I agree with W8 on the ATR, it's had a very chequered history, whilst it has good economic credentials it has had a lot of accidents and incidents due icing.
Some excerpts from a few different online sources;
Roselawn accident;
3.2 Probable Cause
The National Transportation Safety Board determines that the probable causes of this accident were the loss of control, attributed to a sudden and unexpected aileron hinge moment reversal that occurred after a ridge of ice accreted beyond the deice boots because: 1) ATR failed to completely disclose to operators, and incorporate in the ATR 72 airplane flight manual, flightcrew operating manual and flightcrew training programs, adequate information concerning previously known effects of freezing precipitation on the stability and control characteristics, autopilot and related operational procedures when the ATR 72 was operated in such conditions; 2) the French Directorate General for Civil Aviation's (DGAC's) inadequate oversight of the ATR 42 and 72, and its failure to take the necessary corrective action to ensure continued airworthiness in icing conditions; and 3) the DGAC's failure to provide the FAA with timely airworthiness information developed from previous ATR incidents and accidents in icing conditions, as specified under the Bilateral Airworthiness Agreement and Annex 8 of the International Civil Aviation Organization.
Contributing to the accident were: 1) the Federal Aviation Administration's (FAA's) failure to ensure that aircraft icing certification requirements, operational requirements for flight into icing conditions, and FAA published aircraft icing information adequately accounted for the hazards that can result from flight in freezing rain and other icing conditions not specified in 14 Code of Federal Regulations (CFR) Part 25, Appendix C; and 2) the FAA's inadequate oversight of the ATR 42 and 72 to ensure continued airworthiness in icing conditions.[5]
In short ATR knew there was problems with the aircraft in icing and did not promulgate this to operators.
While the ATR 42 and ATR 72 aircraft are now compliant with all icing condition requirements imposed by those 18 ADs, the de-icing boots still only reach back to 12.5% of the chord. Prior to the accident, they had extended only to 5% and 7%, respectively. They still fail to deal with the findings of the Boscombe Down tests, conducted by British regulators, which demonstrated that ice could form as far back on the wing as 23% of the chord, and on the tail at 30% of chord. Both percentages remain well beyond the limits of the extended deicing boots, installed in compliance with those FAA ADs.
The fix to the wing/de-ice boots did not even address half the recommendation of the investigators.
It is likely that the lack of further ATR icing accidents is attributable to the changes in pilot operating procedures, as well as the moving of those aircraft to operating areas where severe icing is not a problem, rather than to the modest extension of the de-icer boots to 12.5% of the chord.[7]
None of the other turbo-props operated in Australia have this record.
If you think the problem was fixed this is a summary of the accident flight Aero Carribean 883, November 2010;
The Instituto de Aeronáutica Civil de Cuba took responsibility for investigating the accident in Cuba with assistance from ATR and the French Bureau d'Enquêtes et d'Analyses pour la Sécurité de l'Aviation Civile (BEA).[19][20] On November 16 the investigation authority presented the final report following analysis of the flight data and cockpit voice recorders. It concluded that a combination of extreme weather conditions caused severe icing which along with bad crew operating procedures ultimately led to the loss of the aircraft in a similar set of circumstances to the American Eagle Flight 4184 incident which occurred with the same type of aircraft in 1994. ATR confirmed that the aircraft was in optimal technical condition.[citation needed]
The report available online for TransAsia Airways cargo flight 791 is worth a read as well. 2002 accident very similar circumstance to Roselawn.
The 2012 UTAir accident in Russia probably had a similar cause but the ice accumulation possibly happened prior to take-off.
One thing is absolutely clear with the ATR, that is you must identify and exit severe icing very quickly or bad things can happen. Other Turbo-props like the Dash and SAAB are susceptible to icing but only to the normal stall issues, all problems these aircraft have had is too low speed with ice. The ATR problem is not a normal stall, it is a control hinge reversal that renders the aircraft almost uncontrollable if not accounted for early on, this can happen at high speed. A big problem with high wing aircraft like the ATR is you can not see how much ice is accumulating on the upper surface of the wing and where. The consistent rate of accidents following Roselawn and possibly numerous undisclosed incidents shows that if mishandled in icing this aircraft is still very unforgiving. If you compare this to the various Dash models there are very few instances of problems in icing, different wing design, these aircraft can motor through large amounts of icing and only need to change level when speed becomes an issue. This also accounts for the big difference in accidents, the ATR-72 having 7 fatal accidents (4 icing related) vs the Dash 8 300/400 having only two (all crew related). I did not include the ATR 42 and Dash 8 100/200 but the ratio is similar, but far less due icing.
Hopefully the 500/600 series ATR have modifications that fix the problem and it never shows again.