Construction of a task for assessing reading comprehension in elementary school: Pilot study

Abstract

Reading comprehension is the process of extracting meanings and building a mental representation of the read content, which requires the proper functioning of cognitive and linguistic skills. In Brazil, there is a lack of instruments for the assessment of reading comprehension that report validity, reliability, and analysis of the items. This pilot study reports some psychometric procedures of a Task of Reading Comprehension for Children. Students (N = 81) from third to fifth year of a public school in the city of São Paulo read seven texts (narrative or expository) and answered open questions. It was analyzed: (a) three main effects on accuracy, namely, the text, the cognitive process used to solve the questions, and the school grade (ANOVA for repeated measures); (b) the psychometric parameters of difficulty and discrimination (Item Response Theory). As expected, the ANOVA showed that the accuracy was lower: (a) for the more complex texts; (b) for three kind of questions: inferential, elaborative, and the ones that required the construction a mental representation of the text; (c) the younger children. Item analysis showed that most questions presented low discrimination indices and functioned to lower levels of ability. The results indicate that, in general, the items proved promising for the assessment of reading comprehension for the age range investigated, but they also point to the need for adjustments in some texts and questions.