The controlled impact demonstration (CID) had two main objectives:
To evaluate antimisting kerosene (AMK) effectiveness on containing fuel fire propagation after a landing resulting in damages to fuel tanks and fuel ignition. This was the only initial objective.
To evaluate floor and seats behavior during crash (crash-worthiness). This objective was added as the initial objective offered a good opportunity for it.
The landing conditions were designed to damage tanks and ignite fuel. The impact was moderate only as "rhino" horns where installed immediately after at the impact point to make an incision in the fuel tanks and engine:
CID rhino barriers, source F.L. Fulton / NASA
23–25% doesn't tell about how many persons could have survived the crash mechanical effects, but how many could have evacuated before losing consciousness due to smoke, disregarding any impact injury/fatality.
Does that happen in actual accidents? Yes, in the British Airtours Flight 28M, 55 people died in an aircraft stopped on a taxiway with a punctured fuel tank.
Smoke and time for evacuating
When in the middle of a fire, the two deadliest hazards are not flames themselves but smoke and hot air. Smoke by itself brings deadly CO and reduces O2, but also triggers a coughing fit preventing, within a couple of minutes, further O2 extraction, leading to incapacitation.
The conclusion you mention only evaluates the number of persons able to evacuate the aircraft before losing consciousness. Related assumptions are detailed in the report, page 55:
Unfortunately there has been many fatalities during actual evacuation of cabins in fire, due to smoke. In particular there was an accident a few months only after the CID, this accident did happen during the takeoff attempt, the aircraft did not even leave the runway and no damage occurred except a puncture to a tank. Still 55 persons died because of the smoke:
What do we know about the survivability aspects of aircraft accident evacuations and how did the industry learn lessons in this area?
One significant accident that contained lessons for the aviation industry and led to changes occurred in 1985 at Manchester Airport in the United Kingdom.
During take-off, at a speed of approximately 125 knots, the left engine of a B737 suffered an uncontained failure which caused a fuel leak. The leaking fuel ignited and burned directly behind the left engine. The take-off was abandoned and the aircraft exited the runway on the right hand side onto a taxiway, which resulted in the wind directing the fire towards the aft fuselage.
The airport fire service attended the accident site promptly, but within 5 and a half minutes after the aircraft came to a stop, of the 137 passenger and crew onboard, 55 persons had lost their lives.
“The major cause of the fatalities was rapid incapacitation due to the inhalation of the dense toxic/irritant smoke atmosphere within the cabin, aggravated by evacuation delays caused by a forward right door malfunction and restricted access to the exits.
More information on this accident on Wikipedia.
Incidentally this accident also tells us something important about cabin crew members role, training, and effectiveness to help people evacuate the cabin at their own risks. In the Manchester accident, cabin crews, which two members died in the accident, were awarded a Queen's Gallantry Medal along with firefighters. At the time they were officially known as "steward/ess".
Floor and seats behavior
A specific experiment was conducted on the seats, to compare standard seats of that time (-9g forward), and improved ones (-18g forward).
Both FAA and NASA concluded CID was a survivable accident (Summary report, page VII) and in the detailed report (on page 31) that the impact conditions were overall mild (thus preventing to fully compare seat performances).
The test objectives which involved the recording of data pertaining to the airframe structural floor loads and seat occupant responses during impact were met. However, because of the mild impact conditions, the objective to demonstrate the performance between the standard and modified seat designs was not completely satisfied.
Dummy bodies set at various locations in the flight deck and the cabin were severely shaken (as visible in this recording, starting at 1:48), but these accelerations were deemed survivable.