Wasn't 2003 a mandatory year? That's the year legislation passed. At the very least, the first half was a mandatory year. Why did you pick 2003 for the start of your PennDOT fatality statistics?
Regardless, pre-law repeal year fatalities by themselves don't give enough information alone to determine the effectiveness of helmets.
I would submit that we need the number of all MC accidents per year pre- and post repeal to compare the percentage of accidents survived. (But even then, it wouldn't address the severity of the accidents or injuries.)
It would however, be a better measure than simply using raw numbers of fatalities or even adjusting for increased number of riders or registered bikes.
The pitfall you seem to be falling in is that any increase in fatalities is due entirely to fewer riders wearing helmets. There are a lot more bikes out there now, something like 75% more over a ten year period, and the VMT per rider is on the decline. That means there are a lot more inexperienced riders skewing the post law repeal stats.
I'll have to dig out the stats, and it will take a lot of time, but I am going to predict that the number of accidents/crashes/collisions (whatever you want to call them) has increased at a greater rate than the number of fatalities over the same period.
And finally, since fatalities are distributed about 50-50 between helmet wearing and non wearing groups, and that the NHTSA believes 37% of of the non wearers could have been saved, the maximum estimated reduction of fatalities would be 18.5%, not 50%. (I do think the NHTSA study is flawed, and that even 18.5% is high. They would be better off focusing on DUI and speeding issues. Save far more lives.)
Go back and read the posts. Some dude said they increased every year since 2003. I posted since 2003 to counter his claim. I have stated the stats for 1999 and stated I did NOT have the 2002 stats. I think 2003 was the first year for the PennDOT statistics.