Biomedical Engineering Reference
In-Depth Information
between males. Male bonobos interact less with each other
than chimpanzee males do.
Bonobos seem to use sexual behavior for greeting and
maintaining affiliative bonds between females and in
heterosexual pairs, not just around the time of ovulation
(
Savage-Rumbaugh and Wilkerson, 1978
). Bonobos have
many signs and gestures to signal readiness or willingness
for sexual interactions. However, they are less aggressive
than chimpanzees are, using fewer gestures and engaging in
less complex interactions. Bonobo parties are more stable
than chimpanzee parties are and individuals are less
competitive, perhaps because of
However, publishing “findings” based on inductive
reasoning at this point may be ill advised. In published
scientific papers, the evidence supporting the tested
hypothesis is carefully explained, and data are presented to
support any conclusions. In contrast, however, before the
hypothesis has been tested, the only data leading to an
inductive conclusion are the observations used to generate
the hypothesis.
Using the same data to generate and to test a hypothesis
leads to several types of errors. Applying statistical tech-
niques to existing data ignores the fact that once an event
has occurred, the probability of that event occurring is 1.0.
One cannot show that the frequency of an event that has
already occurred exceeds its probability of occurrence
(1.0). For example, the probability that a family that
already has five children, all of whom are girls, has five
girls is 1.0, whereas the probability that the next five
children born into another family will all be girls is 0.5
5
(32 chances in 1000). Given a very large number of fami-
lies, one can expect that many will be skewed toward one
sex or another even if the sex of each child is absolutely
random.
Proving that a hypothesis is consistent with earlier
observations demonstrates only that the inductive processes
were sound. Applying statistical analyses to the original
data is conducting a post hoc test to prove the induction
conclusion already reached. This application in no way
displays the value of the theory in predicting new data
outcomes. New data must be collected to test the prediction.
Retroactive predictions that lead to an error of “affirming
the consequent” only prove that the inductive conclusion is
a satisfactory explanation for what is already known. The
value of science is in predicting that which is not already
known. In extreme cases, the error of affirming the conse-
quent consists of accounting for a single observation
the bonobos' more
concentrated food sources.
METHODOLOGIES USED TO STUDY
BEHAVIOR
This section, which introduces research design and meth-
odologies used to study nonhuman primate behavior, is
included to assist those responsible for evaluating the
proper care of animals and experimental protocols that
require the use of animals. The methods discussed are not
unique to studying nonhuman primates but rather are an
extension of the general scientific methods used to obtain
data and test hypotheses.
Observation of Cause and Effect
Any scientific endeavor begins with observation. “Obser-
vation” can mean reading the available literature and
benefiting from others' experiences. It can also mean
watching or recording the subject of interest until one has
seen or heard enough to note some regularities. For
example, when seeking to determine why one animal
injures another or behaves in a certain way, one observes
the animal in various situations and tries to find a common
theme or variable. This type of logical search, for the
answer to why particular behaviors occur in a particular
situation more frequently than is attributable to chance, is
the inductive process. It is a search for some event that
regularly precedes the behavior of interest. Correlations are
calculated using statistical methods, and then one infers
a causal relationship in which one event is hypothesized to
be the proximal cause (stimulus or trigger) that elicits the
behavior of interest.
an
e
anecdote
by inventing a plausible theory and then using
that anecdote as evidence for the theory. Using a collection
of old anecdotes proves no more valuable for this purpose.
e
Hypothesis Testing
The first step in properly testing a theory is to use deductive
logic to produce a specific hypothesis that predicts how the
independent variable will affect the dependent variable.
These variables may be yes/no measures (male or female)
or may involve points along a continuous scale, like a range
of behavioral responses. If the hypothesis is that male
primates are inherently more aggressive than females are
and if we believe that chasing is a good measure of
aggressive behavior, then the experiment should measure
chasing behavior in one male sample and one female
sample. The specific prediction is that the males will chase
other animals more often than females will. The null
hypothesis is that males will not chase others more than
Anecdotal Evidence
When a relationship is hypothesized between an indepen-
dent and a dependent variable of interest derived by
inductive logic, the tendency is to want to share that
information, since the relationship may represent an
extremely important insight that can profoundly alter
thinking in the field or have key practical application.
