"EITHER WAY I go on this, I'm going to be sued." So said Education Secretary Terrel Bell when he decided last month to use 1970 rather than 1980 Census data to allocate $2.4 billion in federal aid to schools with needy students. Why use 12- year-old data? The Census Bureau has tabulated data showing the number of children ages 5 to 17 living in households with poverty incomes by state, but the county-by-county data, which must be used under the law, are not promised until this month, and may be late. Mr. Bell said preparation of county allocation tables would take another six weeks. So he went ahead and used out-of-date data.

Use of the 1970 Census data, however, seriously and unnecessarily misallocates the aid. The pattern of poverty has changed over the past decade. The Northeast and Great Lakes states have more children living in poverty households than in 1970, the South many fewer. Thus in 1970, 24 percent of North Carolina's children but only 12 percent of New York's lived in poverty households. In 1980, 18 percent of children in both states lived in poverty households.

The rising number of poor children in what historically have been our richest and most productive states is a national problem worthy of the most serious and concentrated attention. It is obviously not susceptible of easy or single solution. Existing federal programs that provide educational aid to poor children should ensure that that aid is targeted as accurately as possible. This means using the 1980 Census figures, even if it requires a little delay and even if it means facing lawsuits from states and localities that would get more money under another formula.