From the Publisher

Essentials of Stanford-Binet Intelligence Scales(SB5) Assessment provides comprehensive instructions on optimal use of the Stanford-Binet, including helpful step-by-step administration, scoring, and interpretation guidelines. Using the popular Essentials format, this handy resource also covers test strengths and weaknesses, reliability, and validity, and describes clinical applications of its use. Students in courses on psychological assessment and appraisal will find this to be an indispensable tool to understanding and utilizing this popular measure of intelligence.

Product Details

• ISBN: 0471224049
  
• ISBN-13: 9780471224044
• Format: Textbook Paperback, 224pp
• Publisher: Wiley, John & Sons, Incorporated
• Pub. Date: August 2004
• Edition Number: 1
• Sales Rank: 210,323
• Series: Essentials of Psychological Assessment Series

• Table of Contents
• Read an Excerpt

Table of Contents

Read an Excerpt

Essentials of Stanford-Binet Intelligence Scales (SB5) Assessment

John Wiley & Sons

Chapter One

INTRODUCTION

New editions of nationally standardized tests provide modern wording, illustrations, enhanced measurement procedures, updated theory and research, and new standardizations, enhancing the validity of test interpretations. Such enhancements come at a price, however, in time, effort, and costs to the developers and consumers. After a 7-year revision project, the Stanford-Binet Intelligence Scale, Fifth Edition (SB5) (Roid, 2003b) was published with enhanced features, norms, and procedures. Initial reviews and receptions by users have shown that the revision possful in many respects, to be described in the chapters that follow.

The SB5 combines the point-scale format of the fourth edition (SB4) by Thorndike, Hagen, and Sattler (1986) with the age-level format found in previous editions such as the classic Forms L, M, and L-M (Terman & Merrill, 1937, 1960). Examiners begin a standard test administration by giving the two routing subtests: Object Series/Matrices (nonverbal) and Vocabulary (verbal). Estimates of ability in the nonverbal and verbal domains are obtained from raw scores on each routing test and used to tailor the remaining assessment to the examinee's functional ability. Simple conversion tables show the examiner which functional levels (ranging from easy to hard,Levels 1 through 6) of the nonverbal and verbal scales to continue testing. By adding the nonverbal routing test and an entire one-half of the SB5 in the nonverbal domain, the new edition provides excellent features for testing individuals with limited English or with communication difficulties. The nonverbal section requires a low language demand-that is, minimal receptive language and mostly nonverbal responses (pointing, moving pieces, etc.) by the examinee. Also, the two-stage testing procedure, with routing subtests first and functional-level sections next, provides highly precise estimates of cognitive ability in a relatively short period of time.

Thus, the SB5 is a wide-ranging, individually administered test battery. Norms were designed for ages 2 through 85+ years and the subtests cover five cognitive factors-Fluid Reasoning, Knowledge (crystallized ability), Quantitative Reasoning, Visual-Spatial Processing, and Working Memory-in both the verbal and nonverbal domains. Importantly, the SB5 is the first intellectual battery to cover five cognitive factors in both the nonverbal and verbal domains. Five nonverbal subtests and five verbal subtests measure each of the factors.

Many new features have been added to the SB5 and features of previous editions enhanced, as shown in Rapid Reference 1.1. Many of the new features were designed to enhance the usefulness of the SB5 for assessments with preschool children, individuals with mental retardation, and individuals with intellectual giftedness. In addition, many features were added to make the test easy for examiners to administer and score.

HISTORY

A number of the features of the design of the SB5 become apparent when the history of the Stanford-Binet and its various editions is reviewed. Developments began in France at the turn of the 20th century. Alfred Binet and Theodore Simon (1905) developed a useful tool to assess general intelligence, which is widely cited as the first major breakthrough in intelligence testing. As a member of a French governmental commission working on mental retardation, Binet developed a practical test, sensitive to different levels of cognitive development, that could be given during a clinical interview. Binet's early work is summarized in Rapid Reference 1.2.

Terman's 1916 Stanford Revision

Researchers in the United States, such as Goddard (1908) and Terman (1911), quickly saw the theoretical and practical value of Binet's work and began to adapt the work to the American context. Lewis Terman, of Stanford University, worked with Child (Terman & Child, 1912) to develop a preliminary revision of the Binet-Simon scale. Within a few years, Terman (1916) had extended the scale and collected data on more than 2,300 children and adolescents. The improved scale was published as the Stanford Revision and Extension of the Binet-Simon Scale by a division of Houghton-Mifflin Company called Riverside Press. Alternative versions of the Binet-Simon scales that presented intelligence as a singular dimension were distributed in the United States (e.g., the Goddard, 1910 version). However, Terman's 1916 revision retained Binet's concept of intelligence as a complex mixture of abilities, and is the only revision that has stood the test of time, remaining in publication to the present day. The standardization that Terman accomplished was quite rigorous for the early 1900s and increased the scale's technical quality. Also, the scale had thorough directions for examiners and used the ratio of mental age to chronological age first introduced by Stern (1914). This intelligence quotient (IQ) became the new standard for the assessment of intelligence.

Revisions of the Terman Scales in 1937, 1960, and 1972

Within 20 years of its release in 1916, the Stanford Revision emerged as the most widely used test of intellectual ability in America. The scale had several language translations and was used internationally. In subsequent years, Terman continue to experiment with easier and more difficult items to extend the measurement scale downward and upward and to increase the age range by collecting more standardization cases. As was the practice in test publishing at that time, an alternative form of the scale was also developed. Funding came from Stanford University grants and help came from colleague Maud Merrill, who served as codirector of the revision project. Interestingly, the new edition took 7 years to create-the same length of time devoted to the new fifth edition (Roid, 2003f)-and was called the New Revised Stanford-Binet Tests of Intelligence (Terman & Merrill, 1937).

The 1937 revision was standardized on 3,200 examinees aged 1 year 6 months to 18 years. Terman made efforts to include a broader representation of geographic regions and socioeconomic levels in the normative sample. Two alternative forms, Form L and Form M, were included. Improvements over the 1916 edition included greater coverage of nonverbal abilities, less emphasis on recall memory, extended range of the scale at the lower and upper ends, and more objectified scoring methods. The descriptions of the norming methods employed on Forms L and M and use of the Hollerith sorter (an early data-processing machine that sorted coded punch cards) for statistical analyses are quite impressive and interesting as examples of early psychometric methods (Terman & Merrill, 1937, p. 22).

As happens with any widely used test of ability or achievement, certain items of the 1937 edition became dated by the 1950s, and some of the toys used with the test had become difficult to obtain after World War II. Therefore, Terman and Merrill began a further revision of the scale based on the accumulated information and data collected since 1937. According to the preface in Terman and Merrill (1973, p. vi), "plans for the third revision had been formulated and were well on their way" when Terman died in 1956. Thus, the Stanford-Binet Intelligence Scale, Third Revision (Terman & Merrill, 1960) was published. Several new features were included in the third revision (now called the third edition), including use of the deviation IQ (standardized normative mean of 100 and standard deviation of 16) rather than the ratio IQ, and production of a combined Form L-M consisting of the most discriminating 142 items from the 1937 forms. Extensive item analyses had been conducted on data from nearly 4,500 subjects from the 1930s to 1950s and any items showing substantial change in difficulty were eliminated or adjusted in sequence. No newly created items were added to Form L-M. Six items were placed at each age level of the combined form (reduced from its previous eight items per level) and some directions were clarified.

After Maud Merrill retired, Robert L. Thorndike of Columbia University was asked to lead a project to collect new norms for the third edition. Thus, the same edition was reprinted with the new normative tables-an update of Form L-M (Terman & Merrill, 1973). Because the Cognitive Abilities Test (CogAT; Thorndike & Hagen, 1994) was being standardized at the same time as the 1972 renorming of the Stanford-Binet, Thorndike selected subjects and some siblings of subjects tested on the CogAT to compose the new norm sample. The stratification variables used on the sample (e.g., age, geographic region, ethnicity, community size) were similar to those used today, as were the levels of ability on the verbal portion of the CogAT. The items in the test remained essentially the same as on the 1960 revision, with two minor exceptions.

The 1986 Edition by Thorndike, Hagen, and Sattler

With the SB4 (Thorndike et al., 1986), the test took on a new appearance and structure. The SB4 was based on a four-factor, hierarchical model with general ability (g) as the overarching summary score. The four cognitive factors were Verbal Reasoning, Abstract/ Visual Reasoning, Quantitative Reasoning, and Short-Term Memory. The most significant change from previous editions, however, was the use of point scales for all subtests rather than the developmental age levels used in previous forms. Vocabulary was still retained as a routing test, allowing the test to be tailored to the examinee's verbal ability. Also, many classic Stanford-Binet tasks were retained, including Absurdities, Vocabulary, Matrices, Quantitative Reasoning, and Memory for Sentences-tasks also included in the SB5. Composite scores for each cognitive factor and profile scores for each subtest allowed a comprehensive examination of strengths and weaknesses among abilities within general intelligence.

THEORY AND STRUCTURE OF THE SB5: THE 2003 FIFTH EDITION

Based on the important research of Carroll (1993), the SB5 was constructed on a five-factor hierarchical cognitive model. The five factors were derived from the combined models of Carroll, Cattell (1943), and Horn (1965). The combination of models, now called the Cattell-Horn-Carroll (CHC) theory, normally lists 8 to 10 factors. Many of the supplemental factors, such as processing speed, auditory processing, and long-term retrieval, require specialized timing or test apparatus (e.g., tape recorders). However, the five cognitive factors of the SB5 (see Fig. 1.1) were selected based on research on school achievement and on expert ratings of the importance of these factors in the assessment of reasoning, especially in giftedness assessment. Also, the memory factor was shifted from an emphasis on short-term memory only, as in the fourth edition of the Stanford-Binet, to an emphasis on Working Memory. Therefore, the overall model shown in Figure 1.1 is a hierarchical g model with five factors emphasizing reasoning abilities that can be easily administered within a one-hour assessment. The important change in emphasis in the fifth edition is the duplication of all five cognitive factors in both the nonverbal and verbal domains.

The structure of the SB5 is also shown in Figure 1.1. The verbal and nonverbal domains include five subtests each, for a total of 10 profile scores. The nonverbal subtests require a small degree of receptive language and allow for pointing responses, the movement of puzzle-like pieces, and manipulation of toys to indicate correct answers. The verbal subtests require facility with words and printed material (reading or speaking). A description of each of the 10 subtests is provided in Rapid Reference 1.3.

Psychometric Properties

Extensive studies of reliability, validity, and fairness were conducted as part of the SB5 standardization. The SB5's main technical features are briefly outlined here. The normative sample for the SB5 included 4,800 subjects aged 2 to 96 years. The highest age grouping employed in the norm tables was 85+. The composition of the normative sample closely approximated the stratification percentages reported by the U.S. Census Bureau (2001). Stratification variables included gender, geographic region, ethnicity (African-, Asian-, and Anglo/Caucasian-American; Hispanic; Native American; and Other), and socioeconomic level (years of education completed, or parent's educational attainment). Additionally, subjects were tested (N = 1,365) from officially documented special groups such as individuals with mental retardation, learning disabilities, attention deficit, and speech or hearing impairments.

Internal-consistency reliability ranged from .95 to .98 for IQ scores and from .90 to .92 for the five Factor Index scores. For the 10 subtests, average reliabilities (across age groups) ranged from .84 to .89, providing a strong basis for profile interpretation (see Rapid Reference 1.4). Split-half reliability formulas were used for subtests and composite reliabilities for IQ and Factor Index scores.

Test-retest and interexaminer reliability studies were also conducted and showed the stability and consistency of SB5 scoring.

Evidence for content-, criterion-, and construct-related validity of the SB5 was detailed in Roid (2003f), including extensive studies of concurrent, predictive, and factorial validity. Also, good evidence of consequential validity and fairness of predicting achievement were reported in Roid (2003d). Examples of validity, including the correlations with other assessment batteries, are shown in Rapid Reference 1.5. The correlations shown are quite substantial and similar in magnitude to the concurrent correlations observed for other major intelligence batteries. The substantial predictive correlations between the SB5 and two major achievement batteries-the Woodcock-Johnson III (WJ III) and the Wechsler Individual Achievement Test-Second Edition (WIAT-II)-provide a strong basis for comparing intellectual and achievement scores of individuals.

Extensive studies of the factor structure of the SB5 were conducted, including confirmatory factor analyses using LISREL 8.3 (Joreskog & Sorbom, 1999). The factor analyses were calculated for five successive age groups (2-5, 6-10, 11-16, 17-50, and 51+) comparing factor models with one, two, three, four, and five factors. Split-half scores (scores for odd- and even-numbered items in each of the 10 subtests) were employed to provide more stable estimates of each factor in the maximum-likelihood analyses. The five-factor models showed superior fit, including the non-normed fit index (NNFI), ranging from .89 to .93; the comparative fit index (CFI), ranging from .91 to .93; and the root mean square error of approximation (RMSEA), ranging from .076 to .088.

Continues...

---

Excerpted from Essentials of Stanford-Binet Intelligence Scales (SB5) Assessment by Gale H. Roid R. Andrew Barram Excerpted by permission.
All rights reserved. No part of this excerpt may be reproduced or reprinted without permission in writing from the publisher.
Excerpts are provided by Dial-A-Book Inc. solely for the personal use of visitors to this web site.

---