Actually, I wonder what "unhappiest statistically" means. I can think of "average of the unhappiest year over the population", "year with lowest average happiness", "year with the largest probability of being the unhappiest, when picking random people from the population". All of this doesn't even begin to consider censoring due to death. The more numerically well-behaved definitions seem to be exactly the ones that require comparing happiness between people, so the question of normalization (or lack thereof) enters into play...