Item banking

234
When appropriate statistical analysis of test results has been carried out,
it is possible to construct what is called an item bank. An item bank
is a large collection of previously trialled test items, normally stored
nowadays on a computer, which is placed at the disposal of test
constructors. Usually stored with each item in the bank are:
1. A number of identifying criteria, relating to such things as its
content, class level, stage in the syllabus or course book, the
testing technique used, and number of points.
2. Correct response(s) and scoring instructions.
3. Measurement information on the item, such as difficulty
level and discrimination index, which has been obtained
through previous trialling.
4. Notes on the item (when written, when used, etc.).
Once they have access to an item bank, test constructors simply choose
from it the items that they need for a test. They do this by entering into
the computer details of the kinds of items they need. They might begin for
example, by asking for receptive vocabulary items which have a facility
value between 0.4 and 0.6, and which relate to third year study at their
institution. The computer will immediately present them with all the items
in the bank that meet these criteria, and they are given the opportunity to
‘browse’ through these, choosing those items that they decide to include
in the test. Once they have chosen all the items they need for the test, and
have provided details such as the test title and rubrics, the computer
provides a printed version of the test.
There are a number of benefits to be had from item banks:
1. Once the bank is constructed, there is a considerable saving
of effort. Tests do not have to be constructed over and over
again from scratch.
2. Since the trialling of the items (which makes use of anchor
items, referred to in Appendix 1) is carried out before they are
Appendix 2 Item banking
https://doi.org/10.1017/CBO9780511732980.019 Published online by Cambridge University Press
Appendix 2
235
entered into the bank, the quality of tests that use them will
almost certainly be higher than those made up of untrialled
items.
3. The psychometric information on items gathered during
trialling means that the measurement qualities (including level
of difficulty) of tests made up of these items can be predicted
(before the test is taken) with greater accuracy than when
predictions are made on the basis of test constructors’ judgements. This in turn means that a test constructed in one year
can be made to have the same difficulty as tests set in previous
years, with implications for the maintenance of standards,
fairness, and the evaluation of teaching.
The development of an item bank follows very much the procedures as
those for the development of a test. The only differences are that the
specifications have to be for a bank, not a test; and the trialling process –
making use of anchor items – is absolutely essential.
Item banks are now regarded as indispensable to serious testing
organisations. With the advent of powerful but inexpensive computers,
item banks have become an attractive possibility for all serious testers
who are prepared to put in the necessary initial effort.
Further reading
Suggested sources for investigating item banks and banking can be
found on the book’s website.
https://doi.org/10.1017/CBO9780511732980.019 Published online by Cambridge University Press