This is one of the videos of the crash: 727 Test Crash. It seems it was more of a hard landing without gears. Why wouldn't the experimenters have made it more of a really hard crash if they needed to study the effects of a crash on a plane? Would slamming down the plane really hard (maybe even nose down) given a more realistic scenario and more valuable data rather than making it soft?
If the plane crashes with high vertical speed, the decelerations involved are so insanely huge that there is absolutely no way to make it survivable. That is well explained in Even after years of research, why are planes unable to keep passengers alive in case of a fiery crash?.
However in many cases the aircraft is still, at least partially, controllable, so a crash-landing can be made. And that is the case where survivability can be improved by careful design, so that is the case they were testing.
This is also relevant for accidents on landing where the aircraft simply does not have that much vertical speed. Together these scenarios cover more accident cases than the out of control scenario.
Would slamming down the plane really hard (maybe even nose down) given a more realistic scenario and more valuable data rather than making it soft?
Gone, everything is gone then. What are you going to try to improve from the outcome of such a test? Forget the real airplanes I have never been able to salvage a model aircraft that crashed nose down for any reason whatsoever.
I have on many times crash landed a model safely that had its landing gear malfunction and that's the practice that gives you valuable lessons for future landings. If you had an airplane without a landing gear and you wanted to study how can you improve chances of its survivability you would want to make a survivable attempt in the first place. You wouldn't want to crash it hard deliberately and then try to pick up pieces of the landing gear to find out what could you have done better.
The basic principle of the airworthiness regulations is risk analysis. If you can demonstrate that a particular scenario has a sufficiently low risk of ever happening, you don't need to consider the consequences if it does happen. As an order of magnitude, "sufficiently low risk" means you might expect something to happen at most once during the full lifetime of all the aircraft of that type - which may be a period of 50 years or more for a popular aircraft type where thousands of aircraft were manufactured.
Landing a partly controllable aircraft with the gear up is more likely to happen than that "extremely rare" probability, so you need to demonstrate what the consequences are, hence the test. But there are thousands of possible but extremely unlikely scenarios that could (and most likely would) lead to the loss of the aircraft. For those scenarios, you design to reduce the risk of the event happening at all, not to deal with the consequences if it does happen.
A topical example: you don't try to design planes that will survive explosions in mid flight. Rather, you try to design systems to ensure that planes don't carry explosive material in the first place.