12.01.2015 Views

RESEARCH METHOD COHEN ok

RESEARCH METHOD COHEN ok

RESEARCH METHOD COHEN ok

SHOW MORE
SHOW LESS

You also want an ePaper? Increase the reach of your titles

YUMPU automatically turns print PDFs into web optimized ePapers that Google loves.

COMPUTERIZED ADAPTIVE TESTING 433<br />

How ethical it would be to undertake the final<br />

four of these is perhaps questionable, or indeed<br />

any apart from the first on the list. Are they<br />

cheating or legitimate test preparation Should<br />

one teach to a test; is not to do so a dereliction<br />

of duty (e.g. in criterion- and domain-referenced<br />

tests) or giving students an unfair advantage and<br />

thus reducing the reliability of the test as a true and<br />

fair measure of ability or achievement In highstakes<br />

assessment (e.g. for public accountability<br />

and to compare schools and teachers) there<br />

is even the issue of not entering for tests<br />

students whose performance will be low (see, for<br />

example, Haladyna et al.1991).Thereisariskofa<br />

correlation between the ‘stakes’ and the degree<br />

of unethical practice – the greater the stakes,<br />

the greater the incidence of unethical practice.<br />

Unethical practice, observes Gipps (1994), occurs<br />

where scores are inflated but reliable inference<br />

on performance or achievement is not, and<br />

where different groups of students are prepared<br />

differentially for tests, i.e. giving some students an<br />

unfair advantage over others. To overcome such<br />

problems, she suggests, it is ethical and legitimate<br />

for teachers to teach to a broader domain than<br />

the test, that teachers should not teach directly<br />

to the test, and the situation should only be that<br />

better instruction rather than test preparation is<br />

acceptable (Cunningham 1998).<br />

One can add to this list of considerations<br />

(Cronbach 1970; Hanna 1993; Cunningham<br />

1998) the following views:<br />

Tests must be valid and reliable (see<br />

Chapter 6).<br />

The administration, marking and use of the<br />

test should be undertaken only by suitably<br />

competent/qualified people (i.e. people and<br />

projects should be vetted).<br />

Access to test materials should be controlled,<br />

thus test items should not be reproduced apart<br />

from selections in professional publication;<br />

the tests should be released only to suitably<br />

qualified professionals in connection with<br />

specific professionally acceptable projects.<br />

Tests should benefit the testee (beneficence).<br />

Clear marking and grading protocols should<br />

<br />

<br />

<br />

<br />

exist (the issue of transparency is discussed in<br />

Chapter 6).<br />

Test results are reported only in a way that<br />

cannot be misinterpreted.<br />

The privacy and dignity of individuals should<br />

be respected (e.g. confidentiality, anonymity,<br />

non-traceability).<br />

Individuals should not be harmed by the test<br />

or its results (non-maleficence).<br />

Informed consent to participate in the test<br />

should be sought.<br />

Computerized adaptive testing<br />

Computerized adaptive testing (Wainer 1990;<br />

Aiken 2003: 50–2) is the decision on which<br />

particular test items to administer, which is based<br />

on the subjects’ responses to previous items. It is<br />

particularly useful for large-scale testing, where a<br />

wide range of ability can be expected. Here a test<br />

must be devised that enables the tester to cover<br />

this wide range of ability; hence it must include<br />

some easy to some difficult items – too easy and it<br />

does not enable a range of high ability to be charted<br />

(testees simply getting all the answers right), too<br />

difficult and it does not enable a range of low ability<br />

to be charted (testees simply getting all the answers<br />

wrong). We find out very little about a testee if we<br />

ask a battery of questions which are too easy or too<br />

difficult. Further, it is more efficient and reliable if<br />

atestcanavoidtheproblemforhighabilitytestees<br />

of having to work through a mass of easy items in<br />

order to reach the more difficult items and for low<br />

ability testees of having to try to guess the answers<br />

to more difficult items. Hence it is useful to have<br />

atestthatisflexibleandthatcanbeadaptedto<br />

the testees. For example, if a testee found an item<br />

too hard the next item could adapt to this and be<br />

easier, and, conversely, if a testee was successful<br />

on an item the next item could be harder.<br />

Wainer (1990) indicates that in an adaptive<br />

test the first item is pitched in the middle of<br />

the assumed ability range; if the testee answers it<br />

correctly then it is followed by a more difficult item,<br />

and if the testee answers it incorrectly then it is<br />

followed by an easier item. Computers here provide<br />

an ideal opportunity to address the flexibility,<br />

Chapter 19

Hooray! Your file is uploaded and ready to be published.

Saved successfully!

Ooh no, something went wrong!