Re: Anthropology as Science

Matthew S. Tomaso (Tomaso@MAIL.UTEXAS.EDU)
Tue, 10 Oct 1995 13:40:59 -0500

At 04:21 PM 10/9/95 +0900, John McCreery wrote:

I agree with just about everything that John had to say here, except that I
think we have broadly different conceptions of the value, meaning and
purpose of replication in science.

>Like Read, I am not so pessimistic as Matt about the
>impossibility of replicating ethnographic data. For certain
>stretches of space-time and appropriate levels of abstraction
>replication is certainly possible.

The problem here is that the appropriate levels of abstraction, IMHO, often
make the replicative procedure somewhat less useful in terms of
confirmation. As with statistics, we're forced to deal with refutation or
failure to refute in hypothesis tests - not confirmation of any general
principles. Let's face it, the few times I've seen replication attempts in
ethnography (Goodale's excellent ethnography on the Tiwi, and the old Leslie
Freeman/Margaret Mead stuff), they've been very instructive and significant,
but hardly replicative in any sense.

Where am I coming from?

Well, the way I look at replication in science stems partly from process
geomorphology and archaeology. I'll stick with process geomorphology for
the moment because bringing up the archaeological stuff will likely result
in a tirade of hate mail because of my too-strong opinions. In process
geomorph, as well as many physical/chemical sciences, experimental
replication is differentiated from field testing. Replication, in this
sense, involves wielding as-close-to-absolute control over the variables
influencing a certain problem as possible. It is only really possible in a
laboratory environment. We can't for obvious ethical and rational reasons,
do this in anthropology. Why we should want to is beyond me. Personally, I
don't consider this a problem since I don't consider this kind of
replicability to be particularly important in anthropology. Field testing,
which most geomorphologists would agree is not really replicative in the
scientific sense, involves taking what you've already 'shown to be true' in
a lab setting and applying it in the field interpretively to see whether it
still makes sense of the field situation. Even sediments, as inert as they
are, are too dynamic and complex to allow geomorphologists to achieve any
real 'replicability' regarding field testing (aka 'ground truthing').
Instead, the researcher keeps reference subsamples of the samples obtained
from the field so that others can look at them and verify or refute the

Interpretations of field data, since the variables are unconstrained, always
have an element of subjective, ideosynchratic, (or at least theoretically
biased) judgment, so the interpretations can not really be replicated.
Field testing is crucial in anthropology as it is in geomorphology. Perhaps
moreso. The problem that I see with replicating ethnographic stuff at a
non-abstract level is that you end up with "Micky Mouse laws" like '1.
Sandwiches are a category of food . 2. Food is made to be eaten. 3. Joe
makes sandwiches and therefore 4.Joe makes sandwiches in order to eat' and
then the problem with replicating interpretations at an abstract level is
the fact that we are talking about _interpretations_ when dealing with
absstraction, not results. One could just as easily say that Joe makes
sandwiches in order to produce viable offspring - you just can;t really test
the propositions. I just don't think replication works the same way in
anthropology as it does in other sciences, whether using John's concept of
it or mine. That is, however, not to invalidate or critique it since I
think that many 'revisionist' and feminist 'replication - like' studies have
been very very useful. So, in the end, I hope John and I have found a
common ground, as we usually do.

Matt Tomaso.
Anthropology. University of Texas at Austin.
Phone/Fax 512-453-6256