The notion of washback, that testing influences teaching, is explored and a series of possible hypotheses are advanced. The empirical research in general education and in language education is reviewed to determine whether washback actually exists, how it can be measured, and what accounts for its form. Proposals for future research are suggested.…

Discusses predictive and conceptual validity, reports on two experiments on each type of test validity with relation to language tests, and indicates some problems posed by these two forms of evaluation. (AM)

This report states the issues involved in the use of communicative tests, oral and written, to assess grammatical correctness among students of English as a Foreign Language (EFL). Experience with two English tests administered at the beginning and end of the practice teaching period of Dutch students training to become EFL teachers indicates…

The construction of objective tests for native language reading comprehension is described. The tests were designed for the early secondary school years in several kinds of schools, vocational and non-vocational. The description focuses on the use of the Rasch model in test development, to develop a large pool of homogenous items and establish…

The traditional multiple-choice reading comprehension test of English as a second language, used in the Dutch school-leaving examinations, has been criticized for its apparent lack of construct validity. The Dutch National Institute for Educational Measurement has conducted a number of studies to determine whether there is a different skill…

In the Dutch national examinations, reading comprehension tests are used for all languages. For the native language, reading comprehension is tested with reading passages and related questions to which the test-taker provides his own response, not choosing from a group of alternatives. One problem encountered in testing with these items is…

The Rasch model for test analysis is a latent-trait model, which specifies the relationship between observable test performance and the unobservable traits or abilities assumed under test performance. In most cases, the test constructor has no clue as to whether the latent traits postulated by the model are indeed the abilities he wants to…

Dutch secondary schools offer educational programs on four tracks, vocational and nonvocational. At the end of each program, a final examination is administered, with the test content areas determined by law, by the schools, and by the student's individual program. Assessment is by means of a written, nationally administered exam and…

Since the mid-1960s, the use of objective testing in education, and multiple-choice tests in particular, has increased dramatically in the Netherlands. Since the founding of the Dutch National Institute for Educational Measurement in 1968, the construction of many achievement tests has been concentrated there. This general changeover from…

Presented in this symposium reader are nine papers, four of which deal with the theory and impact of the Rasch model on language testing and five of which discuss final examinations in secondary schools in both general and specific terms. The papers are: "Introduction to Rasch Measurement: Some Implications for Language Testing" (J. J.…