Composition of the LASS 11-15 suite

Outline of tests

The LASS 11-15 suite comprises three attainment tests (single word reading, sentence reading and spelling), one ability test (reasoning) and four diagnostic tests (auditory memory, visual memory, phonic skills and phonological processing). An outline of each test is given in Table 4. Three of the eight tests (Sentence Reading, Spelling and Reasoning) are adaptive, i.e. they are based on statistical item response theory (IRT), whereby each test item is selected from a large bank of items, each of which is of known difficulty for students of that age group. The remaining tests are progressive in format, i.e. they utilise a graded series of items of increasing difficulty for students of that age group. In some of the tests there is a discontinuation algorithm built in, whereby the test will automatically cease once the student’s current attainment or ability level has been exceeded beyond reasonable statistical error; otherwise, the student must attempt all items in the test.

For each test, instructions are spoken by the computer, and practice items are given to familiarise the student with the test requirements. When the student has completed the practice items, the test phase begins.

Adaptive assessment

The term ‘adaptive testing’ refers to any technique that modifies the nature of the test in response to the performance of the test-taker. Paper-based tests are static instruments, fixed in their item content, item order, and duration. By contrast, computer-based assessment can be dynamic. Since the computer can score performance at the same time as item presentation, it can modify the test accordingly, tailoring it to the capabilities of the individual taking the test much more effectively than has ever been possible before.

Conventional tests can be very crude instruments in which, much of the time, the individual’s abilities are not being assessed with great precision because the items are either too difficult or too easy. In an adaptive test the individual can be moved swiftly to that zone of the test that will most efficiently discriminate his or her capabilities, thus making assessment shorter, more reliable, more efficient, and often more acceptable to the person being tested. The savings in testing time are distinctive and can far outweigh any disadvantages of transferring from conventional methods to computer-based methods. For example, Olsen (1990) compared paperbased and computer-administered school achievement and assessment tests with computerised adaptive tests. The computer-based non-adaptive version took 50–75% of the time taken to administer the conventional version, while the testing time for the adaptive version was only 25% of the time taken for the paper-based version.

Table 4. Composition of the LASS 11-15 suite of tests

Sentence Reading Attainment Adaptive Cloze reading - completing sentences by identifying the missing word from a choice of five alternatives. No spoken assistance is given.
Single Word Reading Attainment Progressive Reading individual words out of context — identifying from a choice of five alternatives the printed word that corresponds to a spoken word.
Spelling Attainment Adaptive Spelling individual real words that are spoken by the computer.
Reasoning Ability Adaptive Non-verbal intelligence - analogical reasoning where the correct item from a choice of six alternatives has to be selected in order to complete a spatial matrix.
Mobile Diagnostic Progressive Auditory sequential memory (digit span) — recall of between two and nine digits in correct (forwards) sequential order.
Cave Diagnostic Progressive Visual memory — immediate recall of objects and their spatial positions, beginning with two items and progressing to seven items.
Nonwords Diagnostic Progressive Reading individual nonwords — a pure measure of phonic decoding skills. For each nonword there is a choice from four spoken alternatives.
Segments Diagnostic Progressive Phonological processing ability — segmentation and deletion of syllables and phonemes in real words. For each item there is a choice from four spoken alternatives.

In each of the three adaptive tests in LASS, the program first gives the student a series of ‘probe’ items to determine the range of optimal item sensitivity for that student. These are followed by a series of test items starting in the range of optimal item sensitivity and increasing in difficulty until the student’s current attainment or ability level has been exceeded beyond reasonable statistical error, whereupon the test ceases. The program incorporates a facility to regress to easier items should it transpire that, by chance, the result of the probe items has overestimated the student’s approximate ability or current attainment level.