The first result, as noted at the bottom of the first table, uses the pooled estimate of the standard error of the mean. The second table computes statistics, including the standard error, separately for each level of the categorical factor. That is, the estimates of the standard error in the second table are not pooled.