Calibration of a reading comprehension test for Portuguese students

Reading comprehension assessments are important for determining which students are performing below the expected levels for their grade's normative group. However, instruments measuring this competency should also be able to assess students' gains in reading comprehension as they move from one grade to the next. In this paper, we present the construction and calibration process of three vertically scaled test forms of an original reading comprehension test to assess second, third and fourth grade students. A sample of 843 students was used. Rasch model analyses were employed during the following three phases of this study: (a) analysis of the items' pool, (b) item selection for the test forms, and (c) test forms' calibration. Results suggest that a one dimension structure underlies the data. Mean-square residuals (infit and outfit) indicated that the data fitted the model. Thirty items were assigned to each test form, by selecting the most adequate items for each grade in terms of difficulty. The reliability coefficients for each test form were high. Limitations and potentialities of the developed test forms are discussed.