"you can then very roughly estimate that about 17.8% of winners will be from those two states."
A recent thread in the Jackpot forum listed PB winners by state. I don't know for sure that it's accurate and complete from the beginning of the game (it seems close enough), but it says that CA and NY account for 5.2% of winners. That's well below17.8%, but neither state offered PB until well after the game started in 1992. NY joined in early 2010, and CA joined in April 2013. AFAIK, PB was always twice a week, so NY sold tickets for about 37% of the drawings and CA sold tickets for about 26% of the drawings. If we just call that 1/3 of the drawings and extrapolate it to the entire history of PB then CA and NY would account for about 15.6% of winners.
Of course other states joined at different times and there has never been a period where the entire population lived in a state where they could buy PB tickets, and the per capita sale of tickets probably isn't the same in every state so it's unlikely that CA and NY should really account for 17.8% of sales. Another factor is that sales in CA and NY have been skewed towards higher odds that reduce the number of winners relative to sales; there were typically more winners per year before CA and NY joined the game.
Short of having actual historical sales figures there's no way we can make a really accurate calculation of the theoretical probability distribution of winners, but based on what we do know there's nothing to suggest that the number of winning tickets sold in NY and CA isn't very close to what we'd expect from purely random probability.
Indiana and Missouri, OTOH, do seem to be much different than what random probability would suggest, with 10.2 and 8.1% of total winners. Both states have sold the game right from the start, but if Indiana ever sold the 10.2% of tickets that would correspond to the percentage of winning tickets from IN that didn't last long. Current sales in IN account for roughly 2% of the total, according to lottoreport,com and the oldest figures I found (early 2010) were only slightly higher. OTOH, the odds are high enough that every winning ticket is an anomaly, so I don't find it surprising that some states depart significantly from what random probability suggests. If you flip a coin 1000 times the chances of getting 3% more or fewer heads than the expected 50% is pretty small, but the chances of 3% more or fewer winners in a game with odds of 50 to 30 million to 1 are much greater. Still, if any of you conspiracy nuts want to rail at something that doesn't sound right there are far better candidates than NY and CA.