The Post's Aug. 24 editorial regarding the Office of Management and Budget's review of the 1990 Census "dress rehearsal" implies that OMB's underlying motive is to weaken family-planning, job-training and housing programs, drawing "a veil over many kinds of social distress.. . ." This is a cheap shot that is preposterous and flat wrong.

OMB is in the midst of a public review process and is pursuing its responsibilities under the Paperwork Reduction Act. We do not "want to cut nearly half the questions out of the 1990 Census" but, rather, have asked the Census Bureau for further documentation and justification of a number of the planned questions. No decisions have yet been made, and none will be made until mid-September. Our goal is to ensure that the dress rehearsal and the ensuing 1990 Census provide the government and other users with good-quality information that is statistically valid and useful.

For example, the dress rehearsal long form, to be asked of one out of six households, includes a question on the cost of utilities and fuel for the past year, broken down by electricity, gas, water, and "oil, coal, kerosene, wood, etc." The Energy Information Administration rejected such questions some years ago because it found that they produced answers that were 25 to 50 percent higher than actual energy costs. EIA eventually designed a special survey that was much smaller and less burdensome and that collected much more accurate information. We are asking the Census Bureau why it decided to ask these questions despite EIA's experience.

The editorial notes that Census' proposed list of questions is "similar to those that it asked in 1980" and argues that no questions should be deleted. But the longer the form, the greater the respondent error.

Furthermore, the Census has not been static over time. For the 1990 Census, the bureau proposes to add several questions and eliminate several others included in the 1980 Census. The point is that keeping the Census the same is no virtue, that it is a dynamic document, and that, while everyone loves data, bad information is worse than none at all.

Finally, the editorial states that "nearly everyone dutifully completes" the Census. This is not really so. In 1980, 45 percent of the long forms that were mailed back were so incomplete or deficient that a follow-up was required.

The General Accounting Office has raised questions about the design of the 1990 Census based on response problems in the 1980 Census and pretests for 1990 and on Census Bureau research showing that nonrespondents were deterred by the complexity and length of the form (and this was the "short form"). GAO has recommended that some questions on the short form should not be asked of 100 percent of the population. OMB is reviewing some of the same issues that GAO has raised.

The decennial census is a necessary and invaluable collection of data on basic characteristics of the nation. We are trying to ensure the quality of that information is the best it can be.

WENDY L. GRAMM Administrator for Information and Regulatory Affairs Office of Management and Budget Washington