Reliability and Validity of Selected Criteria of "Light's Retention Scale"

Abstract

Grade is an old and persistent practice in North American schools. Estimates vary from region to region, but the proportion of elementary students repeating grade appears to have altered very little over the past two decades (Canadian Education Association, 1989, p. 3). Meanwhile, researchers have questioned whether grade improves students' performance. A large volume of research suggests that the potential for positive outcomes of grade is consistently outweighed by the potential for negative outcomes (Holmes & Matthews, 1984). Still, minority of students do benefit (Shepard & Smith, 1989; Westbury, 1994). Both educators and parents would be less anxious if they could determine in advance which students would benefit from retention. Light's Retention Scale (hereafter, LRS; Light, 1986a) is meant to provide a format that aids the school professional in determining whether the elementary or secondary school student would benefit from retention (Light, 1986b, p. 5). The scale includes 19 criteria: sex, age, knowledge of the English language, physical size, present grade placement, previous grade retentions, number of siblings, parents' school participation, experiential background, transiency, school attendance, estimates of intelligence, history of learning disabilities, present level of academic achievement, attitude towards possible retention, motivation to complete school tasks, immature behaviour, emotional problems, and history of delinquency. Parents and/or teachers rate the potential candidate from 0 to 5 on the 19 items; score of 0 indicates strong belief that the child would benefit from and 5 indicates strong belief that he or she would not. The sum of the ratings is used to categorize the student's suitability for as excellent, good, fair, marginal, poor, or unsuitable (Light, 1986b, p. 11). Neither LRS nor the accompanying manual (Light, 1986b) provides norms. Light explains that LRS is not psychometric instrument and should be used only as counselling tool (Light, 1986a, p. 1). He does, however, state that the weights assigned to each of the 19 categories were determined after careful analysis of the research on each (Light, 1986b, p. 12). Several researchers investigated earlier editions of LRS. Watson (1979) suggested that the scale may not predict achievement gains during the repeated year. Sandoval (1980) demonstrated that an earlier edition of the scale lacked validity and reliability. Sandoval (1982) further found that Light's scores did not correlate with the academic performance, self-concept, or mental health of retained stu-