On Tue, 17 Jan 2012, Henrique Andrade wrote:
I think I found a small problem in Gretl correlogram. In the file
attached,
if you calculate the correlogram of the series "uhat1" (Variable ->
Correlogram) you will see a confusing behavior.
In the ACF plot there is not any red bar crossing the blue lines, but in
result window we can see asterisks in the 7th and 8th lags.
If the impulses cross the blue lines on the graph that's
saying that the ACF values are significant at the 5 percent
significance level, while the single asterisks for lags 7 and
8 represent significance at the 10 percent level. As in
gretl's model output, 0.01 < p-value < 0.05 is represented by
two asterisks, and p-value < 0.01 by three.
So what you're seeing is quite consistent.
Allin Cottrell