Asymmetric Icing - the Cause and the History
Quite obviously (IMHO) due to the asymmetric build-up of icing caused by both engine's props rotating in the same direction (think it through). I'd be amazed if it was due to engine failure.
They now need to put a scale model in an ice-distributor wind tunnel with the engines being electrically driven i.e. electric motors to drive the props at proportionate (to model-size) speeds) - and see just what the asymmetric ice loading effect becomes over time (in terms of aerodynamic drag differential (L to R), asymmetric lift and weight). By having the flight control loads monitored, they'd be able to determine what the increased stall speeds become and thus the much higher speed at which control is lost (i.e. the inevitable spinning auto-rotation onset speeds-whether on autopilot or not).
All of this was apparent after Roselawn Indiana and the ensuing icing crashes - but no-one bothered to look into asymmetric wing, tail and fuselage ice-loads induced spanwise by "same direction" prop rotation. Maybe now they will..... but wouldn't be surprised if they didn't.
Like wiring-initiated fires, all icing crashes lose their critical evidence post-crash. The problem with rain-ice or freezing rain (or SLD as it's called less alarmingly by the fAA/NTSB) is that it "hits and sticks". It hits (and sticks) more on one side than the other due to prop slipstreaming. That asymmetry is insidious. It's particularly lethal if the autopilot is left to quietly "soak up" that asymmetric ice-loading. Ultimately the autopilot will run out of compensatory trim-loading and be itself over-powered. That can happen quite quickly even in cruise - as the stall speed rises very rapidly in the latter stages of ice build-up. But the REAL problem is that the stall speed difference between the LH wing and the RH wing can be as much as 25 knots - all due to the spanwise distribution of the ice (both its mass centroid and its aerodynamic drag increment on each wing) being so different, left to right. Once that RH wing stalls, the aircraft will roll rapidly and the instinctive pilot input of opposite aileron (to counter the roll) will only embed the aircraft in the then inescapable condition of autorotation (i.e. spinning).
You have to wonder how many turbo-props approach that terminal condition each day, across the globe. The few that reach it and fail to recover just become another non-specific "control loss" statistic. It's less about the Dash-8 and more about the vulnerability of any turbo-prop type that is licenced to operate in icing conditions.