SUPPOSE YOU wanted to create a federal program to help states educate poor children. How would you distribute the money? The obvious way would be to hand it out in direct proportion to the number of poor children in each state. But if that method ended up rewarding states that deliberately failed to spend a reasonable amount of their own money on education, you might decide it wasn't fair.

This dilemma accounts for the current quarrel in Congress over a formula for distributing about $2.66 billion in federal funds to help local school districts educate children from low-income homes.Unlike most politically generated formulas, this one is relatively straightforward. The principal criterion is the number of poor children in each state as measured by the most recent census. In 1974, when the formula was devised, this gave the relatively poor southern states a hefty part of the money.

But northern states argued that only counting children in poverty wasn't fair since it didn't take account of those northern states' higher per pupil costs, which reflected, in part anyway, a willingness to spend more of their own tax money on education. Ignoring that fact, they argued, would simply reward low-spending states for their parsimony.To address this concern, the final formula took account of such costs as well as of povery -- a compromise that reduced the South's share somewhat.

Now that the 1980 Census data are available, however, the South stands to lose substantial money. This is because it has prospered over the last decade, and the number of poor children in the region dropped substantially. Far from rejoicing at this success, southern legislators -- many of whom are no friends of federal aid to education -- are scrambling to keep their share of the pie. With the help of a research institute they devised a new formula.

The new formula would take account of a states' per capita income as a measure of its tax-raising capability. The idea has some merit, but it ignores the fact that, by a variety of means, most states of the South tax themselves very lightly even when their somewhat lower average incomes are taken into consideration.

Apart from its statistical properties, the new formula has the attractive political property of increasing aid to a majority of states -- thus ensuring Senate passage. The big losers would be the big industrial states, where the largest concentrations of poor children live.

Proponents of the change would like to slip it into law as a rider on a piece of must legislation or a major appropriation bill. That would be a bad way to amend so large and important an education program. Formula controversies are often bitter and always complicated. But that is no reason to try to run the thing through without anyone's quite understanding what has happened.