5 Stages of Test Construction
5 Stages of Test Construction
Hoa Nguyen
1
1 Stating the problem
1.
1. What kind of test (proficiency, achievement,
diagnostic, or placement)
2. What is precise purpose?
3. What abilities to be tested?
4. How detailed must the results be?
5 How accurate must the results be
5.
6. How important is the backwash?
7. What constraints (facilities, funding,
technical system)
2
2 Writing specification
2.
• 2 1 Test content
2.1.
– test designers begins with decision about test content (what
will go into the test): imply a view of test construct.
– T t content
Test t t involves
i l careful
f l sampling
li from
f the
th domain
d i off th
the
test: the set of tasks or the kinds of behaviours in the criterion
setting
– Two ways of defining test domain:
• (operationally) a set of practical, real-world tasks
• (abstract construct): a theory of components of knowledge and
ability that underlie performance in the domain.
– Operations: tasks candidates have to carry out
– Types of text. Eg. academic essay or notes (writing), article in
reading journal (reading…)
– Addresses: people whom the candidate is expected to write or
speak to
– Topics
3
2 1 Writing specification (cont.)
2.1 (cont )
2.2 Format and timing
– test structure ((how many
ypparts, items))
– time allocation
– example
– response format
*Relation between test content and response
format (McNamara p. 26)
4
2 1 Writing specification (cont.)
2.1 (cont )
2.3
2 3C Criteria
it i llevell off performance
f
– The required level(s) of performance for
(diff
(differentt levels
l l of)f) success should
h ld beb
specified
– It can be b simple
i l such h as 80% mastery
t off allll
items or more complex such as a detailed
descriptor of different levels
2.4 Scoring procedure
– very important
i t t esp. when
h scoring
i iis
subjective
5
3 Steps in writing the test
3.
3.1
3 1SSampling
li
– Everything under the heading of ‘Content” in the specifications
can be included in any one version of the test.
– Choices have to be made and succeeding versions of the test
should also sample widely and unpredictably.
6
3 Steps in writing the test (cont.)
3. (cont )
3.3 Key
– one correct answer
– alternative correct/ acceptable answers
– full credit and partial credit
– All this should be done through team work
7
3 Steps in writing the test (cont.)
3. (cont )
• 3.4
3 4 Pretesting/test trial
– Pretesting should be done before the test is administered for the
group which is intended to fix problems as even after careful
moderation there are likely some problems with every test
moderation, test.
– Pretesting group should be as similar as possible to the one
which the test is intended
– Problems during administration and scoring should be noted noted.
– Test evaluation: reliability coefficient of the whole test and its
components, analysis of each individual item should be
recorded.
recorded
– Test taker feedback (in the trial and in questionnaire form)
– Note. In many cases, pretesting is not feasible due to test
security time constraints
security, constraints. What can we do? (Noting any
problems with test administration and scoring and fix it
afterwards for future use).