The allocation of the representatives appeared to be by "population." But the actual numbers appear to have been affected by a different view of the population than the 1792 census, as reflected in the "quotas" of the Continental Army units for each state. Under this scheme, the respective state quotas were as follows:
On September 16, 1776, the Continental Congress passed the "eighty-eight battalion resolve," which called for each state to contribute regiments in proportion to their population. (The terms regiment and battalion were virtually interchangeable at that time since nearly every Continental Army regiment consisted of a single battalion). The quota of infantry regiments was fixed at 15 each from Massachusetts and Virginia, 12 from Pennsylvania, 9 from North Carolina, 8 each from Connecticut and Maryland, 6 from South Carolina, 4 each from New York and New Jersey, 3 from New Hampshire, 2 from Rhode Island, and 1 each from Delaware and Georgia."
"New York" at the time, reflected only "upstate" New York. The reason was that New York City was under British control at the time, and was pro British anyway, and therefore could not be counted on to supply its pro rata share of troops to the Revolutionary war effort. At any rate, New York state's representation appears to have been counted without its New York City population in 1787,
It's probable that the above quotas were also based on the "white" population of each state. (Black slaves were not expected to become soldiers.) The subsequent "three fifths" compromise to the U.S Constitution enlarged the populations of the southern states by adding "three fifths" of the number of black slaves to the final count (the northern states had few of these). Thus, Virginia in the 1792 census stood "head and shoulders" above all the northern states (as opposed to "first of equals" with Massachusetts and, to a lesser extent, Pennsylvania), as in the 1776. Just about all the other southern states (Delaware, Maryland, North and South Carolina, and Georgia) benefited.
One issue is that we are comparing apples to oranges to pears. There were 88 regiments in 1776, 65 delegates in 1787 and 102 representatives in 1792. So each state's allocation should be looked at as a percentage of the total, not absolute numbers. Under this scheme, Virginia, for instance, had a representation of 17% (15/88) in 1776, 15% (10/65) in 1787, and 19% (19/102) in 1792.
One possibility raised in the comments was "tax: issues. If so, these would have been reflected in the above troop quotas. And the three-fifths compromise also increased the tax base of the southern states, and may have raised their representation in that manner.
I cannot explain why Massachusetts appears to have been underrepresented in the Constitutional (1787) allocation, unless it was tax issues. North Carolina appears to have been underrepresented as well. They had less of a footprint in the Revolution than their population implied. And perhaps they contributed either proportionately fewer troops or fewer taxes, or both, than other states.