The Coleman Report thus adds a vast weight of fact to the irrefutable moral argument for integration. And it was perhaps expecting raining affirmation of that principle that Civil Rights leaders and educators from all over the Northeast attended a Harvard Ed School colloquium on the Report last October. The conference brought together some of education's foremost scholars, including Coleman, in the first public forum of its kind since the Report's appearance. A unanimous call for integration would have been a genuine breakthrough. And falling that, a clarification of the issues dividing experts would have at least explained past academic silence.

Neither the call nor the clarification was forthcoming. What emerged was a profile of the dilemma of U.S. education. During three hours of morning talks, Coleman and Samuel S. Bowles, assistant professor of Economics, debated methodology to an uncomprehending audience. Then Preston Wilcox, Negro sociologist, delivered the ghettos' demand: put up or shut up; integrate or give blacks their schools, but do it now. On the one hand the experts quarreled. On the other the time bomb ticked in the ghettos.

The Bowles-Coleman controversy may have seemed trivial to the ghetto parent, but it was not. Like the many. other disputes swirling around the Report, it reflects the study's real methodological problems--problems which fundamentally affect its policy implication.

Some of these are matters of technique. Bowles points out, for example, that Colemen measured per pupil expenditure by dividing district expenditures by pupils per district. He thus overlooked any differing expenditures among schools in the same district, or among pupils within schools. (This distinction is possible because of the group of children by ability--Negroes are usually placed in the lowest levels.) Also, Colemen got his information about school facilities by questioning principals--not always the most objective source--rather than conducting independent studies. Both oversights could affect Colemen's conclusions about the effect of facilities on achievement.

More fundamental problems arise from the nature of surveys themselves. There is first the inherent inaccuracy of questionnaires, the source of most survey data. Yes and no, or multiple choice answers never capture crucial nuances in subject response. Take a hypothetical attempt to measure the education of southern Negroes by asking if they have encyclopedias. A yes response could mean a high literacy; it could mean encyclopedias are status symbols, or local salesmen are on the ball.

The most crucial handicap, however, is that surveys are one of social science's least scientific tools. One-shot affairs, they approach a system from