Hello Sean:
Perhaps it might be appropriate to take a fairly critical look at the "first principles" that appear to have been accepted without examination as the basis for some of this discussion.
For many years - particularly the time period from the 1950s through to the mid 1980s - a "first principle" that governed much of the decision-making process applicable to emergency and abnormal procedures was "Don't damage the aircraft". Examples of of this include slavish adherence to power limitations, and aircraft checklists that called for engines to be shut down (on multi-engine aircraft) at the first sign of any sniffle or out-of-limit condition from the engine.
For example (please bear with me for a moment whilst I digress and discuss multi-engine issues), in my own experience of over 5,000 hours of instructional experience given in full motion simulators, I very rarely encountered a pilot who would actually firewall the power levers when confronted with severe windshear on final approach. Instead, pilots would advance the power levers until reaching the first redline, and then do their best to cope with the windshear using maximum rated power. The result was very predictable (especially because I had control over the severity of the downdraft) - the aircraft crashed short of the threshold. The only positive outcome was that because the engines had not been overstressed, I suppose (speaking tongue-in-cheek here) they could be removed from the wreckage and used again in another aircraft.
After running this scenario with the above result, I would suggest to the crew that maybe we could try it again, but this time, to heck with the engine limits, just firewall the power levers and do your best. Often - not always, but often - they had enough power to escape the windshear without hitting the ground. The two engines were toast, the repair bill was going to be half a million dollars or more, but everyone was still alive.
Low oil pressure warnings and illuminated chip detectors are another example of incorrect assumption of first principles. For many years, pilots were taught that if a (turboprop) engine on a twin encountered low oil pressure in flight, or if a chip detector light illuminated, that engine should be shut down IMMEDIATELY - right NOW - lest the precious engine be damaged. A scholarly survey of the results of voluntary engine shutdowns in flight vs. actual engine failures in flight carried out by the Flightsafety Foundation in the early 1990s revealed that more accidents were happening as a result of sequela (loss of control, etc.) arising from the voluntary shutdowns than were happening as a result of actual failures of engines. As a result of this, procedures have now evolved - very few turboprop twins are now equipped with chip detectors, and the checklist procedures for those that have them have been changes to read "Monitor the engine, be prepared for a possible failure, otherwise, report to maintenance upon landing and repair before further flight". The very newest aircraft (e.g. a PC-12-47, notably a single) don't even display the chip detector warning if the aircraft is in flight - the aircraft suppresses the message until after landing.
You can see a change in the philosophy of "first principles" here, away from the old conventional wisdom "don't damage the aircraft" toward a new basic principle "don't damage the occupants'.
A similar example - again with multi-engine aircraft, which is where my personal experience lies - has to do with engine failures after takeoff in a twin. For many years, conventional wisdom (CW) was that you beat it around the circuit as quickly as you could and landed back at the departure airport. The result was a rushed procedure, a highly stressed pilot, no time to evaluate options and choose the best alternative - just crank it around the circuit and land, preferably within 3 to 4 minutes.
This CW has now been replaced by procedures that call for the pilot to carry out the necessary memory items (feather failed engine, add power on good engine), then continue to climb straight ahead on runway heading until a safe altitude has been reached - typically several thousand feet AGL - then, to call ATC and request a hold or a vector to no-where while the pilot assesses the situation and determines what the best course of action is. When a decision to land is made, the pilot carries out a full approach to the airport in as normal a manner as possible.
Why did that CW of 'race it around the circuit and land right away' develop? In part, because of a desire to protect precious machinery. But, quite tellingly, also in large part because of an unexpected influence from the ab-initio training industry: It's expensive to rent multi-engine aircraft when a pilot is getting that first multi-engine endorsement. A properly executed engine failure after takeoff in a twin takes about 20 minutes to complete - as can be seen in this video, which is a perfect example of how to do things correctly according to current best practice:
Thompson 757 EFTO. How many young pilots who are scraping the bottom of their wallets to pay for aircraft rental would be willing to spend this amount of time demonstrating how to properly handle an EFTO during their ab-initio training? Not many.
So - getting back to your original topic, which was whether to turn back or to land straight ahead following an EFTO in a single - I think you need to critically evaluate what the "first principle" behind your decision making process is. If it is to minimize damage to the aircraft, then trying to get back to the airport makes sense. But, if it is to assure the highest level of safety for the occupants, I think that a very strong argument could be made that landing straight ahead - which eliminates the risk of stall/spin in the turn, running short of potential energy and landing short of the runway, etc. - is the safer alternative.
Michael