128 • Alternatives to Animal Use in Research, Testing, and Education 
cedure is applied to data after they are collected 
to adjust for chance differences among groups. 
The analysis increases power, and fewer animals 
may be needed to obtain statistically significant 
results (113). 
Single- Subject Design.— In some instances in- 
ferences can be made about populations from very 
small samples. This is common in psychophysical 
experiments in which there is a substantial prior 
body of evidence indicating that the behaviors un- 
der investigation do not vary appreciably within 
the population at large (e.g., visual sensitivity to 
light). Although such experiments can be con- 
ducted using just one subject, two or three are 
typically used to guard against the possibility of 
misleading results from an atypical subject (198). 
In other than psychophysical experiments, the 
general procedure in single-subject research con- 
sists of choosing a baseline (which involves meas- 
uring the frequency of occurrence of the behavior 
of interest), changing one treatment variable at 
a time, temporarily withdrawing the experimental 
treatment to assess its causal effects, and repeat- 
edly measuring the baseline behavior before and 
after each treatment. (More sophisticated experi- 
mental designs available for single-subject research 
are reviewed in ref. 96.) Single -subject designs are 
increasingly used in animal operant conditioning 
and human clinical research (187). Statistical anal- 
yses of these are reported infrequently due to the 
lack of many statistical techniques for handling 
such data, although using time-series analyses to 
test for changes over time is one acceptable method 
available (111). 
A limitation to studying a single subject is the 
uncertainty of the generality of the findings, a prob- 
lem commonly dealt with by replicating the experi- 
ment with different subjects (96). Thus the reduc- 
tion in animals used may be illusory. 
Inbred Strains.— One way of reducing variabil- 
ity (and hence increasing power) is to use highly 
homogeneous populations of subjects. Inbred 
strains of animals, produced as a result of 20 or 
more generations of brother-sister matings, rep- 
resent one approach, though it is usually much 
more expensive than using randomly bred animals. 
In most inbred strains all subjects are highly iden- 
tical genetically and genetically stable; they change 
only as a result of the slow accumulation of muta- 
tions. In contrast, outbred stocks of animals are 
genetically variable. They contain an unknown and 
uncontrollable degree of genetic variation that may 
obscure or mask experimental treatment effects. 
Inbred strains not only increase statistical power, 
they also reduce variability between experiments 
conducted in different laboratories or in the same 
laboratory at different times (68). 
It can be argued that experiments should rely 
on animals drawn from heterogeneous, outbred 
populations in order to get a broad genetic basis 
for results that can be extrapolated, for example, 
to heterogeneous human populations. Yet the 
differences between different inbred strains are 
usually greater than the differences between in- 
dividuals of an outbred stock. Greater generality, 
then, may be obtained by conducting experiments 
with two or more inbred strains (68). 
Sharing Animals.— A team approach to re- 
search questions across biomedical and behavioral 
research disciplines could reduce the number of 
animals needed for behavioral research (173). For 
example, researchers studying a behavioral phe- 
nomenon by noninvasive means could, at the ex- 
periment’s conclusion, give their animals to biolo- 
gists investigating the anatomy or physiology of 
that species. Likewise, scientists from different dis- 
ciplines could collaborate on research proposals: 
A psychologist may be interested in studying preda- 
tor-prey relations, while a biologist wants to study 
endocrinological changes in response to stress; ef- 
fective collaboration could yield two different data 
sets from the same animals. 
Substitution of Cold-Blooded for 
Warm-Blooded Vertebrates 
The modified use of animals in biomedical re- 
search includes the replacement of mammals and 
birds with fish, amphibians, and reptiles. In be- 
havioral research, however, the often vast differ- 
ences between species are likely to make such sub- 
stitutions difficult. At the moment, researchers 
know more about why warm-blooded vertebrates 
cannot be replaced with cold-blooded ones, as this 
description of seven behavioral research dis- 
ciplines illustrates. 
