Hi all,

Ok, this is not such a great topic but I'm going to submit it anyway...

In an estimation output, gretl uses the expression "using observations...", for example:

for time series,

"Model 1: OLS, using observations 1952-1981 (T = 30)" (1)

or, for undated dataset,

"Model 1: OLS, using observations 1- 30” (2)

And this does make sense when all observations are in a row. When missing/uncomplete observations exist, gretl also provides additional information:

"Missing or incomplete observations dropped:" (3)

Now, when data are sub-sampled on a conditional criteria), (1) and (2) are (always?) reframed, for example,

"Model 1: OLS, using observations 1-13" (4)

and here "using observations..." loses its meaning given that the actual observations used in the estimation are not in a row anymore.

What if “using observations” in (4) be replaced simply by the number of observations used and additionnal info like (4) be provided informing the user that sub-sampled data are used. This will make a clear difference with (1) and (2)* the only cases when all observations used are in a row.

*Note than (2) is correct even for time series sub-sampled data but yet with all observations in a row.

Thank you for reading,

Best,

Artur