Table of Contents: Surmise Relations between Tests
< Previous - Next >


Surmise Relations between Tests


So far, we have referred to single tests. However, in common psychological assessment procedures we often deal with a set of different tests that are usually related. On the background of Doignon and Falmagne's framework, Albert (1995,Albert, Brandt, Hockemeyer, and Schappacher,in preparation; Brandt, Albert, and Hockemeyer,1999; 2000) extended the concept of the nonsymmetric surmise relation between items (i. e. within tests) to surmise relations between tests. The interpretation of a surmise or prerequisite relation between tests, i. e. , is that two tests are in surmise relation from A to B, if one can surmise from the correct solution of a given set of items in test A the correct solution of a particular non-empty subset of items in test B (see Figure 1).

Figure 1: Two tests A and B are in surmise relation from A to B ()

Formally, the relation is defined by
  
is the set of all knowledge states containing item a. For a set of tests , a surmise relation between tests has the property of reflexivity but not necessarily transitivity, i. e. in general, it is not a quasi order. However, there are special cases for which transitivity holds, namely left- and right-covering surmise relations.
The interpretation of a left-covering surmise relation (, see Figure 2) is that for each item there exists a nonempty subset of prerequisites in test B, i. e. a person who doesn't solve any item in B will not be able to solve any item in A, either. There is no need to test this person on test A. Formally, we say that two tests are in left-covering surmise relation from test A to test B. The relation is defined by
    
Right-covering means, that for each item , there exists at least one item for which b is a prerequisite , i. e. failing to solve any item in test B implies a failure on a subset of items in test A (see Figure 3). In other words, a person who solves all items in test A is also able to solve all items in test B. Hence, there is no further need to test the person on test B. Formally, we say that two tests are in right-covering surmise relation from test A to test B.
The relation is defined by
      

 

Figure 2: Left-covering surmise relation from test A to test B ()

 

Figure 3: Right-covering surmise relation from test A to test B ()

Finally, we speak of a total-covering surmise relation, if the relation is left- as well as right-covering. We can also extend the concepts of a knowledge state and a knowledge structure.
A test knowledge state is defined as the combination of item subsets per test person i is capable of mastering. The collection of all test knowledge states is called the test knowledge structure, which is defined as the pair , with denoting the set of tests . If a test knowledge structure is closed under union it is called test knowledge space, if it is also closed under intersection we speak of a quasi ordinal test kowledge space.

Top of the Page