ABSTRACT
The requirement that scientific theories include both abstract concepts and concrete implications, and that the two be logically connected, has been treated rather casually by sociologists. The empirical testing of abstract theories must remain somewhat loose until some strategies for dealing with this problem are devised. To the degree that rules of correspondence are weak and subject to distorting errors, deductions about matters of fact must be regarded as uncertain and possibly misleading. Auxiliary theories would be so constructed that differential bias, if present, would be recognizable empirically. The empirically estimated epistemic coefficients estimate the correlation between each specific indicator and the abstract variable that is assigned a particular role in the model on the abstract plane. The usual procedure in "testing" causal models with only one indicator for each abstract variable is to take the correlation between indicators as the correlation between the corresponding abstract variables.
