2. TEST ASSEMBLY METHODS
In test assembly, items are assigned to a test from an item pool. The
pool contains many items with different characteristics, for
example, content specification, gender orientation, average
response time, word count but also psychometric features like item
information, item difficulty, or discriminating power
3. TEST ASSEMBLY METHODS
These are much more detailed and complicated then from previous
method of test assembly methods.
Different methods for assembling tests with item sets are generally
known to people.
4. MODELS FOR ASSEMBLING TEST ITEMS:
IRT based test assembly
CRT based test assembly
Matching observed score distributions
5. MODELS OF ASSEMBLING MULTIPLE TESTS
Sequential Assembly
Simultaneous Assembly
Big shadow test method
Optimizing bib design
6. SIMULTANEOUS ASSEMBLY
The previous problems are absent if the tests are assembled
simultaneously (that is, as a solution to a single model).
This type of assembly requires a restructuring of the problem
using a new type of decision variable.
7. METHOD 1
SIMULTANEOUS SELECTION OF ITEMS AND SETS:
This method was introduced by Van der Linden(1992)
The key feature of this method is that separate decision variables for
the selection of items and stimuli are defined.
Variables are used to model the constraints imposed on selection of
item and stimuli.
8. Special logical constraints are added to keep the
selection of items and stimuli consistent ,that is, to
prevent the selection of items without selection of
stimuli or the selection of stimuli without selection
of items.
9. METHOD 2
SIMULTANEOUS SELECTION WITH PIVOT ITEM
In MIP one important factor determining solution time is number of
variables in the model.
It is therefore advantageous to find models for test assembly
problems with item sets that are based on fewer variables with
results that closely approximate those for a full simultaneous
approach
10. It is possible to reduce the number of variables by
assigning one item from each item set in the bank
the status of a ‘pivot item’.
Formally a pivot item is an item selected for the
test if and only if its stimulus is selected for the test
11. In practice test constructor can be asked to select as
pivot items, those items that best represent their
stimuli and those that would be their first choices if
the test were assembled manually
12. METHOD 3
SELECTION OF ALL ITEMS PER SET
This test assembly was presented by Luccht and Hirsch(1991)
In method 2 number of decision variables was reduced by
removing the stimulus variables from the model.
13. A more dramatic reduction is possible if the
decision variables for items can be removed.
This option arises when the number of items per
stimulus in the bank has already met the bounds
in its constraints.