54 • Alternatives to Animal Use in Research, Testing, and Education 
Table 3-3.— Reliability of Various Data Sources 
Source 
Years 
covered 
Confidence 
rating 
Strength(s) 
Limitation(s) 
USDA/APHIS: 
Mandated species 
1982-83 
Excellent 
Required by law. Data available 
by institution, thus extrapolation 
to nonreporters is possible 
Mandated species 
1972-81 
Fair 
Required by law. Data by 
institution available, but not 
used 
10 to 20 percent of institutions 
not included in reports to Con- 
gress. Totals not consistent 
(some years include Federal 
agencies, others do not) 
Exempt species 
1982-83 
Good 
Data by institution available. 
Rats and mice were on the 
form so anyone who reported 
probably provided an accurate 
number. Many did not realize 
that these were voluntary since 
they were listed on form. Extra- 
polation possible 
Not required by law 
ILAR Surveys of 
Laboratory Animal Use 
1965-71 
Poor 
Of some use in establishing 
trends for that period 
Old data. Cannot extrapolate 
to missing data 
1968 Survey 
1967 
Fair 
Statistically sound survey. 
Possibility of extrapolating to 
other institutions 
Limited to 683 Federal-grant- 
eligible institutions 
1980 Survey 
1978 
Fair 
Thorough and statistically 
solid. Extrapolation to non- 
reporting institutions possible 
Primary attention given to 
nonprofit Federal-grant-eligible 
institutions. Not required by 
law to be filled out 
W.B. Saunders & Co. 
1965 
Indeterminate 
Company defunct, survey 
methodology unclear; no 
evaluation possible 
Foster D. Snell 
1975 
Indeterminate 
Data appear to be based on in- 
terviews with two breeders 
Methodology unclear. Person- 
nel no longer available 
SOURCE: Office of Technology Assessment. 
breeding and from commercial sources/’ not nec- 
essarily the number actually used in experimen- 
tation. 
The ILAR and APHIS surveys are so different 
in their organization and methodology that it is 
not meaningful to compare the two sources, even 
in years for which data from both are available. 
It is also difficult to point out significant changes 
within this data source because the ILAR method- 
ology varied over time and could not be verified 
adequately, so changes in numbers are difficult 
to substantiate. 
ILAR Surveys of Laboratory Animal Use (20) con- 
sist of tables summarizing the results of question- 
naires on the number of animals used for research. 
As ILAR personnel cannot discern who was sur- 
veyed and who responded, extrapolation for miss- 
ing data is impossible. The surveys could, how- 
ever, be of some use in assessing trends between 
1965 and 1971. A “poor” confidence rating was 
given. 
The 1 968 Survey ofLabora tory Animal Facilities 
and Resources (21) appears to have been a very 
thorough and statistically sound survey including 
all known users of laboratory animals . The results 
shown, however, are only for the 683 organiza- 
tions eligible for Federal grants that responded be- 
cause of the interest of the survey sponsor (NIH). 
It is possible, however, to normalize for missing 
data based on the reported biomedical research 
expenditures for these 683 organizations of $920 
