Evaluating Theories

Evaluating Theories
A first approach
©1999 Edward G. Rozycki

RETURN
edited 11/3/2016

Overview: A theory may be evaluated in two dimensions: the extent to which it is parasitical, and the extent to which it is operationalizable.

1. A theory is parasitical to the extent that it relies on other theories or prior knowledge on the part of the user in order to operationalize it or to apply it intelligently. (This gets somewhat more involved, as we will see below.)

We can imagine a scale of parasitism:

figure 1

2. A theory is operationalizable to the extent that it can be operationalized with few competing operationalizations, that is, with fewer interpretations.

We can picture a scale of operationalizability:

figure 2

The best of theories would rank high on both scales in the shaded area below:

figure 3

How a Theory Works:

A theory divides the world conceptually into variables each of which relates (or not) to the others.

Operationalization helps us relate these variables to observational classes (event-classes), which (may not) relate to each other in the same manner as the theoretical variables to each other. Imagine a simple example:

figure 4

To the extent that our observations relate to each other in a piecewise parallel manner, to that extent is the theory homomorphic (having the same shape) to our observations. Figure 4 gives a precisely homomorphic example. This special case is isomorphic (identically shaped) at both levels, theoretical and observational, both for the variables and observations and the relationships and links between them. Many other variations are possible (and more likely).

A theory is informative to the extent that the theoretical variables do not correlate with what we believe should be the observational variables. So more highly informative theories, provided they work, i.e. they yield results at the observational level, are less parasitical, because they do not rely on, or may in fact contradict, our background information or assumptions about the world.

We saw above that the less interpretable our theory is, the more operationalizable. We can chart the relationships between parasitism, operationalizability and operationalizable in the following way

figure 5

Answering some questions

With the considerations and definitions given above we can answer some questions about the goodness of a theory.

1. Because we can get certain results keeping a specific theory in mind, does this prove that that theory works? No. It depends on how parasitical and how interpretable it is. A competing theory may just as easily explain the situation.

2. If a theory is homomorphic to many practical uses, does this indicate its strength? Not necessarily, because a theory that is too generally homomorphic may be no more than a metaphor. Competing theories should be considered.

3. If a situation can be affected by manipulating a variable the theory does not account for, what does this indicate? That the theory is lacking.

4. If the theory provides a variable (or relationship between variables) that successful practice can ignore, what does this indicate? That the theory is otiose (pointless and overdeveloped).

TO TOP