Preschool Prediction of Low Achievement in Basic School Skills

Abstract

T he purpose of the present study was to determine the extent to which children identified on preschool readiness tests as "high risk" for development of learning difficulties do, in fact, demonstrate low achievement in basic school skills by the time they have completed several years of elementary school. A review by Keogh and Becker (1973) cited nine studies that attempted to locate children with potential learning problems in kindergarten or earlier. However, only four of these studies reported the percentage of correct predictions. A recent review of research published through 1973 (Schaer & Crump 1976) points out the lack of consensus regarding the feasibility of early identification of learning disabilities. Studies conducted from 1973 to the present provide no more reason to be confident about the validity of early identification of learning problems. False-positive rates are often high, and rates of correct prediction differ for failing as well as successful students (Dimitrovsky & Almy 1975, Eaves, Kendall, & Crichton 1974, Feshbach, Adelman, & Fuller 1974).If significant correlations are found between scores on a preschool instrument and later school achievement, it has generally been assumed that the preschool instrument has been established as a valid means of identifying children at risk for developing future learning problems. However, educators who are currently classifying children as at risk or not at risk based on the child's performance on such preschool measures may be making unwarranted assumptions regarding the validity of predictions for individual children. SUBJECTS AND MEASURESThe 732 subjects in the present investigation were drawn from the 1,613 participants in the Educational Follow-Up Study (EFS) (Rubin & Balow 1977), a prospective longitudinal investigation of the educational and behavioral outcomes associated with perinatal and early childhood conditions and events. The EFS subjects, born at the University of Minnesota Hospitals over a five-year period during the 1960s, are normally distributed on measures of IQ. The Metropolitan Readiness Tests (MRT) were administered to EFS subjects at age 5 the summer prior to kindergarten entrance and again at age 6 prior to entering the first grade. The word meaning, spelling, and arithmetic computation sections of the Stanford Achievement Test (SAT) were individually administered during the summer of the calendar year in which subjects reached their ninth birthday. At the time of the SAT administration, 212 (29.0£) of the subjects had completed grade 2, and 520 (71.0*) had completed grade 3. All EFS subjects who had been administered the MRT at both ages 5 and 6 as well as the SAT at age 9 at the time these data were collected were included in the present study. METHODS Correlations between scores on the MRT andeach of the three subtests of the SAT were first computed separately for the five-year and sixyear MRT performances. The MRT scores from 62 Journal of Learning Disabilities at UNIV OF CALIFORNIA SANTA CRUZ on April 5, 2015 ldx.sagepub.c...

Improving the Prediction of Reading for the Individual Child

“…Rubin, Balow, Dorle & Rosen (1978) concluded that present evidence will not support attempts to make individual predictions as to which preschool children will later experience serious academic difficulties. Lindsay and Wedell (1982) expressed the opinion that there is a lack of evidence for the usefulness of screening measures in identifying educationally "at risk*' children.…”

Comparative Validity of Two Preschool Screening Tests

“…Therefore, the variance of scores for 4-year-olds functioning at or above average level may be primarily determined by similar items on both tests, while the variance of scores at the low end of the distribution may be determined by dissimilar items. Other predictive validity studies (Feshbach, Adelman, & Fuller 1977, Rubin, Balow, Dorle, & Rubin 1978 showing a similar discrepancy between correlational and classificational analyses have interpreted correlational validity coefficients as more appropriate for predicting criterion success (i.e., academic competence), while classification analyses better indicate prediction of criterion failure.…”

The Prediction of Good and Poor Reading Before Kindergarten Entry: A 4-Year Follow-Up

“…Jansky (1978) states that no kindergarten battery will predict failing readers at better than a 75% level. It has also been stressed that a significant correlation between preschool test scores and later school achievement does not mean that the preschool instrument is valid for identifying children at risk, and that assumptions with regard to prediction for individual children are unwarranted (Rubin, Balow, Dorle, & Rosen, 1978). In many studies, prediction of average or high achievement appears to be more accurate than the prediction of low achievement (Powell & Parsley, 1961;Rubin et al, 1978;Satz, Taylor, Friel, & Fletcher, 1978).…”

“…It has also been stressed that a significant correlation between preschool test scores and later school achievement does not mean that the preschool instrument is valid for identifying children at risk, and that assumptions with regard to prediction for individual children are unwarranted (Rubin, Balow, Dorle, & Rosen, 1978). In many studies, prediction of average or high achievement appears to be more accurate than the prediction of low achievement (Powell & Parsley, 1961;Rubin et al, 1978;Satz, Taylor, Friel, & Fletcher, 1978). For example, the mean accuracy in predicting good and poor reading ability for each of two kindergarten measures was about 75% a year later (Feshbach, Adelman, & Fuller, 1974); accuracy was very high for predicting good reading, yet approximately 70% of the poor readers were not detected by either measure.…”