We investigate the source of the discrepancy between R-matrix and distorted-wave (DW) collision strengths for J− J = 0-0 transitions in Mg-like ions, for example 3s 2 1 S 0 -3p 2 1 S 0 , as reported previously. We find it to be due to the neglect of coupling, for example via 3s3p1 P 1 , as done by most DW codes. We have implemented an option to account for such coupling as a perturbation within the autostructure DW code. This removes the discrepancy of a factor ∼10 and ∼100 for Fe 14+ and S 4+ , respectively, for such transitions. The neglect of coupling would have affected (to some degree) the atomic data for a few weak optically forbidden transitions in other isoelectronic sequences if they were calculated with DW codes such as FAC and HULLAC. In addition, we compare the Fe 14+ line intensities predicted with the R-matrix collision strengths against observations of solar active regions and flares; they agree well. For Fe 14+ , we suggest that the best density diagnostic ratio is 327.0/321.8 Å.