This question is actually easily generalisable to any battle across almost any period of time, and is actually a well-known statistical phenomenon - the losing side tends to rout, and routs tend to have higher casualties. More often that not, any battle will always have one side suffering a disproportionate number of casualties than the other, even if the forces were evenly matched to begin with. In WWII, this was not an exclusively Japanese phenomenon. The Malayan Campaign for example, even ignoring the number of soldiers captured, had British casualties 2x that of the Japanese even with double the British force. The Battle of El Alamein had almost 3x the number of Italian/German casualties as British, despite being described as a hard-fought victory. Even the Battle of Crete, widely held to have been an Axis Phyrric victory, still featured 6,000 Axis casualties against 23,000 Allied casualties.

The main reason we see this is that military effectiveness decreases exponentially with casualties. This is most visible at the squad level. A squad of 7 men can execute a variety of tactical manouvres, from alternating covering fire, to pin and flank manouvres. This ability deteriorates rapidly as men are lost. It becomes almost impossible to pin and flank with less than 5 men, and alternating covering fire becomes ineffective with less than 3 men. The result is that a force which has lost half its men is not half as effective, but perhaps a third or even a quarter. A single soldier fighting against a 3-man squad does not have a 25% chance of winning, but almost 0%.

Applied to a battle, it is easy to see that an evenly matched beginning can rapidly devolve into a one-sided pummelling. This was the case in most of the American island hopping invasions like in the Battle of Tarawa. The initial stage of setting up a beachfront was typically a bloodbath, as well placed Japanese troops massacred defenceless Marines. But this situation reversed rapidly as the beachhead was formed, and eventually with sheer numbers and firepower, it would be the Marines who could massacre effectively defenceless isolated Japanese units. Consequently the bulk of American casualties were incurred in the start of the battle, and the bulk of Japanese casualties towards the end. None of this suggests that the Japanese fought any less well, or that the victory came at any lower a cost to the Americans, but is rather a statistical feature of battles. With the Battle of El Alamein, the bulk of Axis casualties were also incurred during the rout and disorderly retreat. This is why rearguard actions are so important, since most casualties are actually sustained during the rout; though there is no rearguard action possible on an island.

This is not always true, and one can easily find examples to refute it. For example, the Battle of Stalingrad was won by attrition and commitment of large numbers of men and materiel by a Soviet force willing to expend the men and materiel. This serves to highlight the fact that numbers are less consequential than what we can broadly call combat ability, which is a function of a great variety of different factors.