Processing speed and timed academic skills in children with learning problems

Abstract

Information processing speed is commonly measured in intelligence and neuropsychological testing, and the scores from speed measures are considered in diagnostic and management recommendations for students with academic learning problems. However, this score usage often depends on assumptions about strong relationships between cognitive speed and the ability to perform actual academic tasks under time pressure. The primary purpose of the present study was to test the strength of these relationships empirically. In the present study, children with prior learning disability diagnoses (146 girls and 301 boys, ages 10–14 years old) completed diagnostic batteries that included measures of cognitive speed as well as timed academic skills. The relationships between the two types of measures were often modest (median r = 0.25), and the gap between processing speed and timed academic scores was typically approximately 1 standard deviation. The pattern of relationships suggested that superficial similarity in stimuli and task demands affected the strength of associations. These results suggest that timed academic skills cannot be reliably estimated based on processing speed scores, and there will often be significant gaps between the two. Therefore, making diagnostic judgments (e.g., learning disability diagnoses) or management recommendations (e.g., for extended time testing accommodations) should be based on more direct measures of relevant academic skills.

Keywords:

• Academic skills
• assessment
• processing speed

Notes

1 Generally, the scores and tasks from different editions of the same test were based on virtually identical stimuli. However, the WJ-IV-COG processing speed composite was based on different tasks in the 3^rd and 4^th editions.

2 Owing to the archival nature of the study, we did not have access to test protocols for all participants, which would have allowed for sample-specific reliability analyses based on item-level data.

3 In the vast majority of cases, the bivariate sample sizes for our correlational analyses also had at least 50 participants; see below.

4 Consistent with typical presentation of statistical results, the numbers in the parentheses here represent degrees of freedom (rather than sample sizes) for the statistical tests on the significance of r.