Many authors have looked for the relations between the g factor and cogniti
ve tasks in order to find out the nurture of the g factor (Jensen, 1998). N
evertheless, some methodological problems must be solved to understand if t
he individual differences in g are differences in cognitive processes. One
of them is the low reliability of the scores in this kind of tasks. On the
grounds of the standard error of the ability level estimation, we hypothesi
zed that the correlations between the g factor and a working memory task wo
uld be larger in the group of the subjects better measured (less standard e
rror). The results support this hypothesis.