Ch. 3— Patterns of Animal Use • 53 
In general, if the numbers cannot be justified 
through some rational process (such as the above), 
too much significance should not be attached to 
them. 
These four criteria were used to assign a confi- 
dence rating to each data source. The confidence 
categories are: “excellent,” "good,” “fair,” “poor,” 
and “indeterminate.” (These ratings refer only to 
the published numbers, not to their usefulness as 
a predictive tool.) Such a confidence rating is nec- 
essarily subjective; the categories are comparative 
and should not be viewed as absolute. 
Upon reviewing all the data sources avail- 
able for predicting the laboratory- animal use 
in the United States, it is clear that no source 
accurately portrays the number of animals 
being used. Each has methodological prob- 
lems that prevented it from accurately count- 
ing all users of animals. What follows is an anal- 
ysis of the available data sources and how they 
rank in comparison with the other surveys in terms 
of confidence and reliability. 
USDA Animal and Plant Health 
Inspection Service (APHIS) 
The 1982 and 1983 data were analyzed on a case- 
by-case basis. Copies of the original report forms 
were obtained from USDA; they were sorted by 
institution type, checked, coded, and entered into 
a computer database. Comparing the 1982 and 
1983 APHIS data (see table 3-5, in the "Summary 
and Analysis of Estimates” section) with the USDA 
Animal Welfare Enforcement Report for 1980 (the 
APHIS 1980 data in table 3-5) reveals a large dis- 
crepancy. The USDA reports invariably contain 
lower numbers for all species, as the data sheets 
received after the December 31 cutoff date are 
not included in reports in either the current or 
the next fiscal year. It is estimated that between 
10 and 20 percent of the reports are not used to 
compile the report to Congress in a given year (17). 
This limitation does not apply to the results con- 
tained in the present compilation for 1982 and 
1983, since all data for a given year were used no 
matter when received. The assumption is made 
that copies of virtually all of the data sheets re- 
ceived by USDA in the 1982 and 1983 are used 
in this study. No verification was made of which 
institutions did not report. 
The number of institutions reporting to APHIS 
has hovered around 1,000 since 1972. The num- 
bers for 1982 and 1983 (shown in table 3-6, in the 
"Summary and Analysis of Estimates” section) were 
tabulated from the actual summary data sheets 
provided to APHIS by the institutions and include 
all possible reports . Even these figures— 1 ,127 for 
1982 and 1,146 for 1983— are probably low, as not 
all institutions submit reports. (The total number 
of institutions registered by APHIS was 1,113 in 
1982 and 1,166 in 1983; this excludes Federal agen- 
cies, which are not required to register.) Some of 
the institutions may not report because they have 
not used any animals that year, or because they 
have only used exempt species. 
For the six required species listed on the form 
(dogs, cats, guinea pigs, hamsters, rabbits, and pri- 
mates), the numbers reported provide a very close 
approximation of the animals actually used. Thus 
these data were assigned a confidence rating of 
"excellent.” (For a summary of all the data sources 
discussed in this section and their confidence rat- 
ings, see table 3-3.) For exempt species (primarily 
rats and mice), it is possible to estimate the num- 
ber of unreported rats and mice by extrapolating 
from the numbers reported (see the section on “Es- 
timate Using Corrected APHIS Data”). Some com- 
mentators (1,3,27) claim, however, that a certain 
number of exempt animals go unreported— and 
would be missed in an extrapolation— because they 
are purchased directly by the user and not re- 
ported to the central facility. This contention could 
not be confirmed. Therefore, the voluntarily re- 
ported data on rats and mice on the 1982 and 1983 
APHIS annual reports received a confidence rat- 
ing of "good.” 
ILAR Surveys 
The Institute for Laboratory Animal Resources, 
a component of the National Research Council, 
periodically surveys users of laboratory animals 
(18,19,20,21,22,23), although it is generally more 
concerned with facilities and personnel than with 
quantity of animals used. The ILAR data repre- 
sent the number of animals "acquired by own 
