Another factor is we tend to think that if we lose power, we have the ability to act RIGHT NOW.
In real life, count on about 3 seconds of "deer in the headlights" before ANY action is taken. And think about what's happening to airspeed in that 3 seconds.
A turn begun at that point is very likely to go bad in a hurry.
True, though 3 seconds is forever. I consider myself pretty slow on the reaction time thing but I believe from my experience that part of the problem is that one will tend to act (incorrectly) in less than that time. But no data to support that so...
What I observed when practicing this in my old Maule is that if I pulled the power simulating a total and sudden power loss during climbout, reacting immediately was a bit of a problem airspeed wise. Given the deck angle, I wanted to dump the nose. That was a mistake. The plane is already roughly trimmed for the right speed and the nose falls by itself if you do little pitch wise.... a non-obvious response.
The proper response in that particular plane with that particular setup was to not dump the nose, but immediately bank into a steep turn and maintain more back pressure than expected (because of the bank angle) just like you would on a power-off steep turn. The steepness of the turn required is the difficult part.
If there is a xwind, you must also turn into it. Turns out that comes pretty naturally because you've already crabbed into it when under power so unless some external factor makes you favor a turn downwind, you kind of pick the right direction naturally.
After several practice attempts, I was surprised to see how little altitude would actually be lost when the whole maneuver is anticipated (no response time) and the reaction is optimized. Of course any obstruction really changes the equation. And a narrow runway requires a good bit more turn than simply a 180.
I'm not sure a sudden and total loss of power is very realistic as others have mentioned. And an optimized response is no more likely than a perfect engine out recovery on a multi if it's not practiced on a regular basis. I for one, have never practiced it in my current plane - 350 hours in. I may go out and do that today along with some 'landing straight ahead' visualization.
But I think it's worth practicing at least once just to get some part of the picture in one's head. Dumping the nose in the Maule cost more than I ever would have thought. The turn part I knew and understood from glider flying. Doing countless low altitude turns in a heavily ballasted glider provides some insight too - the speeds must higher due to the weight but the bank angle still must be steep in order to get it around. High density altitudes can change the visual picture significantly. Again, non-obvious unless practiced.