769515

so that I guess you have 6 df.

Here is a check

. di 2 * ttail(6, 0.43)
.68221048

So, looks like 6 df.

Hence a secondary question is whether your dataset is large enough to do this well. But then the results are telling you that any way.

Thanks Nick

my data set based on 1251 observations (unbalanced panel data, with 7 year and 183 N).