ated grants. While some high-quality statistical information was being produced, the quality was variable. In two programs, quality was maintained or improved on some dimensions but in the Common Core of Data, data quality problems have persisted for several decades. The major influence on information production was severe reductions in funding levels. Activities that did not carry congressional mandates were most vulnerable to funding declines and changes in priorities, which also were linked to rapid changes in leadership. Expert review of specific information-gathering activities had a positive influence on quality in some instances. Results were clearest when several of these factors coexisted and worked in the same direction.

Principal Findings

Information Production

The number of grants and contracts awarded for research decreased 65 percent from 476 in 1980 to 168 in 1985. The number of evaluation contracts peaked at 119 in 1980 and progressively dropped 79 percent to 25 in 1985. Statistical surveys, planned or conducted, fell 31 percent between 1980 and 1983 from 55 to 38. The intervals between data collections increased and technical support to the states for data collection was sharply reduced. (See pages 20-24.)

The information that was produced by awards also changed. Sixty-five
percent of NIE'S 1980 awards but only 11 percent of the 1985 awards
were for new data collection. Awards for service activities such as dis-
seminating information and providing expert testimony in civil rights
cases increased from 35 percent to 89 percent of all awards. Fewer edu-
cational areas were investigated in 1985 than in 1980 through research
grants. In 1980, for example, 56 of 293 awards for new data collection
went toward studies of special populations such as minorities and
women. In 1985, there were five such studies. Some areas such as learn-
ing in nonschool settings and areas identified as "school problems"
(including such issues as dropouts and delinquency) received no new
data collection funds at all in 1985; in 1980, there were 33 awards. (See
pages 31-36.)

Further, those who carried out the work shifted. The proportion of research awards made to department-sponsored institutions (for example, laboratories and national centers) increased substantially from 1980

Executive Summary

Quality of Statistical
Programs

Influences on Information
Production and Quality

to 1985. In 1980, institutions received 25 percent of the awards in three major program areas, compared to 56 percent in 1985. In 1980, 23 percent of NIE's awards were made through contracts; in 1985, 86 percent. OPBE funded nearly all 1985 evaluations through competitively awarded contracts; in 1980, the award process was more diverse. Thus information-gathering was increasingly more likely to be prescribed by the agency than to have been proposed from the field. (See pages 36-38.)

A review of relevance, timeliness, technical adequacy, and impact shows
that quality varied. The National Assessment of Educational Progress
received generally high marks, although efforts to optimize one aspect
of quality were associated with losses in other dimensions. The Fast
Response Survey System received relatively high marks on relevance
and medium ratings on technical adequacy and timeliness. However, the
Common Core of Data, adequate in some respects, was generally poor in
its quality of information. Many problems-some of which had been
identified by others several decades earlier-remain. (See chapter 3.)

Support for research has decreased since the early 1970's by more than
70 percent in constant dollars, despite the fact that the federal invest-
ment in education increased by 38 percent and federal support for
research in general increased by about 4 percent in constant dollars
between 1980 and 1984. Funding for statistics and evaluation also
declined more than in these areas for the government in general. The
patterns of fiscal declines in research, evaluation, and statistical activity
corresponded to reductions in information production. (See pages 68-
72.)

Although all information-gathering activities were affected by budget constraints, congressionally mandated activities received smaller reductions and thereby consumed an increasing share of available resources. Activities that were not required by law were vulnerable to changes in priorities, funding, and policies. Rapid turnover of top leadership, especially in NIE, was associated with decisions not to fund areas of research initiated under other directors. (See pages 76-78 and 83.)

In the three statistical programs, relevance was increased by adding data elements, tailoring data collection to the needs of specific requesters, and making dissemination flexible. Timeliness was improved by releasing data early and diversifying their formats. Technical adequacy was higher for surveys than for data from state administrative records.

Some information-gathering activities reviewed comprehensively by technical experts improved in quality. (See pages 60-66.)

GAO does not present recommendations in this report.

Recommendations

Agency Comments

The Department of Education generally agreed with GAO's findings, stat-
ing that the report will perform a valuable function in documenting a
long-term decline in resources for educational information. However, the
department expressed three concerns about the report. First, citing the
many organizational changes initiated since 1985, the department
believed GAO's analyses did not reflect the current situation. Second, it
questioned GAO's assessment of shifts in priorities, stating that greater
emphasis on dissemination represented a positive step, changes in lead-
ership did not affect research priorities, and almost all important areas
were being investigated through a variety of strategies. Third, the
department took issue with GAO's analysis of the implications of changes
in who is producing information and how it is funded. Department offi-
cials also provided detailed descriptions of recent organizational
changes, documentation on budgets and activities not covered in GAO'S
review, and further specific comments. (See appendix IV.)

GAO acknowledges the numerous changes since fiscal year 1986. However, it is too early to determine whether these changes will adequately address the problems identified in this report or the new problems that the changes themselves might create. Empirical assessment of the production and quality of information will be necessary.

With regard to shifts in priorities, GAO maintains that dissemination can remain a critical part of the research process only if the data that are being disseminated are relevant and timely. GAO Continues to conclude that changes in leadership did affect priorities and notes that while information is being collected on contemporary problems, the department seems to lack formal mechanisms for identifying emerging issues. GAO continues to believe that while contracts provide a needed basis for accountability, widespread use of contracts has other, less positive consequences. For example, requests for proposals often specify the scope of work, leaving little flexibility for the imaginative researcher. While GAO Commends the department's efforts to restore some of the avenues for new data collection-such as the unsolicited-grants program-current levels of support are dramatically lower than in 1980.