I think the one classic automation versus raw data instrument interpretation accident that sticks in my mind, was the Thai Airways Airbus A310 that flew into a mountain at Kathmandu on 31 July, 1992. The accident was well illustrated in one of the excellent books by the Australian flight safety author Mac Job, called Air Disaster Volume 3.
At the beginning of a VOR/DME approach to runway 02 at Kathmandu in IMC, initially the flaps failed to extend. Instead of following the published missed approach procedure, and sorting things out in the holding pattern, the captain, in attempting to rectify the flap problem (which was successful a few minutes later), allowed the aircraft to wander in a large orbit until it finally took up a heading on autopilot towards an area of MSA 21.000 ft.
The captain maintained 11,500 ft in the mistaken belief that the aircraft was in safe sector where the MSA was 11,500 ft. The CVR revealed that pilots were heavily engrossed in staying heads down into the FMC trying to type in waypoints that were behind them and which could therefore not be easily seen on the MAP.
It was obvious that no notice was taken of VOR or ADF RMI readings which, coupled with the VOR/DME position would have immediately shown that the aircraft was in fact 24 nm NNE of the airport and not 24 nm south as the captain thought. Lack of simple raw data instrument rating skills was the cause of the accident which killed all aboard.