4
Psychology in the Schools 1980. 17, 442-445 RELIABILITY AND CONCURRENT VALIDITY OF LIGHT’S RETENTION SCALE’ JONATHAN SANDOVAL Universiiy of Cali/ornia, Davis Light’s Retention Scale was completed for 123 candidates to repeat the first grade. These same first-grade children were tested with a number of measures of maturity, including achievement tests, measures of intellectual functioning, visual-motor in- tegration, self-concept, and physical maturity. Results indicated that Light’s total score is not sufficiently reliable and has little concurrent validity for making retention decisions. The total score should never be used. H. Wayne Light designed his Retention Scale (Light, 1977) as a device to assist school personnel in thinking about children they were considering for grade level reten- tion. The problem of deciding when to require a child to repeat a grade in school is one that causes great concern and anxiety to school personnel and parents alike. The need for an objective instrument for aiding in this decision making is obvious. Light’s scale was not constructed through a large-scale research program. Rather, he reviewed the available literature on the topic and summarized it in 19 factors he believes to be related to an appropriate retention. Items such as intelligence, physical size, and whether or not the child has been previously retained are found on the scale. Although he intended his scale to be used as a counseling device, he did affix numbers to the various options in each item, reflecting his belief in each option’s importance. In addition, in the manual, he has listed categories derived from total scores summed across all of the items. These categories are to be used in making the decision. No norms are provided for the scale in the manual. Recently, Dan Watson (1979) presented data indicating that Light’s scale may not be related to gains in achievement made during the retained year in elementary school children. Watson did not report reliability or concurrent validity data for the scale in his study. In addition, he collected information from children in a number of primary grades. The present study was designed to add to the information available on the psy- chometric use of Light’s Retention Scale. METHOD Subjects Fifty-three first-grade teachers from 37 participating public and private schools in the Sacramento-Solano County area of Northern California completed the scale. The cooperating teachers identified in February and March all children they believed might profit from retention, and about whom they would have to make a retention decision in May. Approximately 180 parents of these children were contacted, and 148 agreed to participate in a study of nonpromotion. Teachers were able to respond to all items on 123 children. Thus, the children were candidates to repeat the first grade during the 1979-80 school year and were the type of children for whom the scale was designed. Requests for reprints should be sent to Jonathan Sandoval, Dept. of Education, College of Letters & ’This study was supported by grant MH 28765-01 from the National Institute of Mental Health 442 Science, University of California, Davis, Davis, CA 95616.

Reliability and concurrent validity of light's retention scale

Embed Size (px)

Citation preview

Page 1: Reliability and concurrent validity of light's retention scale

Psychology in the Schools 1980. 1 7 , 442-445

RELIABILITY AND CONCURRENT VALIDITY OF LIGHT’S RETENTION SCALE’

JONATHAN SANDOVAL

Universiiy of Cali/ornia, Davis

Light’s Retention Scale was completed for 123 candidates to repeat the first grade. These same first-grade children were tested with a number of measures of maturity, including achievement tests, measures of intellectual functioning, visual-motor in- tegration, self-concept, and physical maturity. Results indicated that Light’s total score is not sufficiently reliable and has little concurrent validity for making retention decisions. The total score should never be used.

H. Wayne Light designed his Retention Scale (Light, 1977) as a device to assist school personnel in thinking about children they were considering for grade level reten- tion. The problem of deciding when to require a child to repeat a grade in school is one that causes great concern and anxiety to school personnel and parents alike. The need for an objective instrument for aiding in this decision making is obvious. Light’s scale was not constructed through a large-scale research program. Rather, he reviewed the available literature on the topic and summarized it in 19 factors he believes to be related to an appropriate retention. Items such as intelligence, physical size, and whether or not the child has been previously retained are found on the scale. Although he intended his scale to be used as a counseling device, he did affix numbers to the various options in each item, reflecting his belief in each option’s importance. In addition, in the manual, he has listed categories derived from total scores summed across all of the items. These categories are to be used in making the decision. No norms are provided for the scale in the manual.

Recently, Dan Watson (1979) presented data indicating that Light’s scale may not be related to gains in achievement made during the retained year in elementary school children. Watson did not report reliability or concurrent validity data for the scale in his study. In addition, he collected information from children in a number of primary grades. The present study was designed to add to the information available on the psy- chometric use of Light’s Retention Scale.

METHOD Subjects

Fifty-three first-grade teachers from 37 participating public and private schools in the Sacramento-Solano County area of Northern California completed the scale. The cooperating teachers identified in February and March all children they believed might profit from retention, and about whom they would have to make a retention decision in May. Approximately 180 parents of these children were contacted, and 148 agreed to participate in a study of nonpromotion. Teachers were able to respond to all items on 123 children. Thus, the children were candidates to repeat the first grade during the 1979-80 school year and were the type of children for whom the scale was designed.

Requests for reprints should be sent to Jonathan Sandoval, Dept. of Education, College of Letters &

’This study was supported by grant MH 28765-01 from the National Institute of Mental Health

442

Science, University of California, Davis, Davis, CA 95616.

Page 2: Reliability and concurrent validity of light's retention scale

Light’s Retention Scale 443

Measures At the same time the teachers completed the scale, during April and May, the pupils

were administered the Letter-Sounds and Reading Comprehension tests from the California Achievement Tests (Tiegs & Clark, 1977), and the Letter Identification and Word Identification Tests of the Woodcock Reading Mastery Tests (Woodcock, 1973). They also completed the Numeration, Geometry and Symbols, Addition, Subtraction, Word Problems, and Time subtests of the Key Math Diagnostic Test (Connolly, Nacht- man, & Pritchett, 1976). In addition to these achievement measures, they took the Vocabulary and Block Design Subtests of the WISC-R, the Developmental Test of Visual-Motor Integration (Beery, 1967), the McDaniel-Piers Young Children’s Self- Concept Scale (McDaniel, 1978), and had their height and weight measured. All measures were selected to tap different dimensions of maturity or readiness: achievement-intellectual, visual-motor, emotional, and physical.

The Retention Scale was scored, using both Light’s weights assigned to the response options and a nonweighted score based on the implied continua in the options. For items Intelligence, Present Level of Academic Achievement, and Student’s Attitude About Possible Retention, options assigned the identical weight by Light were given separate simple weights. Since items have from 2 to 6 options, the simple weights for the options ranged from 0 to 5 . Procedure

Once the simple weights and Light’s weights had been assigned, the results of both the simple and Light’s scoring were factor analysed by principal factoring with iteration and varimax rotation, using the SPSS Factor Program (Nie, Hull, Jenkins, Stein- brenner, & Bent, 1975). Next, Cronbach’s Alpha, a measure of internal consistency, was computed for the scale total (both weightings) and for the scale derived from the sum of the items most highly correlated with the first rotated factor (short form total). Finally, the correlations of the total scores and selected items were calculated with the concurrent measures of maturity.

RESULTS AND DISCUSSION Table 1 presents the results of the factor analysis, listing the eigenvalues, percent of

explained variance, and the varimax rotated factor matrix. Only factors with eigenvalues above 1.00 are tabled. The three factors are nearly identical, with the transposition of factors two and three, whether the simple weighting or Light’s weighting is used. The simple weighting produces factors with slightly higher amounts of explained variance. The first factor, which may be termed “School Problems and Deprivation,” accounts for the majority of the variance and may be considered a short form. The other less impor- tant factors may be labeled “attendance” and “sex and starting school late.”

Table 2 displays the means, standard deviations, and Alpha reliability statistics for the total score and the short form total for both Light’s weighting and the simple weighting. The distributions of scores from both weightings are markedly positively skewed. Clearly, the short form has higher reliability, but the reliability is the same for either weighting.

The concurrent validity was tested by correlating the total score with the measures of maturity and achievement collected at the same time. For there to be concurrent validity, moderate correlations are necessary. In this instance, none of these correlations for any of the total scores reached significance at p1.05.

As appropriate, individual items were correlated with parallel measures, with the following result: Item Intelligence (sample weighting) correlated .37 with WISC-R

Page 3: Reliability and concurrent validity of light's retention scale

444 Psychology in the Schools, October, 1980, Vol. 17, No. 4.

TABLE 1 Eigenvalues of Factors and Varimax Rotated Factor Matrix of

Light’s Retention Scale with two Response Weighrings

Light’s Weighting Simple Weighting

Factor

Eigenvalue

Variance

ITEM Attendance Intelligence Achievement Size

Age Sex Siblings Previous Retention Learning Disabilities Attitude Participation Motivation Delinquency English Grade* Transiency Emotional Problems Experiential Background Immaturity

1

4.12

51.9

-.11 .03

-.04 -.01 -.oo

.02 -.06

.02

.64

.25

.73

.63

.83 .78 - .64 .55 .50 .59

2

1.20

14.9

.47 -.19

.42

.21

.03

.01

.07

.04 -.01

.39 -.35 -37 -.02 -.13 - .12 .06 .31

-.I 1

3 1

1.04 5.39

12.9 55.8

.12 -.12

.31 -.01 -.07 -.11 -.01 -.01

.09 -.02

.51 .02

.02 -.04

.71 .01

.06 .84 -.01 .27 -.03 .90

.13 .76

.05 .90

.04 .92 - - .14 .68

-.04 .66 .21 .46

-.11 .66

2

1.48

15.3

.05 -.17 -.lo .04 .04 .61

-.05 .97 .02 .02

-.06 .06 .06

-.01 -

.25 -.02 -.09 -.09

3

1.24

12.9

.a0

-.lo .02 .04 .o 1 .04 .02 .m

-.02 .37

-.05

-.18 .01

-.01 - .09 .04 .17

-.07

*All children were in the first grade,

Vocabulary and - .01 with Block Design. The Item Achievement (simple weighting) cor- related with Letter Identification .23, with Letter Sounds .16, with Word Identification .5 1 , with Reading Comprehension .54, and with the total of the Key Math subtests .33. Item Size correlated .70 with Height and .65 with Weight. Item Student’s Age correlated .22 with actual age. All of the above correlations are significant at p1.05 except Block Design and Letter Sounds. The item History of Learning Disabilities correlated with none of the achievement measures. Item Emotional Problems did not correlate with any of the scores on the McDaniel-Piers Young Children’s Self-concept scale.

Light’s Retention Scale would not seem to be a reliable or valid instrument in its present form. Light’s weights do not seem to be a problem, although they are not superior to simple weights.

Half of the items factor together (the items on the back page of the scale) to make an acceptably reliable short form total. Salvia and Ysseldyke (1978) recommended that .90 be used as a standard for accepting the reliability of a score, if the score is to be used

Page 4: Reliability and concurrent validity of light's retention scale

Light’s Retention Scale 445

TABLE 2 Means, Standard Deviations, and Alpha Reliability Statistics for

Total and Short Form of Light’s Retention Scale by Weighting System

Standardized x S D Alpha Alpha

Light’s Weights Total Score 40.48 16.76 .73 .76 Short Form 10.21 18.12 .88 .90

Simple Weights Total Score 10.78 19.72 .73 .76 Short Form 5 .m 16.79 .90 .93

for making educational decisions about individual children. The total score reliability does not meet this minimum, although the short form total does.

Possibly as a consequence of low reliability, the total score was not found to have concurrent validity in the present investigation. It is not surprising that Watson (1979) found no predictive validity.

Many teachers completing the scale found the task hard. On some items the options offered did not suitably describe many children. For example, options on Academic Achievement are: (a) one year behind in all academic areas, (b) more than one year behind in all areas, (c) at grade in reading, but one year behind in others, (d) at or above grade in reading and spelling, and (e) at or above in all areas. These choices did not cover all of the combinations perceived by teachers. A most common situation, behind in reading but at grade level in other subjects, was not available as an option. This item plus others on motivation and attitude were often left blank. The options on Emotional Problems were particularly vague. The scale needs revision in item content if it is to be accepted by teachers.

So far, Light’s retention scale does not meet the conventional standards for a psy- chometric device intended for use in schools. It may have utility as a counseling aide, but a total score must not be computed or used in decision making.

REFERENCES BEERY, K. E. CONNOLLY, A. J. , NACHTMAN, W., & PRITCHETT, E. M. Key Math Diagnostic Arithmetic Test. Circle Pines,

LIGHT, H. W. Light’s Retention Scale. San Rafael, CA: Academic Therapy Publication, 1977. MCDANIEL, E. The McDaniel-Piers Young Children‘s Selj-Concept Scale. West Lafayette, IN: Purdue

University Department of Education, 1978. NIE, N., HULL, C., JENKINS. J., STEINBRENNER, & BENT, D. SPSS: Statistical Package for the Social

Sciences, 2nd ed. New York: McGraw-Hill, 1975. SALVIA, J. , & YSSELDYKE, J. E. Assessment in special and remedial education. Boston: Houghton Mifflin,

1978. TIEGS, E. W., & CLARK, W. W. California Achievment Tests. Monterey, CA: California Test

Bureau/McGraw Hill, 1977. WATSON, D. The relative eflciency of the Light Retention Scale in identifying children for retention. Paper

presented at the annual convention of the National Association of School Psychologists, San Diego, March, 1979.

Woodcock Reading Mastery Tests. Circle Pines, MN: American Guidance Service, 1973.

Developmental Test of Vkual-Motor Integration. Chicago: Follett, 1967.

MN: American Guidance Service, 1976.

WOODCOCK, R. W.