Explanation of Variation and Detection of Covariation

Abstract

Two research strategies, the most similar systems and the most different systems, have each been promoted as being the correct method for conducting comparative research. These two strategies are shown here to be identical in their logical structure, and neither strategy is found to be superior to the other in principle. Their sole differences lie in the amount of variability found in the dependent phenomenon and which independent variables this permits to be eliminated as irrelevant. A research strategy that mixes the two alternatives is proposed as being more feasible, given reasonable expectations about actual research settings.