Cross-validation Group Research Articles

The WAIS Adaptability Scale and vocational behavior

A Short Form of the Wechsler Adult Intelligence Scale was developed by selecting subtests that were consistently related to work performance across three groups (89 disabled, 140 retarded, and 111 disturbed) work-shop clients. Four scales were so selected: Similarity, Comprehension, Digit Symbol, and Block Design. The sum of the scores on these four subtests was significantly related to Full Scale IQ's (construct validity) and to work-shop performance (concurrent validity). Scores on this “Adaptability” Scale was also significantly related to employability measures in two other cross-validation groups. In fact, in all five groups Adaptability scores yielded higher correlations with employability measures than Full Scale IQ's.

Meaningfulness (m̄) indices for 120 nouns for children aged nine years

In this experiment a table of meaningfulness (m) was derived for 120 nouns. Noble's production method was used. Data were taken from a randomly selected sample of 135 9-year-old children who were attending school in Victoria, B.C., Canada. Range of meaningfulness was 8.41-.41 with a median of 4.25 and a mean of 4.29. A sample of 204 randomly selected Ss attending school in an adjacent school district was used as a cross-validation group. A correlation coefficient (Pearson product-moment) of .92 was obtained between two measures of meaningfulness for each of 48 selected nouns.

Projective drawings of brain-injured subjects.

The House-Tree-Person drawings of 90 Ss were evaluated with respect to diagnosed brain injury. Seven objective items such as window drawn on house edge and relative size of door were selected and weighted to give a scale of 22 points. This scale then yielded a mean score of 6.0 for 30 brain-injured Ss, 2.8 for Ss with various forms of physical disability, and 1.6 for 30 University of Washington students. The differences among these means were significant beyond the .001 level. A mean score of 4.8, obtained for a cross-validation group of an additional 10 brain-injured Ss, was significantly higher than the means for the control groups but not significantly different from the mean for the first brain-injured group. The relationship of diagnosed brain injury to Circular Pencil Maze scores and WAIS subtest scores on Digit Span, Vocabulary, and Block Design, as well as MMPI T scores on the schizophrenic scale were also investigated. Three illustrative cases showed that quality of drawing and “intelligence” were not major factors in the H-T-P scores for brain-injury.

Do objective professional examinations involve reasoning?

An objective professional examination in medicine and three relatively “pure” tests of reasoning were administered to a key development group of 88 medical interns and a cross-validation group of 83 interns. Results suggest that professional examination items involving reasoning can be identified by a physician or by item-analysis against reasoning tests. Judgmentally identified items correlated .35 to .49 with general and deductive reasoning; items identified by item-analysis correlated on cross-validation, .37 to .46 with these reasoning factors. (Spearman-Brown estimates were in the .50-.70 range.)

The Validity and Reliability of the Human Trait Inventory Designed to Measure Under- and Over-Achievement

The problem of the study was to determine if edited and rescaled personality items extracted from previous studies would elicit differential responses from validation and cross-validation groups of underand over-achievers. All analyses were done separately for the sexes. A fur ther purpose of the study was to determine a) the va lidity of the cross-validated items in predicting grade point average and b) the effect of adding the scores based on the cross-validated items to an aptitude measure to predict gradepoint average. The cross-validation analy sis yielded 32 male and 31 female items which correlated .42 for males and .36 for females with grade point aver age. Hoyt's analysis of variance reliability coefficients hovered around the .80 level. The items, when treated as a composite score, did make a significant contribution to increase in precision of predicting grade point average from an aptitude predictor alone. CURRENT RESEARCH emphasis on non-in tellectual factors in achievement is uncovering a positive relationship between certain per sonality variables and academic performance (Jensen, 1960; Krug, 1959; Middleton and Guthrie, 1949). Other research has also demon strated a relationship between personality and motivation (Mitchell, 1959). It is the purpose of this study to investigate the interaction of per sonality, achievement and motivation.1 That is, a research scale was 1) devised using personality items found by previous investigators to be related to academic achievement, 2) validated and cross validated on criterion groups of high and low motivated students, and 3) assessed for validity and reliability in estimating high school grade point average. Motivation Definition and Operationalization For the purposes of this research, academic motivation was broadly defined as a combination of forces which initiate, direct, and sustain be havior towards a scholarly goal. Historically, overand under-achievers have been assumed to represent extremes in academic drive. Therefore, it was decided that criterion groups of discrepant achievers would be used to test the motivational hypotheses. Methodology The methodology of the study evolved hypoth eses formulation, instrument construction, sample selection, and item and scale analysis.

The Relationship of Minnesota Teacher Attitude Inventory Scores to Certain Biographical Information

The purpose of this study was to determine if certain biographical data was associated with attitudes toward teacher-pupil relationship as measured by the MTAI. Data were collected on 532 female Ss near the beginning of their course work in education. These subjects were divided into groups scoring high and low on the MTAI for both and purposes. Biographical data were compared for differences in per centages between high and low groups. Some items were found which significantly differentiated between the two groups. IT IS A popular belief that if we had informa tion regarding an individual's past we could un derstand his present attitudes and behaviors. Several studies in recent years have demon strated the usefulness of biographical data in studying human behavior. It has been shown that certain biographical data are related to occupa tional choices of individuals (4). Ryans (3) found some relationship between childhood and adolescent activities of teacher with their later teaching behaviors. Statement of Problem The purpose of this study was to investigate certain biographical data that might be associ ated with attitudes, as measured by the Minne sota Teacher Attitude Inventory (MTAI). Bio graphical data that are associated with attitudes of pupil-teacher relationships may give us clues regarding the development of attitudes postu lated to be important in the classroom situation. The problem may be stated in the following question: Can it be empirically demonstrated that there exist certain biographical patterns associated with differences in attitudes as meas ured by the MTAI? The problem stated in the null form is as follows : There is no difference between the pattern of responses made by a group that scores low (lower quartile) on the MTAI, and a group that scores high (upper quartile) on the MTAI to certain biographical data. The Subjects Data were collected from 476 female subjects who were registered in a psychological founda tions course (Human Development and Learn ing), which was their first professional course in education, and from 56 female subjects reg istered in a nonprofessional education course (Introduction to Teaching) at the University of South Florida (USF). Students at USF normally do not declare their specific major in education until after they have completed their first pro fessional education course. The standardization group (N = 166) con sisted of those who were registered in a psycho logical foundations course during the 1961-62 academic year and trimester I of the 1962-63 academic year, and whose scores fell in the upper or lower quartiles. The biographical data were complete for those who scored in the upper quar tile, but were missing for two subjects in the lower quartile. The cross-validation group (N = 86) con sisted of those who were registered in a psycho logical foundations course (N = 49) and an in troduction to education course (N = 37) during trimester II of 1962-63 academic year, and whose MTAI scores fell in the upper or lower quartiles of the distribution for trimester II, 1963. Bio graphical data was incomplete for two subjects whose MTAI scores were in the lower quartile, and for nine subjects whose MTAI scores were in the upper quartile. A constant of 100 was added to all of the MTAI scores to avoid negative numbers. The means and standard deviations for MTAI scores of the and cross-validation groups is presented in Table 1. The reason for the unequal number in the upper and lower quartiles was the failure to col lect biographical data from all individuals whose scores fell either in the upper or lower quartiles. These may be thought of as nonrespondents. This content downloaded from 207.46.13.86 on Sat, 15 Oct 2016 04:13:08 UTC All use subject to http://about.jstor.org/terms

The lie scale of the Maudsley personality inventory

Summary 1. The Maudsley Personality Inventory, in a form containing 20 items designed to test “lying”, was administered to an experimental population of 99 Ss. 2. An item analysis of the Lie items was carried out and an 18 item Lie Scale with an average item/scale r of .59, and split 1/2 r of 66, was isolated. 3. It was found that “Lying” (L) had no significant correlation with Extraversion (E), but a negative and non-linear correlation with Neuroticism (N). 4. The distribution of L scores had a marked positive skew. 5. Between scale correlations were calculated for two cross-validation groups (a) 244 Maryland Students (b) 100 British apprentices. 6. For the two cross-validation groups, the shapes of the L score distributions were the same as for the experimental population. The relations between the scales were also found to be closely similar to those obtaining with the experimental population, except that the negative correlation between L and N was linear for group (b). 7. The implications of these findings are discussed.

Cross-validation groups, extreme groups, and the prediction of academic achievement.

Cross-validation groups, extreme groups, and the prediction of academic achievement.

High School Achievement and the Bell Adjustment Inventory

High school achievement, or the lack of it, has been related to various nonintellectual variables, including personality and psychopathology as measured by personality inventories. The personaliry instruments chosen have typically been the Bernreuter, the Bell, andthe MMPI. Two studies (Finch, 1932; Sappenfield, 1938) found no relationship between achievement and adjustment as measured by the Bermeuter inventory. Two studies with the Bell inventory were contradictory, with a positive report by Resnick (1951) and negative findings by Griffiths ( 1945 ). Gough (1949) was unable to predict achievement from selected MMPI items. The present study was designed to clarify the status of the Bell Adjustment lnverxory as a predictor of high school achievement. METHOD There were separate validation and cross-validation studies of Bell items predictive of high school achievemenr. The validation study included 470 male and female juniors and seniors from rwo high schools.' The students were selected from an equal number of high and low achievement sections in two required courses, History and Political Science. The cross-validation study contained 239 junior dass Ss selected in the same manner. All Ss were administered the Bell Adjustment Inventory. A 6-mo. interval (i.e., May to October, 1959) separated rhe validation and cross-validation testings. Grade-point-averages (GPA) were obtained for all Ss from grades during the immediately preceding semester. Distributions were then made in terms of GPA, where the letter grade A=l and F=5. Three validation groups were formed, Group 1 , the 100 Ss with the highest GPA; Group 2, the 50 cases on either side of the median GPA; and Group 3, those 100 Ss with the lowest GPA. Three cross-validation groups were formed in the same manner, with 50 Ss per group. Only juniors were used for the cross-