TEST CONSTRUCTION AND ASSEMBLY

After specific objectives have been defined in terms of student behavior and performance and before test item composition actually begins, it would be useful to establish a Table of Specifications. This will serve as a basic guideline for systematically ordering and assembling your evaluation instrument (heretofore known as "a test").

Essentially, the table will consist of cells giving information on the representation of various item types based on your specification of evaluation needs. These needs will be expressed in subject matter area delineated as to type of student performance expected. Although learning objectives may be graded or categorized in any manner suiting the evaluator's preference, Bloom's Taxonomy of Educational Objectives, may provide a useful and logically consistent item typology. The subject matter coordinate would be derived from statements of general objectives pertinent to this evaluation period.


Here is an example of a typical Specification Table constructed for use with a six-week block of a Basic Electronics course.

Learning Objective Levels – Cognitive Domain

% of representation on test

Subject Matter

% of representation in the unit of study

Knowledge only

Comprehension

Nature of Electricity

30

15

15

Ohm’s Law

10

5

5

Series Circuits

15

15

10

Parallel Circuits

15

10

5

Series-Parallel Circuits

30

15

5


When constructing the Specification Table, a specific course objective may be stated in a 2-3 word summary description and entered under "Subject Matter". A more thorough description of what the "Ohm's Law" category of objectives entails could be found by referring to an initial list of course objectives.

The percentage of items on the test devoted to a particular topic should roughly correspond to the emphasis given the topic in actual teaching of the course, and as originally implied in the general objectives.

As in this illustration, you may decide that to be consistent with general course objectives, no test item need be composed outside of the levels of "knowledge" and "comprehension." Also, you may wish to enter actual numbers of test items, rather than their percent of coverage.

Under or over-represented cells may not necessarily require revision of representation. Such problems can be dealt with by differential weighting of items. (Complex scoring formula could, however, excessively complicate the grading process.)

Although you are not likely to follow your derived representation figures rigidly, you'll probably find that becoming adept at using a Specification Table will notably improve your test composition efficiency. This table contains information that you should expect to automatically take into consideration anyway while constructing test items. Even our simple example suggests that more variables deserve consideration than most of us can keep in our heads or remain intuitively responsive to.


A Final Word On Mechanics

To allow more concise documentation of your test you may wish to construct a second Specification Table, entering item numbers in the cells. This will facilitate later revision or editing of particular item levels and types

Now that you have taken the first step in assuring that course content will be fairly represented, actual item construction may begin.


Continue on with...

Guidelines for Constructing Multiple Choice Items