Educational and Psychological Measurement

A simple formula for calculating semipartial correlations is presented and illustrated.

MULTIPLE-choice test item responses may be influenced by a number of factors other than knowledge of content. Two factors, language difficulty and option set convergence, were experimentally manipulated and their effects on item difficulty assessed. Option convergence was found to have a significant effect on item difficulty while the effect of language difficulty was not significant. Distracter selection patterns differed across levels of language difficulty for only 6% of the items. It was suggested that a model of task difficulty be developed through investigation of additional characteristics of multiple-choice items.

Over the past several years, numerous questions have arisen pertaining to response alternatives for Likert scaling. Specifically, both two-column and one-column Likert formats are commonly used in educational and psychological measurement. Which format, however, is to be preferred? Is one format superior to the other and under what restraints? This study makes a start toward clarifying these issues by analyzing the results of applying two different methods of Likert-scale construction (single-column and discrepancy-column formats). The findings indicate that the discrepancy format clearly provides stronger discrimination for purposes of measuring need than does the single-column approach.

A comparative inspection of the WAIS and WAIS-R IQ tables in their respective manuals indicated that there would be variable IQ discrepancies at different ages because of differential age adjustments on the norms of the two scales. The use of equivalent WAIS and WAIS-R sums of scaled scores derived from Wechsler's (1981) comparison of the two instruments to determine comparable IQs on the two tests at different ages demonstrated the predicted variable discrepancies. It was concluded that it would be fruitless to attempt to equate WAIS and WAIS-R IQs directly but that the process could be achieved by the equation of sums of scaled scores on the two instruments and the use of the tables in both test manuals to obtain the WAIS and WAIS-R IQs representing the same levels of ability at different ages.

The purpose of the study was to determine the realiability of the spanish translation of the Self Esteem Inventory (SEI) with a group of Puerto Rican students on the island and another on the mainland. The sample was composed of 296 high school students ages 15-18. The study concluded that the Spanish translation of the SEI was a reliable instrument in evaluating the personal judgment of worthiness for Puerto Rican students in the island and the mainland United States.

The presently viewed role and utility of measures of instructional sensitivity were summarized. A case was made that the rationale for the assessment of instructional sensitivity can be applied to all achievement tests and should not continue to be restricted to criterion-referenced mastery tests. Next, a line of research and theory concerning context dependence in tests containing interpretive exercises was outlined. The logic of this theory was integrated with the rationale of instructional sensitivity. Finally, practical interrelationships among content validity, reliability, instructional sensitivity, and related measures were discussed and illustrated by use of a varied sample of test items and data.

Narcissism has become a matter of increasing concern in recent years. In this paper it is referred to as selfism and construed as a problem solving generalized expectancy in the framework of Rotter's social learning theory. The development of a 28-item objective scale is described along with preliminary evidence for its validity.

The effects of some restraints not included in the classical assumptions of the F- and H-test (e.g., correlation of mean and sample size) were examined in a simulation design of 1000 samples per condition. Also simulated was a situation in which two assumptions were not met simultaneously. The major conclusions were: H was not an appropriate alternative for F with samples of N ? 20; in all cases of unequal variances combined with unequal sample sizes H should be applied; and neither H nor F should be applied if more than one assumption of either test is not met.

A number of statistical procedures are available for detecting bias in individual test items. A supplementary treatment is proposed which helps identify sources of bias affecting groups of items. Although applicable to other statistical procedures, this treatment is illustrated with the transformed item-difficulty method as applied to an evaluation of a test used to help select applicants to be admitted to an aviation training program.

After a review of the disadvantages of linear models for estimating the probability of academic success from previous school records and admission test results, we propose the use of a probit model. In contrast to a number of linear regression models which use academic results as a dependent variable, the probit model allows the retention of drop-out cases within the sample; in addition, the coefficients as well as the probability estimates obtained from the probit model by the maximum likelihood method have satisfactory asymptotic properties. The model is calibrated on admissions data for the Ecole des Hautes Etudes Commerciales in Montreal for the years 1979 and 1980. It is noted that both of the variables used in selection have a significant impact on the probability of success, and each variable seems to be associated with a specific aspect of the phenomenon. Furthermore, when the candidates are grouped into quintiles, the deviations of the estimated probabilities from actual success rates are extremely small.

The present study compares the robustness of two different one way fixed-effects analysis of covariance (ANCOVA) models with respect to the effects of unequal regression slopes. The purpose of this study is to investigate whether the model which uses a test statistic incorporating estimates of the separate slopes will be more robust than the conventional model which assumes the slopes are equal. A Monte Carlo simulation technique was employed to generate data under 64 different situations. Two treatment groups, five different sample sizes and twenty pairs of regression slopes were used. The number of replications in each simulation was 1827 to enable 0.95 confidence that each actual alpha value did not differ from the estimated alpha by more than .01. Both equal and unequal error variance were examined. A different random number seed was used for each of the 64 simulations. The results indicate that when the two standardized regression slopes differed by less than .4, both models were robust. When the difference exceeded .4 and the sample sizes were equal, the model which incorporated estimates of individual regression slopes was more robust than the conventional model which used a pooled within regression coefficient. When the difference between slopes exceeds .4 and unequal sample sizes were associated with unequal error variance, neither of the models were robust.

Dependent variables in psychological research often take the form of test profiles. A computer program is described which provides comprehensive information on profile differences among independent groups based upon profile centroid separation, shape, elevation, and scatter. Descriptive statistics, univariate and multivariate measures of strength of association, test of homogeneity of covariance matrices, and comparative mean plots also are provided.

This program computes multiple judge reliability levels under the following conditions: different sets of judges perform the ratings; the number of judges is a constant; and the scale of measurement is nominal.

When selection occurs on the basis of two or more predictors, multivariate restriction of range can reduce various parameters of a validation study. An SAS and a FORTRAN IV program are available that allow for correction of criterion standard deviation(s) and zero-order validities.

This paper describes a microcomputer program that extracts principal components from a correlation matrix and rotates them to a varimax solution. Eigenvalues, variances associated with each component, unrotated component loadings, and varimax rotated loadings are output by the program.

The TUKEY FORTRAN program was designed to facilitate pairwise comparisons among means following a significant F ratio in an analysis of variance. Intended for use on CMS, the program has been written to allow the user to obtain Tukey's HSD statistic by entering information regarding the design and the analysis of variance results. Two variations of Tukey's test, the Tukey-Kramer test and the Games-Howell test, are available as alternatives for designs in which the assumptions of the HSD test are not met.

This series of studies examined the effect of adhering to the universally endorsed item-writing practice of making the correct answer of a multiple-choice item grammatically consistent with the stem. Formats were designed to investigate whether the practice of providing relevant grammatical cues in itself created a sensitivity for item performance. ACT Assessment experimental social studies items were constructed to investigate grammatical compliance in two particular situations: plural-singular and vowel-consonant agreement. In no instance were grammatically inappropriate situations created. Results suggest some subject sensitivity to these cues. Rephrasing into neutral, but still precise and grammatically correct, language may eliminate this sensitivity.

The Structure of Intellect Learning Abilities (SOI-LA) tests are widely used in gifted screening and the diagnosis of learning difficulties. Although the Figural, Symbolic, and Semantic dimensions of the tests have been employed as a prescriptive tool for matching students to appropriate methods of reading instruction, the factorial validity of these dimensions has not been widely investigated. Using a normative sample of 2nd-grade pupils (N =364), this study relied upon multiple-group factor analysis to confirm the hypothesized dimensions. The dimensions were generally confirmed with the Figural factor being the most clearly defined.

A 35-40 minute battery of eight tests was constructed to measure the following four specific cognitive abilities in young, preschool children: verbal skill, memory, perceptual speed, and spatial ability. The battery was administered to 98 preschool children. A factor analysis of the intercorrelations of the eight tests revealed four interpretable factors each representing one of the targeted abilities.

Prior to this study, evidence for this particular organization of cognitive abilities in young children had not been reported in other factor analytic studies of existing preschool mental tests. An explanation for this fact might be that the earlier test batteries were not designed to measure particular cognitive abilities. On the other hand, the eight considered for this study had been designed expressly for that purpose.

Based on results of factor analyses, a shorter version of the test battery, consisting of four tests, was devised. This shorter version which can be administered in 20 minutes, shows promising validity relative to the Stanford-Binet Intelligence Scale.

In a longitudinal investigation with an initial sample of 179 seventh grade students (90 girls and 89 boys) on whom data were collected at the end of the first three semesters, at the termination of the next four semesters (N = 147), and at the conclusion of the following two semesters (N = 141), predictive validity coefficients of five subscales of the Dimensions of Self-Concept (DOSC), Form S were ascertained relative to total grade point average (GPA) and to the GPA earned in each of several academic subjects during each of three periods of enrollment. Although not showing quite so high validity coefficients as those realized for subtests from a standardized achievement test battery, four of the five DOSC subscales afforded promising predictive validities-especially the one entitled Level of Aspiration. Additional validity data for the DOSC in relation to the last two semesters of high school work (twelfth grade) will be reported in the near future.

At an upper-division university, students entering the Bachelor of Science in Nursing (BSN) program were graduated from Associate Degree in Nursing (ADN) programs or graduated from hospital-based (diploma) programs. The admissions criteria for both groups of students were identical with one exception. Diploma nurses were required to pass three ACT Proficiency Examination Program (PEP) tests with a standard score of at least 45. Thus, the hospital-based graduates, although awarded credit to increment their transfer hour credential, had also validated their prerequisite for some of the advanced nursing curriculum. Past research was confirmed by this inquiry; the ensuing academic success of the diploma nurses was significantly higher than that of their ADN peers. Regarding retention, the rate for hospital-based graduates was significantly higher than that for those graduated from junior and community colleges. Thus, the PEP credit-by-examination testing program to place diploma nurses into the BSN program at an upper division university was found to be successful by two criteria: academic success and retention.

The purpose of this paper was to report the findings of a study concerning (a) the relationship between academic achievement and student-faculty personality congruence in terms of field dependence and (b) the predictability of field dependence by age, sex, and cumulative grade point average (GPA). The level of field dependence of 386 students and their instructors was identified by the Group Embedded Figures Test. A formula was employed to describe the congruence of personality traits between students and faculty. The correlational analysis supported the belief that student's course grades and GPA were very slightly greater when there was similarity rather than dissimiliarity of personality type with that of their instructor. A multiple regression analysis indicated that the cumulative GPA and sex as predictors contributed a statistically significant amount of variance to the field dependence/independence measure but that age did not.

The Mathematics Attribution Scale-Algebra (MAS) was designed to assess attributions to ability, effort, task, and environment for success and failure in algebra. The present study examined the reliability and-validity of the MAS with high ability high school students. The results of this study present a complicated picture of the psychometric properties of the MAS.

The Mathematics Attribution Scale (MAS) (Algebra) was designed to assess attributions of success and failure in algebra to ability, effort, task, and environment. The present study examined the MAS (Algebra) for a separate dimension of attributions for success and a dimension of attributions for failure. The two hypothesized dimensions did not emerge.

The 1969 revision of the Strong Vocational Interest Blank (SVIB) introducted to that instrument the concept of homogeneously keyed scales presenting 23 Basic Interest scales, and the 1974 revision, the Strong-Campbell Interest Inventory (SCII) integrated the six Holland types into the scoring in terms of six Theme scales (Campbell and Hansen, 1981)-29 scales in all. These changes acknowledging the theoretical structures of occupational interests make comparisons between the SCII constructs and other homogeneously keyed instruments more meaningful. The purpose of the present study was to provide construct validity data by evaluating the relationship between results obtained from the college level form of the Carrer Occupational Preference System, Interest Inventory, Form P (COPS-P) (Knapp-Lee, Knapp, and Knapp, 1983) and those of the Theme and Basic Interest scales of the SCII. Correlations were computed between the COPS-P scales and both the SCII Theme and Basic Interest scales for a sample of 122 college students. The magnitude of correlations between conceptually similar COPS-P and SCII Theme scales ranged between .29 and .72 with the median being .60. In virtually every instance the COPS-P scale was more highly related to the theoretically consistent SCII Theme scale than to any other Theme scale in the array. In a comparison of the COPS-P scales with the SCII Basic Interest Scales, an examination of relationships between those scales which have identical, or nearly identical, names did show a range in magnitude of .37 to .81 with a median of .68. The highest correlation was between the COPS-P Communication, Written scale and the SCII Basic Interest scale, Writing.

The pattern of correlations between the COPS-P scales and the SCII Theme scales provided support for the theoretical constructs measured by the two instruments.

The purpose of this study was to evaluate the convergent and discriminant validity of counselor and client ratings of common presenting problems at a counseling center for unviersity employees. Each subject was asked to complete the Employee Problem Appraisal Form (EPAF) at three points in time: during the intake interview (N = 306); after eight weeks of counseling (N = 114); and at termination after 16 or more counseling sessions (N = 80). The counselor for each client also completed a parallel form of the EPAF at each time interval. Five scales on the EPAF were selected for multitrait-multimethod analysis on the basis of internal-consistency estimates. Non-zero convergent validity values (correlations between independent measures of the same trait) ranged from .27 to .66. That four of these non-zero convergent validity values were higher than correlations between independent measures of different traits (heterotrait-heteromethod values) provided some evidence of discriminant validity. Elevations of correlations between repeated measures of different traits (heterotrait-monomethod values) were indicative of shared method variance in both counselor and client ratings. Although non-zero convergent validity values in this study were not large, they increased over time and compared favorably with measures of similar psychological variables based on independent ratings of the same trait (Campbell and Fiske, 1959). These results were discussed relative to improvement of the EPAF, and suggestions were offered for further research using this instrument.

In an examination of what employees considered to be characteristic of excellent managers, it was found that trust and respect dominated all other categories of managerial behavior. An instrument to assess trust and respect was constructed by using the Rasch model. In an examination of item difficulty estimates, it was found that the item content could be described in terms of a developmental model of supervisor-subordinate relationships. The present paper describes the construction and preliminary validation process for the instrument and offers a probabilistic hierarchical developmental model.