One of the problems in dealing with population statistics (vis a vis war death responsibility) is the absolute impossibility of being 100% accurate when it comes to causal relations. Population growth is historically steady and using the census as a way to determine causal impact of events, is one methodology that has been acceptable to historical writers. For example, how many people died during the Chinese famines as related to the Great Leap Forward? One method is to extrapolate what the population should have been given the annual population growth rate, minus what the census is in reality, and one can say with fairly assumed accuracy that the event cause x amount of impact on the population.
Of course, nothing is absolute. One has to remember that the Japanese invasion did, in fact, interrupt an ongoing civil war. Chinese had been killing themselves in droves during the republican warlord era and continued even into the war with Japan despite a notable truce between the Nationalist and Communist Chinese. http://en.wikipedia.org/wiki/New_Fourth_Army_incident
IMHO, I don't think the death toll was inflated insofar as numbers dead, but rather that the causal relations of those deaths are complex, multifaceted, and cannot be easily attributed solely to one particular war faction. It is clear that the IJA did cause a tremendous number of deaths, but it is important to remember that they were not the sole cause. Also (although a bit tangential to your question but nonetheless important), if one looked to history as a metric of world wide deaths from armed conflict, then far and away the biggest claims to number of people killed is within the China region. http://en.wikipedia.org/wiki/List_of_wa ... death_toll
IMHO, it appears death indeed rides a pale horse and he seems to speak fluent Chinese.