Psychometric Evidence for the Reading Screening Test–Digital Version in Portuguese Grade 1 Students

Early identification of reading difficulties may facilitate timely assessment, monitoring, and educational support. This study examined validity evidence for the Reading Screening Test–Digital Version (TRL-DV) among Grade 1 students in Portugal. A validation sample (N = 208) was used to examine internal structure, internal consistency, relations with external variables, and classification accuracy. An independent large psychometric sample (N = 2708) was used to examine internal structure, internal consistency, item functioning, precision, and item reach. The operational total score showed high internal consistency in both samples (KR-20 = 0.928 and 0.952), although these values may be inflated by common sequences of zero scores associated with unreached items. The correlated two-factor model outperformed the unidimensional model on several fit indices but did not yield a consistently admissible, adequately fitting solution. Scores correlated positively with word and pseudoword reading, rapid automatized naming, digit span, and teachers’ literacy grades. Item position and the decreasing probability of reaching later items contributed substantially to operational performance. An exploratory ROC analysis against a sample-derived criterion yielded AUC = 0.920. The findings support interpreting the total score as an indicator of decoding-related reading performance under timed digital administration and suggest potential screening value; component scores and classification thresholds require further validation.