Methods of Assessment

Reliability versus Validity in the use of Selective Deletion (Cloze) written placement tests.

Reliability versus Validity

Firstly, the selective deletion format is purely a written test of grammatical structure and written vocabulary skills (e.g. collocation) and knowledge of the right word. It is not entirely receptive because the students are required to produce and fill in the right word with the correct spelling.

However, aural / oral skills remain untested - the test is not long enough to enable a teacher to get round a large group of students and to administer individual aural/oral interviews. The emphasis on written skills is unfair to those students who can speak and comprehend English, but who are not familiar with the Roman alphabet. It allows students with incomprehensible and socially unacceptable pronunciation to gain higher scores than they should.

It is by no means a text of communication. It is mostly exposition. There is hardly any written dialogue. There are too few items in the test, yet it is still very tedious to mark.

Because of these short-comings, I produced an entirely different test for one of our courses. There is a listening element recorded on audio-cassette to ensure reliability. Our cassette-players perform perfectly in all environments! The rubric for the grammar and vocabulary sections is MULTIPLE CHOICE - one that is familiar to most students and only requires them to select between Answers A, B, C, or D. The process of selecting between four possible alternatives makes the test very receptive, but the test is long enough to enable us to get round to talk individually with each student. Students who finish early are encouraged to write freely about their need for English, which adds a productive element to the test and allows us to diagnose their purposes and linguistic weaknesses.

The ANSWER SHEETS for the test papers contain grids (students check ONE box only for each item) so they can be marked very quickly using black overlays which are only transparent in the areas where correct boxes can be ticked. It is necessary to do a quick check to see that the student has not ticked all the given alternatives for each item. Since there are only FOUR alternatives for each item, students with no knowledge of English can be expected to achieve scores of at least 25 %. This is taken into account in the marking scheme.