Yeah, icing was only a peripheral factor. I have 2300 hours in the Q400, much of it across the Cascades, Sierras, Rockies, & Canadian Rockies in winter. It does just fine in ice, largely because it has an ungodly amount of power. Horizon did a good job of maintaining the boots, which helps a lot.
I studied the accident pretty closely at the time due to my experience in the airplane and wrote a series of in-depth blog posts about it (here are Parts
1,
2,
3,
4, and
5 if you're interested). But here's the cliff notes version: the crew was engaged in non-pertinent conversation for most of the flight, continuing well below 10,000, and neglected to do the descent checklist - including entering v-speeds - until a few minutes before landing. They had the Refs Speed switch set to "Increase" as per SOPs in icing conditions; on the Q400 this activates the stick shaker at 8° AoA instead of the normal 12°. They however, forgot to use icing speeds, which at most weights is a 10-15 kt increase. Colgan's training in this area was deficient; the NTSB interviewed about 30 Colgan pilots (including check airmen) after the accident and only a few could give the correct answer to when you use Ref Speeds Incr and when you using ice speeds.
Basically, the CA just decelerated from 170 kts very quickly when he was slowing to approach speed and forgot to call for landing flaps until they were fairly slow. They were still above the (incorrect non-icing) approach speed, but because of the refs speed switch being set to Incr & the landing flaps not being set yet, the plane hit 8° AOA, the stick shaker went off, the autopilot disconnect, and the CA hauled back on the yoke with about 20-30 lbs of force while bringing the throttles up to 80% torque and not touching them again (130% is available in an emergency).
The plane wasn't even close to stalled when the shaker went off. The speed deteriorated to 105 kts while pitching up to 30°, with the CA using aileron/spoilers to counteract the considerable torque/p-factor instead of rudder; therefore when it stalled, it immediately snapped right, to 105° bank. At this point the stick pusher was on but he overrode it with 60-120 lbs (!) of force the whole way down with one momentary interruption when they had somewhat recovered from the upset, but still had a huge AoA. It's true that the FO put the flaps up without being commanded during the initial stickshaker event, but given that the CA's handling resulted in nearly continuous AoAs over 25°, it's doubtful that leaving the flaps alone would have made a difference. After the upset, they basically had one chance to save it, but the CA continued to haul back on the yoke.
As to why the CA hauled back on the stick, many have suggested he was thinking tailplane stall, and I guess it's possible as he had previously flown the Saab, but honestly everything he did looks more like panicked reaction than any conscious decision. The way that Colgan (and pretty much everyone) taught stalls at the time resulted in having motor memory of 20-30 lbs of back force on the yoke during stall recovery, because you normally stopped trimming at a high airspeed, and were expected to maintain altitude during recovery. Well, if you use the exact same motor response when the autopilot trims all the way down to stick shaker, combined with 80% torque, you get a very quick 30° pitch up.
Colgan's training was infamously bad. The Q400 was a quirky airplane and IOE at Colgan was half of what Horizon gave pilots new to the airplane. Both CA and FO were tired from punishing commutes. FO had commuted across the country because she couldn't afford to live on the east coast and had moved back in with her parents. She also had a head cold and talked on the CVR about not daring to call in sick. CA of course had many checkride busts, a few of which he hadn't disclosed on his app. Was a former Gulfstream PFT kiddie. Yeah, the crew screwed the pooch pretty royally, but this was also a system accident if we ever saw one. Part of that system was CO pilots allowing their company to outsource their flying to a dirtbag operator. Part 117 and the ATP rule mean that dirtbags now can't be quite so dirtbaggy, but the system that put an airplane like the Q400 at an operator at Colgan is still in place and still being defended by certain parties within ALPA.