Does third grade discrepancy status predict the course of reading development?

Annals of Dyslexia, 2001 by Flowers, Lynn, Meyer, Marianne, Lovato, James, Wood, Frank, Felton, Rebecca

Lindamood Auditory Conceptualization Test (LAC) (Lindamood & Lindamood, 1979). This test assesses several aspects of phonemic processing including the ability to discriminate sounds, hold strings of sounds in short-term memory, and sequence and manipulate sounds within word-like strings (e.g., changing "tif" to "tof" or "tofs" to "ofs"). Difficult items are weighted more heavily and the authors report that a maximum converted score of 100 is expected by the end of sixth grade. Converted scores are reported as there is no standard score conversion. Lindamood, Lindamood, and Calfee (1974) reported LAC correlations between .66 and .81 across the grade span with Wide Range Achievement Test (WRAT) (Jastak & Bijou, 1946) reading and spelling subtests.

Test of Auditory Analysis Skills (TAAS) (Rosner, 1979). The TAAS measures phonemic awareness by requiring subjects to segment syllables, delete initial and final consonants, and segment consonant blends. Students are expected to perform this task perfectly by the end of the third grade. Raw scores are reported on this 13-item test as there is no standard score conversion. Rosner reported validation of his original test using Scholastic Aptitude Tests (SAT) scores (1971). Rapid Automatized Naming Test (RAN) (Denckla & Rudel, 1976). This naming test is a measure of fluency and automaticity in retrieving labels for highly practiced items (colors, numbers, objects, and lower-case letters) under strictly timed conditions. Stimuli are presented on individual cards containing five rows of randomly arranged items, ten to a row. Scores are reported as mean seconds with graphological symbols (letters/numbers) and nongraphological symbols (colors/ objects) combined because there is no standard score conversion. (See Meyer, Wood, Hart, & Felton, 1998, for the rationale behind this scoring procedure.) Wolf , Bowers, & Biddle (2000) summarized the evidence that the RAN contributes significant independent variance to reading skills as compared to tests of phonological awareness.

Diagnostic Interview for Children and Adolescents (RICA) (Herjanic & Reich, 1982). This structured diagnostic interview is keyed to criteria described in the Diagnostic and Statistic Manual of Mental Disorders, 3rd Edition. Parents or guardians were interviewed on the scale assessing inattentive, impulsive, and hyperactive features of attention deficit disorder in their third-grade children.

Four Factor Index of Social Position (Hollingshead, 1957). Socioeconomic status (SES) is determined from parental occupation and educational level. Information was obtained by interviewing parents of third-grade students.

STATISTICAL ANALYSIS

T-tests comparing discrepancy groups on subject demographics and reading skills measured in the third grade were conducted separately within poor readers and normal readers. Chi square tests were performed on the categorical variables of sex and race.

For each reading score of interest, we potentially had measurements on each child at third, fifth, eighth, and twelfth grades. We assumed that the measurements were independent from child to child but that they were correlated within each child. Developmental course was analyzed using a mixed effects regression growth curve model incorporating both fixed and random effects. We preformed separate analyses for poor readers and normal readers. Fixed effects-race, gender, SES level, and grade level-were associated with particular levels of variables used in the model and represent deviations from an overall mean. When modeling each of the reading measures, we allowed fixed effects to enter the model and also allowed for these effects to vary by grade by allowing interaction terms. We included an individual child random effect to account for the correlation of scores within each child, and because we were interested in making inferences about the population of children from which these particular children are drawn. After a good explanatory and predictive model was built, developmental course was investigated by adding to the model the effect of discrepant status and its interaction with grade level. The full model with discrepant status and its interaction with grade level was tested against the model without the interaction term using the likelihood ratio test. Developmental course was said not to differ between discrepant and nondiscrepant readers if this test was not significant.