The Deep Cloze Test of Global Inferencing in Reading Comprehension: Structure and Validity Among Early Adolescents.

Abstract

Recent declines in Finnish students' reading performance highlight the need for efficient assessment tools that identify specific processes underlying reading comprehension. This study aimed to adapt and validate the deep cloze test for Finnish sixth graders. The test is designed to measure the ability to draw global coherence inferences for reading comprehension and has previously been validated for adults. The test was adapted for Finnish-speaking early adolescents through a structured, two-phase process designed to ensure developmental and linguistic appropriateness while preserving the intended inference demands of the original measure. Students (N = 275, M_age = 12.72) completed the deep cloze test as well as tests of sentence verification, vocabulary, and reading comprehension. A confirmatory factor analysis supported the unidimensional structure of the deep cloze test. Consistent with the design of the task and previous research, the underlying dimension was interpreted as reading comprehension that requires readers to draw global coherence inferences. Item Response Theory (IRT) analyses showed that each item had reliable discriminatory power. Deep cloze test scores were associated with scores on a normed multiple-choice reading comprehension test, a vocabulary knowledge test, and a sentence verification test, providing evidence of convergent and concurrent validity. The results suggest that the deep cloze test is a promising measure for assessing reading comprehension processes that require readers to draw global coherence inferences.