A construção e validação de um teste com referência a critério para medir as competências básicas de leitura e escrita para o Programa de Alfabetização Funcional do MOBRAL

• Ano de defesa: 1981
• Autor(a) principal: Pereira, Regina Coeli Avila Alves
• Tipo de documento: Dissertação
• Tipo de acesso: Acesso aberto
• Idioma: por
• Link de acesso: http://hdl.handle.net/10438/9182
• topic: Educação; Testes e medidas educacionais - Brasil; Alfabetização de adultos - Brasil

The need for developing adequate instruments to evaluate educational programs is quite urgente presently. According to recent studies in the field of educational measurements, this need is fulfilled by the criterion-referenced tests. The construction of two criterion-referenced tests to evaluate basic competences in the areas of reading and writing of MOBRAL's Functional Literacy Program was the objective of this study. Firstly, four basic competences were selected and described to be measured by each test. The selection of such competencies was based on the objectives of the in the Teacher's Manual (1977). Secondly, specifications for each selected in the two areas were created. Based on Program stated competence these specifications, thirty-two items for the reading test and fourty items for the writing test were constructed. Such items were randomly distributed into five experimental forms for each teste The reading test has multiple-choice items and the writing test has subjective answer. The correcting criteria of minimum standards of performance were established for each test. Means and standard deviations were the statistics used in the analysis of the tests. The results of this analysis allow these minimum standards of performance to be kept in subsequent studies. Test-item analysis was developed in two stages: (a) discriminatory rates were calculated in the traditiona1 way, and (b) sensitivity rates were calculated through the process developed by Gronlund (1976). The results of the studyallowed the utilization of all items. The items were divided into two equivalent forms for each test, each one subdivided into three subtests to assess an isolated measure of the described competence. In order to examine the validity, rates of descriptive validity and area selection were established. The reliability was assured with the construction of more than five items for Gronlund's competence (1976).