How would you calibrate a DIY PAR meter? Calibration is by far the hardest part.
Same way you would calibrate any other instrument - against a known standard.
The advantage of the Arduino/Sensor combo is that you're counting frequencies with which you're doing math. The math can be factored as necessary. You have control of the software that does the calculations.
And your probe consists of a clear plastic encased IC with attached leads built however you choose to consider durability. The IC is definitely not the standard three-lead part that most attach to their voltmeters and try to whiz by on.
Custom work is often within the realm of research, is it not? The idea is to get what you need to perform the task at hand, and to do so with verifiable, repeatable results. If you need multiple copies for all involved, you can purchase your "factory certified" model as the standard against which all of the custom builds are calibrated, leaving you with but the main, purchased meter to be sent elsewhere to be certified by a party outside your research group.
Of course, not all research is tightly budgeted. If your own budget allows for the expediture of $200+ a pop for each student, then by all means let loose with the money. But then you also lose the ability to teach adaptation in achieving the main goal as well as first-hand knowledge in the construction of your equipment and its limits and configurability.
But do not make the mistake of dumping something simply due to its "DIY" label without actually considering what it is or how it operates. Research history is as full of examples of amateurs and their projects making firm discoveries as it is with government-funded projects, if not more so...