“E535 ’51-'93 ““i-‘y- 5-1“. ~ .H.\ .a' ~.-;;-; 3‘ 3 1293 104507 rmmflm ** "" " Mi illlWllHHll!“IUHHWINIHIWlHIHLWIlW This is to certify that the thesis entitled SPATIAL ABILITY: RELATIONS WITH GENDER, SPATIAL COGNITION AND ENGLISH AND MATH ACHIEVEMENT presented by Jane Leslie Pearson has been accepted towards fulfillment of the requirements for Master of Arts (“geekl Psychology (AMOS? {ix/Aim ijor professor Date AX q L‘q {2/ D 7 MS U is an Affirmative Action/Equal Opportunity Institution MSU LIBRARIES “ BETURNING MATERIALS: Place in book drop to remove this checkout from your record. FINES wili be charged if book is returned after the date stamped below. .1 A ‘ a) «r i' ‘l. 050031399 SPATIAL ABILITY: RELATIONS WITH GENDER, SPATIAL COGNITION AND ENGLISH AND MATH ACHIEVEMENT by Jane Leslie Pearson A MASTERS THESIS Submitted to Michigan State University in partial fulfillment of the requirements for the degree of MASTER OF ARTS Department of Psychology 1982 ABSTRACT SPATIAL ABILITY: RELATIONS WITH GENDER, SPATIAL COGNITION AND ENGLISH AND MATH ACHIEVEMENT By Jane Leslie Pearson This study investigated the relation between measures of spatial ability-~Embedded Figures Test, Building Memory Test, Mental Rotation Test (MRT), Differential Aptitude Spatial Relations Subtest (DAT)--and spatial cognition--Landmark Potential, Map Drawing, Water-level Task (H20)--and the influence of scholastic achievement (ACT) and gender on these tasks. Subjects were 353 undergraduates--199 women and 154 men. Initial analyses indicated gender differences on ACTS and rela- tions among ACTS and spatial tasks, necessitating the controlling of ACTS in subsequent analyses. Results showed that one Spatial ability measure (DAT) correlated with three of the four Spatial cognition tasks. When ACT Math was controlled, gender differences appeared only on the H20 and MRT tasks. No gender differences appeared on environ- mental spatial tasks. It was concluded that 1)"traditional" measures of Spatial ability do not measure Spatial (environmental) cognition; and 2) gender/task relations are differentially affected by Math and English achievement. ACKNOWLEDGEMENTS My foremost thanks go to Dr. Lucy Ferguson for supporting my many inquires in Spatial ability, and for her patience and editorial help in my efforts to complete this manuscript. My other committee members, Dr. Ellen Strommen and Dr. Lauren Harris, also provided me valuable guidance in my "Spatial explorations". I also want to thank my data collectors and coderS--Mark, Mark, Dave, Susan, Susan, Michelle, Renee, Chaz and Howard. A Sincere thank to Sarah Whiteher who saved my sanity by helping me with the Hewlett- Packard computer is also in order. Finally, I want to express my appreciation to my parents, to Kyle, and to Nick, for their constant support that makes this type of effort possible. 11 LIST OF LIST OF LIST OF Chapter TABLE OF CONTENTS TABLES. O O O O O O O O O O O O O FIGURES O C O O O O O O O O O O O APPENDICES. O O O O O O O O O O O I. INTRODUCTION. . . . . . . . . . . . Purpose 0 O O O O I O O O O I I O O I. MEASURES AND RATIONALE. . . . . . . Embedded Figures Test . . . . . . Building Memory Test. . . . . . . Mental Rotations Test . . . . Differential Aptitude Spatial Relations Subtest . . . . . . . Water-level Task. . . . . . . . . Landmark Potential, Map Drawing, and Landmark Location Tasks . . ACT Subtest and Composite Scores. II. METHOD sample. C O O I O O O O O O O O O 0 Procedures. . . . . . . . . . . . . Instruments . . . . . . . Landmark Potential an Map Drawing Tasks . . . . . . . . . . Landmark Location Task. . . . . . Scoring and Reliability . . . . . . III 0 RESULTS 0 O O O O O I O O O O O O 0 Mean Scores and Gender. . . . . . . Correlations Among Spatial Measures Factor Analysis on Spatial Measures iii Page vii viii 1O 11 11 12 12 18 19 19 19 21 24 26 27 27 27 3O Men and Women's Correlations. . . . . . . . Separate Factor Analyses on Spatial Measures. . . . . . . . . . . . . . . . . Comparison of Men and Women's Correlations. Factor Analysis of Spatial Measures and ACT Scores. . . . . . . . . . . . . . ACT Scores as Covariates. . . . . . . . . . Partialled Comparisons of Men and Women's Correlations. . . . . . . . . . . Point-biserial Comparisons of Partialled and Original Scores . . . . . . . . . . . Factor Analysis on Partialled Scores. . . . Summary . . . . . . . . . . . . . . . . . . IV. DISCUSSION Relations Between Spatial Thought and Spatial COgnition . . . . . . . . . . Internal Consistency. . . . . . . . . . Specific Predictions. . . . . . . . . . Sex-related Differences . . . . . . . . . ACT and Sex-related Differences . . . . . Math and English Achievement. . . . . . Conclusion. . . . . . . . . . . . . . . . Summary . . . . . . . . . . . . . . . . REFERENCES NOTES . . . . . . . . . . . . . . . . LIST OF REFERENCES . . . . . . . . . . . . . . . APPENDICES O O O O O O O O O O O I O O O O O O 0 iv 30 33 33 35 38 38 39 39 42 43 45 46 46 48 49 5O 53 54 55 59 TABLE 10 11 12 13 14 LIST OF TABLES Instruments and available reliabilities . . . . . . Means, standard deviations, and t-tests on Spatial tasks for men and women . . . . . . . . . . Means, standard deviations, t-tests and point- biserial correlations on ACT standard scores for men and "om en 0 O O O O O O O O O O O O O O O 0 Correlation coefficients for total sample on Spatial taSkS O O O I I I O O O O O O O O O O O O 0 Factor analysis on spatial tasks . . . . . . . . . Pearson correlation coefficients for men and women on Spatial tasks. . . . . . . . . . . . . . . . . . Comparisons of men and women's correlations using FiSher's Z O O O O O O O O O O O O O O O O O O O 0 Factor analysis on Spatial tasks and act Scores . . Correlations between Spatial tasks and ACT scores for men and women . . . . . . . . . . . . . . . . . Partial correlations on Spatial tasks: Controlling for ACT Composite . . . . . . . . . . . Partial correlations on spatial tasks: Controlling for ACT English . . . . . . . . . . . . Partial correlations on Spatial tasks: Controlling for ACT Math. . . . . . . . . . . . . . Partial correlations on spatial tasks: Controlling for ACT Social Studies. . . . . . . . . Partial correlations on spatial tasks: Controlling for ACT Natural Science . . . . . . . . 20 28 29 29 31 32 34 36 37 59 59 6O 6O 61 15 16 17 18 19 20 21 Comparisons of men and women's correlations using Fisher's Z controlling for ACT Composite. Comparisons of men and women's correlations using Fischer's Z: Controlling for ACT English. Comparisons of men and women's correlations using Fischer's Z: Controlling for ACT Math. . Comparisons of men and women's correlations using Fischer's Z: Controlling for ACT SOCial Studies. 0 O I O O O O O O O O O I O O 0 Comparisons of men and women's correlations using Fischer's Z: Controlling for ACT Natural Science . . . . . . . . . . . . . . . . Comparisons of point-biserial correlations on Spatial tasks controlling for ACT scores with original point-biserial correlation . . . . . . ACT Math or English partialled factors for men and women . . . . . . . . . . . . . . . . . vi 62 62 63 63 64 39 41 FIGURE I II LIST OF FIGURES Route indicating location of high (H) and low (L) landmark scenes . . . . . . . . . . . . . . . . City-block map used for map drawing. . . . . . . vii 22 23 LIST OF APPENDICES APPENDIX A Partial correlations on Spatial tasks. . . . . . . 59 B Comparisons of men and women's correlations. using Fischer's Z. . . . . . . . . . . . . . . . 62 viii INTRODUCTION Research exploring ability measurement (Cronbach, 1970) and gender differences in intellect and ability (Shields, 1975) has been reported Since the latter half of the 19th century. More recently, Maccoby and Jacklin (1974) reviewed over 2,000 books and articles on gender dif- ferences in intellectual ability, motivation, and social behavior, and concluded that three cognitive gender differences appear stable; com- pared to boys, girls excell in verbal skills, and boys have better math and visual-Spatial ability than girls. Of interest to this study is the gender-related difference found in several Spatial ability measures, and also the construct of spatial ability itself. Since Maccoby and Jacklin (1974) conducted their massive literature review, several studies (e.g. Sherman, 1978; Hyde, 1981) have re-reviewed the evidence on cognitive gender differences and have pointed out that even for these supposedly "well-established" differences, the magnitude of gender differences is very small. Furthermore, the concept of "Spa- tial ability" has broadened Since Maccoby and Jacklin's review. New methodologies from areas such as geography, sociology, architecture and urban planning, are now found in the psychological literature, investi- gating Spatial representation and Spatial behavior in the environment. With the wide variety of disciplines exploring Spatial ability a broad Spectrum of spatial behaviors and representations have been investigated. What has resulted is some confusion as to what "Spatial ability" actually means. Liben (1981, Chap. 1) in a discussion of Spatial representation and behavior, identifies three "types" of Spatial repre- sentation--Spatial products, Spatial thought, and Spatial storage--and two "contents" of Spatial representation--Specific and abstract. Under the rubric of Spatial representation contents, knowledge about specific Spaces that enables an individual to maneuver in an environment is called "environmental cognition". Alternatively, investigators are in- terested in the domain of Spatial abstraction. Piaget's work on projec- tive and Euclidean understanding of Space would be one example of Spa- tial abstractions (see Piaget and Inhelder, 1967). AS for types of Spatial representation, Spatial products encompass any sort of external representation, regardless of medium, such as Sketch maps, models, or verbal accounts. Siegel (1981) argues that al- though a distinction exists between knowledge and communication of Spa- tial ability (or cognitive mapping), we can only study external repre- sentations directly; internal knowledge can only be inferred. The second type of Spatial representation--Spatial storage-refers to any information mentally stored (as in neurophysiological structures), where the individual is not cognizant of the information. Truth pro- positions, stimulus-response bonds, and animals' efficient movement through space, illustrate this concept. What is crucial to spatial storage is that it is not cognizant; once the individual reflects upon this information, it becomes Spatial thought. Therefore, spatial stor- age is often inferred on the basis of activity or way finding, and Spa- tial thought requires additional evidence (e.g., sketch maps, verbal descriptions). Spatial thought—-Liben's third category of spatial representation—is "knowledge that individuals have access to, can reflect upon, or can manipulate, as in Spatial problem Solving or Spatial imagery" (Liben, 1981, p. 12). Many of the standardized spatial ability measures, such as the mental paper folding task of the Differential Aptitude Spatial Relations subtest (Bennett, Seashore & Wesmann, 1974), and the Shepard and Metzler (1971) Mental Rotation Task, tap spatial thought. In regard to gender differences in spatial behavior and representa- tion, it is in the "Spatial thought" area of Spatial ability-often called visual-Spatial ability--where gender differences have been most often noted, with males excelling. However, sex-related differences in large-scale environment Skills, such as "sense of direction" accuracy, pattern-walking and geographical knowledge have also been found (see Harris, 1981, Chap.4). Thus is appears that gender differences may be found in many types of spatial representation Skills and behavior. Purpose Although Liben's categories are very straitforward, the extent to which the types of Spatial behavior Share common Skills is generally unknown. Another issue, as in much psychological research today, is the question of the generalizability of laboratory findings to field "facts". If Spatial products can be considered environmentally valid measures of spatial behavior, the degree to which standardized spatial thought measures relate to Spatial products could be considered a test of environmental validity for spatial thought measures. Such questions are theoretically and practically significant when attempting to assess and explain the gender differences found in Spatial skills. Therefore, it is the purpose of this study to examine the relation among measures of Spatial "thought" (i.e., paper and pencil spatial assessment tests) and Spatial "products" from a Slide-Simulated walk through an environment. Of equal interest is the degree of gender dif- ferences found in both types of Spatial tasks (spatial thought and spatial products), and the possible differences between separate factor structures for men and women. Furthermore, the contribution of mathematical, verbal and overall scholastic achievement to Spatial task performance is crucial to discern spatial ability from "general" ability (Fennema & Sherman, 1977). Therefore a standardized measure of Scholastic achievement is also included in the correlational analyses of the study. CHAPTER I This chapter describes the measures selected for the study, and the rationale for their selection. The first four measures to be compared--the Embedded Figures Test '(Witkin, Oltman, Raskin & Karp, 1971), a revised version of the Educa- tional Testing Service Building Memory Test (Note 1), a paper and pencil version of the Mental Rotations Test (Vandenberg & Kuse, 1978) and the Spatial subtest of the Differential Aptitude Test (Bennett, Seashore & Wesman, 1974)—-can be considered "traditional" paper and pencil mea- sures, each purporting to tap some aspect of visual-spatial ability. These measures have evolved from psychological testing, and are often used a assessment tools. The fifth measure to be compared is a paper and pencil, multiple- choice version of the Piagetian water-line task (Harris, Hanley & Best, 1977). Several studies (e.g. Liben, 1978; Ray, Georgiou & Ravizza, 1979) have found Significant relations between the water-level task and Spatial ability tests. The remaining measures to be compared--landmark potential, map drawing, and landmark location tasks, come from geography and urban planning literature. As was noted previously, from this more practical perspective, Spatial ability is interpreted as Spatial knowledge, or cognition, and related spatial behavior. (One could argue that the water-level task is also a measure of spatial knowledge. That is, one must be familiar with the "behavior" of water in a tilted container to succeed in the task.) In this type of spatial research, investigations attempt to examine actual or Simulated spatial behavior by means of Spa- tial "products" such as map reading, map drawing and landmark potential estimations. The remainder of this chapter reviews the Spatial measures to be used. For each test, whatever available information concerning develop- mental differences in general, differential abilities (i.e., mathemati- cal and verbal abilities), and sex-related differences in Spatial cogni- tion or spatial ability that the measures have revealed, is presented. Embedded Figurengest The Embedded Figures Test (EFT) has been chosen as one of the spa— tial assessment measures to be examined for several reasons. According to Witkin et al. (1971), the EFT is a differentiator of cognitive style. The EFT requires that an individual locate a previously seen simple figure within a larger, complex figure, which has been so organized as to obscure or embed the Sought-after target figure. Those adept at such disembedding are considered to be "field independent", while those in- dividuals less capable of quickly discerning the figure from the field are said to have a "field dependent" cognitive style. In the strictest interpretation, the EFT measures competence at perceptual disembedding. It seems plausible that determining the location of an object in an en- vironment as a reference point, such as landmark, would require Similar disembedding Skills; those individuals who are more field independent (FI) would perform better at determining landmark potential. Both re- quire an individual to separate an element from its background. A second reason for selecting the EFT measure is that it has re- vealed sex-related differences in past research (Witkin et al., 1971; Maccoby and Jacklin, 1974), with males being more field independent that females. However, significant sex differences occur with less consis- tency in non-Western studies (Witkin & Berry, 1975). Where significant sex differences do appear, they are mainly from sedentary, agricultural communities, as opposed to migratory cultures. This may suggest that the sex-related differences are due in part to experiential differences in role assignments for males and females. In migratory, less strati- fied societies, females are more independent, relative to sendentary societies where women are often assigned the roles of home and child care. Witkin addresses the sex-related differences controversy found among field-dependence-independence and spatial abilities studies (Witkin & Berry, 1975). He states that . . . Confusion has been introduced into the sex-differences literature by the inappropriate lumping of "perceptual dis- embedding" (field-dependence-independence) with "spatial abilities". Actually, they are discrete dimensions; in fac- tor analytic studies, tests of these dimensions have repeatedly been found to load different first-order factors. The existence and causes of sex differences must therefore be specifically and separately identified for each dimension (note 9, p. 75). As will be seen below, however, spatial ability is not always a "separate dimension" from perceptual disembedding. The relation between the EFT and spatial ability brings up a third reason for examining the EFT. Depending on the type of Spatial ability measure being used, factor analytic studies have reported conflicting findings. Many of the earlier factor analysis studies compared the EFT to intelligence subtests that are more spatially (i.e., not verbally) oriented. Studies by Cohen (1957, 1959), Goodenough and Karp (1961), and Karp (1963) report factor loadings of the EFT on three factors of the Wechsler intelligence scales. The three factors of the total IQ have been identified as analytic (Block Design, Object Assembly and Pic- ture Completion subtests), verbal-comprehension (Vocabulary, Informa- tion, and Comprehension subtests), and concentration (Digit Span, Arith- metic Digit Symbol subtests). The four previously named reports sug- gested that EFT scores correlate Significantly with analytic subtests, and correlate to a low degree with both verbal-comprehension and atten- tion concentration. Similarly, Coates (1975) found that for four-year- olds, the Preschool Embedded Figures Test (PEFT) loaded on a common fac- tor Shared by the Block Design and Geometric Design subtests of the Pri- mary Scale of Intelligence (WPPSI). Although moderate correlations may be found between the EFT and full scale IQS, it would seem that this correlation would be attributable to the strength of the analytic factor subtests as only one of three possible factorial contributors. Thus, the EFT ought to be a valid measure in determining analytic abilities. Reports contrary to such factor-loading findings come from studies of sex-related differences in spatial and verbal abilites. In their review of sex differences, Maccoby and Jacklin (1974) and Wittig and Petersen (1979, Chap. 1) have described spatial ability as having a visual nonanalytic component, and a visual analytic component. Non- analytic Spatial visualization involves visually rotating an object without verbal mediation (e.g. mental rotation tasks). Because of the time limitations on these tests, Wittig and Petersen theorize that a verbal mediation approach is not likely, assuming that it requires more time to solve the tasks and is thus a less efficient strategy than a nonverbal, rotational approach. Gildemeister and Friedman (1978) pur- sued the question of the relation between verbal and spatial ability, investigating gender differences in performance that correlated with verbal ability. They compared first graders from the high and low quar— tiles in verbal ability (measured by the WISC-R verbal subtests) on several tests that included the Children's Embedded Figures Test (CEFT), Spatial visual-analysis tests (as measured by Similarities and dif- ferences in photos of faces). In general, the intercorrelations among tests differed with verbal ability. The authors suggested that verbal ability needs to be taken into account when inferences about style dif- ferences are made. In the inter-test correlations, field independence was consistently related to Spatial—visual ability. Focussing more on Spatial test relations, Wolf (Note 2) found that sex differences disappeared when the EFT and Primary Mental Abilities (PMA) Space measures were compared; the gender difference on the EFT was accounted for by the PMA Space measure. In other studies of sex-related differences, Hyde, Geiringer and Yen (1975) found the Identical Blocks test accounting for sex differences on the Rod and Frame Test (the ini- tial measure of F1 deve10ped by Witkin) and on a mental arithmetic test, but not on the EFT. They did find that removing the effects of Spatial ability (Identical Blocks Test) produced Significant sex differences in favor of females, with these latter differences being accounted for by 1O vocabulary proficiency on the WAIS subtest. It appears that these findings, like Witkin's, provide even more reason for concern when making inferences about sex differences in perceptual disembedding and Spatial abilities. Recalling that Gildemeiester and Friedman noted that differences in verbal ability affected performance in other tests, what seemed to dif- ferentiate verbal abilities was, not surprisingly, the ages of the children. Most of the children in the top quartile of their study were older, and age appeared to be significantly related to performance on Spatial visualization tasks. Deve10pmental differences also have been investigated in the EFT. Witkin, Lewis, Hertzmer, Machover, Meissner and Wapner (1954) reported that PI increased sharply between the ages of 10 and 13, increased Slightly between 13 and 17 years, and then reached asymptote by 17 years. In another study of F1, Witkin, Goodenough and Karp (1967) again found that FI increased with age, and also that individuals maintained their relative position along the FI-FD dimension with increasing age. That is, an individual's field dependence relative to peers at one age stayed the same relative to his peers at a later age (Goldstein & Blackman, 1978). Building Memory Test The revised ETS Building Memory Test (BMT) has been included because of its correlation with performance on map drawing and perspective taking tasks, and with the ability to reconstruct a model urban environ- ment after a video simulation of a trip through the environment (Note 11 1). The BMT requires the individual to memorize the locations of several buildings on a map, and then to locate the positions of each building on an outline map. At this time, I have not been able to find any reports of sex-related or age differences on this task. Performance on the BMT ought to be related to the Simulated-walk Slide tasks (de- scribed below) because of the Similarity between the slide and video Simulations. Mental Rotations Test The Mental Rotations Test was included to test Moore's (1979) sug- gestion that rotation is a Specific cognitive ability in cognitive map~ ping. If so, then performance on this test ought to be related to per- formance on the slide Simulation tasks. The Mental Rotations Test (MRT) consists of a comparison of complex cube configurations presented in different orientations to determine whether they are the same or dif— ferent. Also, the MRT is considered to be a nonanalytic measure of spa- tial ability aS mentioned earlier. If there is such a division of Spa- tial Skills of analytic and nonanalytic components, it Should be re- vealed in a Significant correlation between performance on the MRT and the spatial subtest from the Differential Abilities Test (nonanalytic) and low correlations with the EFT and water-level tasks (analytic). Differential Aptitude Spatial Relations Subtest Perfbrmance on the DAT Spatial subtest is of interest when comparing the mental rotation aspect of Spatial ability with other spatial tasks that may appear more disembedding (or analytic) bound. The subtest con- sists of a type of mental rotation in which the subject has to decide 12 which of four drawings of three-dimensional figures matches a standard figure. The alternative drawings appear in various orientations, or various Shaded codes, and must be mentally manipulated and rotated to determine the match. As in many of the spatial relations tests, gender differences are reported in the DAT Spatial test, with males excelling (Harris & Wagner, Note 4; Hartlage, 1970; Flanagan, Dailey, Shaycroft, Gorham, Orr, Gold- berg & Neyman, 1961, Note 4). Flanagan et al. reported that high school boys' average scores exceeded the girlS' scores by at least .40 standard score units. Wgter-level Task The water-level task is included to investigate its purported rela- tion to general spatial ability, its demonstrated gender difference in performance, and its posited relation to the EFT. In the classical Pia- getian test of conservation, the child is shown a bottle half-filled with water, asked to note the position of the water in the bottle, and then asked to predict where the water-level will be when the bottle is tipped. Presumably, 12-year-olds would succeed at this task. However, many 12-year-olds, and many adults--most frequently women--do not demon- strate knowledge of the principle (Harris, Hanley & Best, 1977). Gender differences also have been reported in the degree of correlation between the water-level task and Spatial ability tests. For example, Ray, Georgiou and Ravizza (1979) found that the correlation with performance on the Minnesota Paper Form Board Test was Significantly higher for col- lege men than for college women. 13 Although Piaget and Inhelder developed the water-level task to tap underlying coordinate systems (spatial abstraction), this task also taps environmental cognition. That iS, to accurately complete the task, one must have knowledge of the physical principle governing the position of water in a tilted container. Therefore, its relation to the spatial cognition tasks (to be described) will also be of interest. A significant relation between the EFT and the water-level task has been reported by Liben (1978). Liben hypothesized that both tasks in- volved disembedding skills. She also found sex differences in degree of relation in performance on the EFT and water-level task, with 12th-grade boys' scores correlating to a greater degree on the two measures than 12th-grade girls' scores. Thus, the inclusion of the water-level task would provide additional data on the relation between the EFT and water-level tasks, and possible support for the sex-related difference in degree of relation between the measures. The findings of this study may also extend the disembedding theory to relations between the EFT, water-level task, and the present study's landmark potential test; it may be that all three measures involve a disembedding Skill. This hy- pothesis would be tested by the degree of intercorrelations between the three tasks. Landmark Potential, Map Drawing and Landmark Location Tasks Gender differences in performance and intercorrelations with other tests will be examined on the Landmark Potential, Map Drawing and Land- mark Location tasks. To provide a context for the consideration of the landmark potential task, some recent Piagetian theories of environmental 14 knowledge will be briefly reviewed. Based on Piaget's research on children's abilities to represent Space, Hart and Moore (1973) hypothe- sized a three stage system of the development of references: In the first stage, the child progresses from a coding system based on the re- lation of objects to his own body (egocentric frame of reference), to a coding system based on axes that facilitate full coordination of Space (coordinated frame of reference). Siegel and White (1975) have proposed a Similar model also based on Piagetian ideas but geared Specifically for large Scale Space. In this model, landmarks and routes are essen- tial elements of cognitive maps. Sequentially, landmarks are first noted and remembered with routes later acting in context of the land- marks as links between them. Like Hart and Moore's theory, configura- tions of routes are the final stage, becoming integrated into an overall framework. Siegel and White (1975) define landmarks as . . . unique patterns of perceptual events [configurations] at Specific locations, they are predominantly visual for human adults, they are strategic foci to and from which one travels, and they are used as proximate or intermediate course-maintaining devices (p. 23). Studies investigating the development of large-scale environmental re- presentation have used frames of reference (landmarks) and route mapping as reasonable foci. Because the study of landmark potential and route mapping is relatively new, at least to most develOpmental psychologists, an extensive description of several studies is included. Several studies have attempted to assess adults' route knowledge by 15 the use of photographic Simulation of environmental routes. Using a series of 50 color Slides, Allen Siegel and Rosinski (1978) tested col- lege undergraduates' knowledge of distance relationships among various landmarks in a Simulated walk through a commercial-residential area. An analysis was conducted on the estimates of distance between a reference scene and each of 25 target scenes, using as a measure correlations of log-estimates to log-actual distances. A second type of analysis was used to assess the interrelationships among the distances between the environmental targets. Nonmetric multidimensional scaling (MDS) was used for such analysis (Subkoviak, 1975), allowing for the reduction of complex matrices to a simple picture portraying spatial interrelation- ships among objects. Results indicated that the subjects performed Sig- nificantly more accurately after a second viewing of the Slides. Sex-related differences in cognitive mapping skills have been re- ported. Bettis (cited by Harris, 1981, p. 87) found that fifth-grade boys out-performed girls Significantly on knowledge of geographic facts about the state of Michigan. The work of Siegel and Schadler (1977) with four- and six-year—olds indicated that boys were more accurate than girls in constructing a model of their classroom. In larger model studies, Herman and Siegel (1978) again found boys were more accurate in reconstructing a model town. Of greater interest to this study is a report by Herman, Kail and Siegel (1979) of gender differences in col- lege freshmen's landmark knowledge. In assessing the spatial knowledge of their campus, it was found that men's knowledge of landmarks signif- icantly exceeded that of women in recall, recognition and accuracy in naming buildings. Measures of knowledge of routes or configurations of 16 the campus did not reveal sex-related differences, however. Age differences in landmarks and route mapping also have been studied. Allen, Kirasic, Siegel and Herman (1979) asked second-graders, fifth-graders and college students to select and use environmental land- marks. In one experiment, the subjects viewed the same 50 slide presen- tation cited above and selected high potential landmarks. In a second exPeriment, a new group of children and adults from the same age levels were chosen as subjects to view Slide selections of their peers, and those scenes selected by others not of their age group as well. The re- sults indicated that adults and children did not always select the same high potential landmark, and that children are less capable than adults in judging landmarks as distance cues. The researchers suggest that the age differences are amenable to either a cognitive or perceptual inter- pretation. In the cognitive interpretation, intersections, which by na- ture represent possible changes in heading, may be used to demarcate cog- nitive units of distance. Allen et al. (1979) posit that practical ex- perience indicates that urban distances are frequently communicated in terms of number of blocks. The developmental increase in selecting scenes with potential landmark value may thus reflect an increasing use of a culturally determined metric. The researchers suggest that a per- ceptual explanation for the trend is equally plausible. Changes in heading (which alter the focus of the visual flow field) and movement through intersections (which transforms high contrast regions in the visual field) are both rich sources of perceptual information (Gibson, 1969). Given the fact that children are not as adept as adults at per- ceptual differentiation (Gibson, Gibson, Pick & Osser, 1962), it may be 17 that the deve10pmental pattern of Scene selection reflects an increase in the ability to discern perceptual cues. Another developmental study accentuates two factors involved in de- termining landmark potential. Acredolo, Pick and Olsen (1975) asked to what degree two environmental variableS-—familiarity and differentiation (landmarkS)--affected the recall ability of three-, four- and eight- year-old children in locating a Spatial event. They hypothesized that in the absence of understanding projective and Euclidean Spatial rela- tions, the preoperational three- and four-year-olds' performance would become partly dependent on differentiation, or the presence of land- marks. Their results confirmed their hypothesis; memory for location was Significantly more accurate in a differentiated environment (equipped with landmarks) versus an undifferentiated environment, especially with three- and four-year-olds. Familiarity, or past experience, has been found to affect strongly the type and number of landmarks used in an individual's image of his environment (Lynch, 1960). The value of maintaining a distiction be- tween familiarity and differentiation is demonstrated by the existence of environments containing landmarks so salient that "newcomers" could successfully find their way using them, or be able to rank them as high potential landmarks. Siegel (1981) sees high landmark potentials as referring to "the relative values of various environmental features such as reference points--nothing is expected as to why these features make better landmarks" (featnote 5, p. 181). It is of key importance to this study to keep the distinction between familiarity and differentia- tion in the attempt to understand the developmental process of learning 18 to select landmarks, and the relation of this process to other measures of spatial ability. Familiarity perhaps could be analogous to training or practice effects and could be considered a confounding variable. Therefore, the landmark potential, map drawing and landmark location measures are based on an unfamiliar environment. ACT Subtest and Composite Scores Another potentially confounding variable that must be considered in research on Spatial abilities is "general ability", Specifically verbal and mathematical abilities (Fennema & Sherman, 1977). The discussion of factor analysis with the EFT, and the presence or absence of gender dif- ferences in spatial ability when other abilities are controlled for, emphasize the importance of taking into account such moderator vari- ables. The pr0posed study, therefore, will include in its analysis the participants' ACT subtest and composite scores in order to control for possible mathematical, verbal and overall scholastic achievement moder- ator variables. In sum, the present study investigates the relation between measures of Spatial thought-~the EFT, BMT, MRT, DAT space relations, and more environmentally oriented measures of Spatial cognition-—the water-level task, landmark potential, map drawing, and landmark location. CHAPTER II METHOD Sample. The subjects were 154 men and 199 women from the Introduc- tory Psychology classes at Michigan State University. All received credit for partial fufillment of the requirements of their introductory course. Procedure. Testing was conducted in a classroom equipped with a projection screen and writing surfaces for up to 30 peOple. The experi- menter, along with four research assistants, administered the tests to 19 groups with an average of 20 peOple in each group. The testing lasted for 2 1/2 hours, with a ten minute break half-way through the testing session. Instruments. The instruments used, and their available reliabili- ties, are shown in Table 1. For ease of administration, the Group Emb- edded Figures Test (GEFT) was used in place of the EFT. Pilot data on the BMT Showed a strong ceiling effect. Therefore, the study time of four minutes for the stimulus features to be memorized, as well as the response time of four minutes, were reduced to one minute and two minutes, respectively. The water-level task (H20) (Harris et al., 1977) consisted of pictures of 12 groups of four bottles of various Shapes, tilted at various angles, with water lines drawn in them. Only one of the bottles had a true horizontal line; other lines were drawn at various degrees from 180°. The subjects were instructed to choose the one of the four bottles that indicated "how the water line 19 20 TABLE 1 INSTRUMENTS & AVAILABLE RELIABILITIES INSTRUMENT PAPER g PENCIL TESTS ABBREVIATIONS RELIABILITY Group Embedded Figures Test GEFT .828 Building Memory Test BMT .76b Mental Rotations Test MRT .88° .83d Water-level Test H20 .91f Differential Aptitude Spatial Relations DAT .83e SPATIAL COGNITION TASKS SCORING RELIABILITY Map Drawing Task MAP .981‘ Landmark Potential Task TENBEST Landmark Locations Tasks: Good Landmark GOODLM .3uf .97f Poor Landmark POORLM .11f .93f aSec. 1 a 2. Witkin et al., 1971 bSpearman-Brown on original form. Thorndyke & Stasz, 1980 cinternal consistency. Vandenberg a Knee, 1978 dtest-retest, Vandenberg a Kuse. 1978 etest-retest, Bennett et al., 1974 fCronbach's alpha 21 will look". Procedures and materials for the GEFT, MRT, BMT (with time adjustments), and DAT spatial subtest were as described in the respec- tive test manuals (Witkin et al., 1971; Vandenberg & Kuse, 1978; Krauss et al., Note 1; Bennett et al., 1974). Landmark Potential and Map Drawing Tasks. The materials and proce- dures described below are a modification of those used by Allen, Siegel and Rosinski (1978). A simulated walk through a commercial neighborhood in Windsor, Ontario, was developed by means of a series of Slides taken at standpoints three meters apart. A Canon AE1 SLR camera with a stan- dard 50mm lens provided a viewing angle of 47°. The Slides were taken at a constant angle parallel to the ground. When changes in heading took place (i.e., turning a corner), the camera was rotated 24° so that half of the field of vision in the preceding picture was present in the "change of heading" Slide. For the most part, the Sidewalk was directly before the observer with store fronts across the street to the left. Many of the storefront Signs to the right, and in front of the viewer, were visible as well. A final presentation set of 60 slides was selected from the approxi- mately 300 original Slides. Selection criterion was based simply on the complexity of the route and visual complexity of the scenes. For ex- ample, in the later part of the route (fifth, sixth and seventh blocks), no turns were made and few salient Signs could be seen. Therefore, large distances between selected scenes (up to 42 meters) did not inter- fere with the visual continuity of the "walk". In the early portion of 22 the route (third and fourth blocks), where visual and turning complexi- ties were greater, distances between selected slides were smaller (Six to 24 meters). Thus, the presentation set contained scenes separated by gaps from six to 42 meters. Slides were projected at a rate of five seconds each from alternating Kodak Ektagraphic projectors with a dis- solver unit. The dissolver unit allowed for a more realistic simulation by eliminating the "black pauses" between slides that would occur with- out such a device. (See Figure 1 for an illustration of the route.) Prior to the slide tasks, participants were told that: 1)they were first going to see 60 Slides depicting a walk through a commerical neighborhood and were instructed to watch this initial presentation in such a manner as to be able to retrace the steps of the person who took the Slides; 2)in a second viewing, they would be asked to select from a numbered display of the same Slides ten scenes that would be most help- ful to them to remember "how far they had gone along the walk"; and " L_ I ----- d>-—--q—-—-—.' u—-——————u—-—-———-. — _ Figure 1. Route indicating location of high (H) and low (L) landmark potential scenes. 23 3)after the second presentation, they would be asked to draw the route on a city-block map. In the landmark potential task the participants wrote down during the slide presentation the numbers representing the ten scenes they be- lieved to be most helpful to judge distance. The number of each slide (ordered from one through 60) was projected to the right of the screen by a third projector. The ten choices could be recorded in any order. In the map drawing task, participants were asked to draw on a city—block map, the route they believed they had just "travelled". The beginning and end points (first and last Slides) were indicated on the maps (See Figure 2). Subjects were cautioned that the walk might not be the most 0.... 81m .’ 1' I If 1,[ “3 W W‘ Figure 2. City-block map used for map drawing. 24 direct route. Landmark Location T_§k¢ Finally, 16 slightly enlarged scenes con- taining nine "good" and seven "poor" landmarks were projected in random order (see Figure 1 for location of good (H) and poor (L) landmark Slides). Numbers were also projected to the right of these Slides, and subjects were asked to indicate on their route (from the map drawing task) where each slide could be "seen". The landmark potential of these Slides had been determined prior to the experiment by two pilot groups of 22 and 28 undergraduate "judges" from a similar pOpulation. The judges had twice examined the slide series, and selected good and poor landmarks from a display of 5 X 7" prints of the Slides. The prints were mounted on a board in the same order that the judges had viewed them as slides. Prints with high land- mark potential were described as "scenes containing information that would be helpful to know how far you have gone on the walk"; prints with low landmark potential were described as "scenes containing information that would be least helpful to know how far you have travelled on the walk. The instruction "how far" was used to prime distance cues (i.e., city blocks as a metric) in order to enhance performance on the map drawing task. The judges of both pilot groups were asked to rate each slide, indicating on their scoring Sheets to which of the following categories the scene belonged: 1)the 15 "most helpful" scenes; 2)the 15 "least helpful" scenes; and 3)the remaining 30 in a "neither" category. In the initial pilot presentation, very little landmark potential agree- ment was found. The problem was determined as similar information 25 (e.g., same Sign) being visible in more than one Slide. A second and final presentation set of 60 Slides was then selected, avoiding same— content scenes, but keeping a visual continuity in the simulated walk. Using this presentation set with the second pilot group, interrater agreement was improved on the "good" (most helpful) and "poor" (least helpful) scenes. However, using those scenes selected good or poor at least 55% of the time, the interrater reliability is still very low (see Table 1) in comparison to Allen et al.‘s (1978) report of coefficient alpha = .72 for high and .68 for low landmark potentials, using a 25% (at least two of their eight raters) selection criterion. It Should noted that although reliability across raters was low, futher analyses on "intraindividual" (test-retest, parallel forms) reliability could Show that individuals are quite consistent in their landmark potential choices. Equally important is the fact that the route photographed by Allen et a1. (1978) was more differentiated, spanning both commercial and residential neighborhoods, and thus increasing the possibility of agreement among raters. Therefore, despite the low interrater relia- bility for the pilot judges in this study, the 55% criterion was deemed sufficient, with nine Slides considered good landmarks, and seven con- sidered poor landmarks. All "good" landmarks were in "critical areas", or slides of scenes that had salient signs not present in the slide. AS can be seen in Table 1, the scoring of the good and poor landmarks in the final sample yielded alphas of .97 and .93, respectively. In order to control for order effects (keeping the slide tasks in sequence), a Latin Square design was employed to determine test order for each testing session. 26 Scoring and Reliability Total scores for each measure were tabulated as described in each testing manual where available (GEFT, DAT, MRT, BMT). The water-level test (H20) total possible score was 12. "Correct" responses for the landmark potential task were determined by tabulating frequencies and percentages of responses on each Slide. The percentage frequency of the subjects' choices were standardized. Each of the respondent's ten pre- ferences was given a value (standard score), and an average of the values of the ten choices was calculated for each respondent. The map drawing task was scored by three raters. Correct corners crossed (or not crossed) and correct distances drawn were summed, so as to avoid later penalties for early mistakes drawn in the route. Inter- rater reliability for the three raters across the first 86 maps was co- efficient alpha .98 (See Table 1). This was believed to be more than sufficient agreement for scoring, and no further reliability tabulations were done for the remaining maps. Landmark location scores were deter— mined Simply by correct placement of good and poor landmarks. Inter- rater reliability across 2 raters on these scores was a coefficent alpha of .97 for the good landmarks, and .93 for the poor landmarks across 53 maps. The good landmark Score was then divided by 9; the poor landmark score was divided by seven. CHAPTER III RESULTS Mean Scores and Gender The means, standard deviations, numbers of men and women completing each task, and t-values and point-biserial correlations comparing men and women's performance are shown in Table 2. T-tests were also con- ducted on good and poor landmark locations (GOODLM, POORLM) scores. Both men and women scored significantly higher on the GOODLM task than on the POORLM task, with both men and women's t-values 2.55 (p >.05). This seems to indicate that good landmarks are better remembered than poor landmarks, and may be essential for the accurate drawing of the route. With the exception of the landmark potential task (TENBEST), men had higher scores on all Spatial tasks. Men and women appear Significantly different on the MRT, H20, DAT, MAP and GOODLM tasks. Similarly, in this sample, men significantly out-performed women on the ACT Composite and subscores, with the exception of the English subscore. Here women scored Significantly higher than men (see Table 3). Correlations Among Spatial Measures Most of the correlations among the spatial measures for men and wo- men combined were positive and significant, as would be expected. Table 4 shows the paper and pencil tasks--GEFT, BMT, MRT, H20, DAT--all cor- relate with each other at the .001 level. The Slide taSkS--TENBEST, MAP, GOODLM & POORLM--also intercorrelate at the .01 level or less. 27 28 TABLE 2 MEANS, STANDARD DEVIATIONS, AND l-TESTS ON SPATIAL TASKS FOR MEN AND WOMEN Number of 1 Point-biserial [grighlg cases Mean §p Value babilit Correlation ""' (twozta%)e (Probability) osrr ”5 Men 153 10.75 “.61 2 -006 Women 198 10.15 4.70 "2 ‘ 3 21:23; BMT _ 01 Women 196 16.92 4.92 (,hj) ”RT “031+ Men 150 11.83 14.68 6.63 .00 N=31+Ll Women 194 8.47 h.64 (.001) H20 -.36 Women 197 6.75 3.92 (.001) DAT -.13 Women 198 36.52 9.90 (.007) MA? '010 Men 152 12075 1+0“? 1.8} .07 N=31+8 Women 196 11.88 9.36 (.03) GOODLM _ ,1 Men 152 ~22 ~‘9 1.9a .05 N;353 Women 196 .18 .14 (.02) POORLM - men 152 .10 .11; 1.08 .28 N;gg} Women 196 .08 .13 (.1h) TENBEST + 0 Men 154 295°““ 63'97 -.48 .63 N;353 Women 199 248.53 54.19 (.31) 29 TABLE 3 MEANS, STANDARD DEVIATIONS, L-TESTS & POINT-BISERIAL CORRELATIONS ON ACT STANDARD SCORES FOR MEN AND WOMEN Number of t Point-bi i 1 Variable cases Mean §Q Vthe Probability correljfgg: (two-tailed (ProbaEiIity) ENGLISH test) Men 122 19.00 3.71 +.15 -206 g ‘- Women 175 20.27 4 7 3 009 I-Sgg) MATH Men 122 23.76 4.7? . .29 . .36 .000 - women 175 200 39 600,-" 5 Ig:ggl17) SOCIAL STUDIES Men 122 20.53 6.00 -.13 2.2 . 2 - Women 175 18.91 6.20 7 O h ":83?) NATURAL SCIENCE Men 122 24.57 4.95 -.24 4.38 .000 _ Women 175 21.92 5-38 ?:031) COM§?SITE - 20 " en 122 22.15 3.73 3.“? .001 “:29? Jomen 175 20.43 n.77 (.001) TABLE 4 CORRELATION COEFFICIENTS FOR TOTAL,SAMPLE ON SPATIALgTASKS BMT 35“” ' p (.05 (3“;) q, P<001 MET h1"‘ 33i§§ one P (.001 (342) (337) ( ) = Numbers of men and women H20 44"' 24"* 45"‘ included in the analysis (345) (343) (343) DAT 55". “8.1!. 59"! 1,5!!! (345) (340) (341) (346) MAP 09* 08 19... 11- 23". (346) (340) (339) (345) (342) TENBEST 04 OO -01 04 O7 16 (351) (345) (344) (350) (347) (348) GOODLM 16"” 19... guess 17"; 30*** 41'** 13" (346) (340) (339) (345) (342) (348) (348) POORLM 10‘ 05 1t,” 11* 19;** 50"; 16" 36"“ (346) (340) (339) (345) (342) (348) (348) (348) GEFT BMT MRT H20 DAT MAP TENEES GOODLM 30 With the exception of the TENBEST task, the Slide measures are Significantly correlated with the spatial thought measures. The GOODLM task correlates with all the spatial thought measures at the .001 level. With the exception of the insignificant correlation with the BMT, POORLM correlates with all the spatial thought tasks at the .03 level or less. Factor Analysis on Spatial Measures A VARIMAX factor analysis on the complete sample Shows a similar pattern to the simple correlations among Spatial tasks. As can be seen in Table 5, two factors emerge. One factor consists of spatial thought measures and the water-level task; the other factor contains the slide tasks. Men and Women's Correlations The correlations among the spatial tasks for men were similar to the combined sample (see Table 6). The Spatial thought and water-level tasks all correlated with each other at the .001 level. The men's Slide task scores intercorrelated to a greater degree than in the combined sample, with all correlations signficant at the .003 level or less. This is consistent with the findings of Ray et al. (1979) and Liben (197s). Correlations of the men's spatial thought and water-level tasks (paper and pencil tasks) with slide tasks differed from the combined sample, with only the GOODLM task Significantly correlating with all the pencil and paper tasks. The POORLM task significantly correlated with only the GEFT and DAT (p <.02). Correlations among all paper and pencil tasks for women were highly 31 TABLE 5 ACTOR ANALYSIS ON SPATIAL TASKS -——-—— FOR TOTAL SAMPLE FOR MEN FACTOR 1 FACTOR 2 FACTOR 1 FACTOR 2 GEFT .66370 ‘ .03054 GEFT .65515 * .06804 BMT .5198? ' .05322 BMT .57589 * .02240 MRT .67575 ' .15082 MRT .6767} t .11109 320 .58138 a .08884 H20 .6129? * .04330 DAT .63441 t .17894 DAT .81259 a .19068 MAP .10696 .73298 * MAP .09105 .72783 * TENBEST .00198 .20736 * TENBEST -.00643 .31590 * GOODLM .23570 .52909 * GOODLM .2414} .62419 * POORLM .07069 .67346 . POORLM .05050 .66550 a Eigenvalue 2.57650 1.07731 Eigenvalue 2.61310 1.22870 EQBJEEEE FACTOR 1 FACTOR 2 FACTOR 3 GEFT .70425 * -.02707 .19848 BMT .50299 t .1032? .01910 MRT .66032 a .20859 -.20021 H20 .5107} a .0852? .12075 MAP .07766 .71595 * .1035? TENBEST .02819 .07164 .26720 n GOODLM .19782 .uuzha . .0152? POORLM .04882 .68308 a .12971 Eigenvalue 2.47461 1.02955 .17262 It indicates highest factor loading 32 TABLE 6 PEARSON CORRELATION COEFFICIENTS FOR MEN AND WOMEi N SPATIAL TASKS Men 36"“ Women 34'" ' P< .05 (195) ., p< .0, Men 39... 38059 0.9 p< .001 MDT (149) (115) ( ) N b r d ‘” . ... es. = um era 0 men an women Women 7703) %?92) included in analysis Men “8.0! 27... A2004 H20 (152) (148) (150) women “30.. 2h... 35... (196) (195) (193) Men 63... A895. 5700. 46... DAT (148) (144) (147) (149) women 56... “80.. 60.00 “2.99 (197) (196) (194) (197) Men 09 1O 15‘ 05 23" , (151) (147) (148) (151) (147) “A? Women 08 08 19" 11 2OH (195) (193) (191) (194) (195) Men 02 -O9 05 1O 08 23 szggsT (153) (149) (159) (155) (149) (152) Women 06 08 -O4 O1 O7 O9 (198) (196) (194) (197) (198) (196) Men ‘8. 19¢. 23;. 144 32cc. “7.00 21.9 GOODLM (151) (147) (148) (151) (147) (152) (152) Women 14" 21" 20" 15' 25“" 34“” O4 (195) (193) (191) (194) (195) (196) (196) Men 2?“ -02 08 06 19a “9... 2200 40"“ POORLM 51) (1h?) (IAB) (151) (1A?) (152) (152) (152) ‘ Women 04 1O 16' 12 19" 49". 10 319.. (195) (193) (191) (194) (195) (196) (196) (196) GEFT BMT MRT H20 DAT MAP TEN BEST GOODLM 33 significant (p (.001), Similar to the men's performance (see Table 5). However, correlations among the slide tasks differed from the men's in- tercorrelations. The GOODLM task correlated highly significantly (p <.001) with the MAP and POORLM tasks, but not with TENBEST. TENBEST did not Significantly correlate with any of the remaining slide tasks (MAP, GOODLM, POORLM). In addition, POORLM correlated highly Signifi- cantly (p <.001) with the MAP task for the women. ' Correlations 0f the women's paper and pencil tasks with the Slide tasks differed from the men's performance. Whereas none of the paper and pencil tasks significantly correlated with MAP for men, MAP Signifi- cantly correlated (p <.OO4) with the MRT and DAT tasks for women. Cor- relations Of POORLM with the paper and pencil tasks also differed from the men's significant correlations (POORLM with GEFT, DAT, p <.02). In contrast, women's POORLM scores intercorrelated with those tasks that did not correlate Significantly for men. That is, POORLM correlated significantly (p <.05) with the MRT, H20, and DAT tasks. Similar to the men's performance, none of the paper and pencil measures correlated Sig- nificantly with TENBEST, and all paper and pencil tasks correlated sig- nificantly with the GOODLM task. Thus for both men and women, TENBEST does not appear to tap Skills required by the Spatial thought, or water- level tasks. And, relations between the landmark location tasks and paper and pencil measures appear quite different for men and women. Separate Factor Analysis on Spatial Measures Factor analyses conducted separately for men and women (Table 5) summarize the simple correlation differences. The factor analysis on 34 men's scores shows nearly the same factor loadings as the combined ple. However, the TENBEST score loads somewhat higher on the second factor for men. This would suggest that women Show a different pattern of performance from men on TENBEST. And as can be seen in Table 5, TEN- BEST ends up on a third factor itself for women. The low interrater reliability of the pilot judges, who consisted of both men men and wo- men, may reflect this gender difference in performance. The extent to which men and women's intercorrelations differed was determined using Fischer's Z (Table 7). Significant differences (p <.01) appear in the following correlations: GEFT/DAT, GEFT/POORLM, BMT/TENBEST, BMT/POORLM, MRT/TENBEST, MAP/TENBEST, MAP/GOODLM, and TEN- BEST/GOODLM/POORLM pairs. The gender difference in the GEFT/DAT, with men's correlations significantly higher than women's, is consistent with the findings of Hyde et al. (1975) of Identical Blocks/GEFT relations. TABLE 7 COMPARISONS OF MEN & WOMEN'S CORRELATIONS USING FISCHER'S Z B‘VlT 8 Negative values indicate higher positive ‘ '5 coefficients for women. Positive values indicate higher positive MRT "1'65 1’"3 coefficients for men. H20 1.58 .80 2.089 * P<-05 DAT 2.70" 0 -1.13 1.25 9* p<.01 MAP 025 .50 ‘1003 -1050 .78 ..* p<.001 TENBEST -1.00 -6.75**‘ 4.75*** 2.38. .50 3.759. GOODLM 1.03 .53 .78 -.25 1.63 3.751.. n.20404 POORLM 3.30" -5.50*’* -2.15* -1.40 .15 0 3.23’* 2.58" GEPT BMT MRT H20 DAT MAP TENBEST GOODLM 35 Factor Analysis of Spatial Measures and ACT Scores A second factor analysis that included both spatial measures and ACT scores was computed (Table 8). For men and women combined, the three factors of 1)paper and pencil Spatial tasks; 2)spatial Slide tasks; and 3)ACT scores appeared. Separate factor analyses for men and women yielded different results from the combined sample. Beyond the three factors similar to the combined sample, ACT Math loaded highly and separately on a fourth factor for men. For women, three factors again appeared, but the TENBEST score loaded with the ACT scores, rather on a factor by itself (of. Table 5). 36 TABLE 8 FACTOR ANALYSIS ON SPATIAL TASKS & ACT SCORES FOR TOTAL SAMPLE EQRJWQMEN FACTOR 1 FACTOR 2 FACTOR 3 FACTOR 1 FACTOR 2 FACTOR 3 GEPT .61993 ’ .23396 -.01361 GEFT .2841? .65711 0 -.04213 BMT .49289 t .15660 .04224 BMT .24025 .46331 t .12026 MRT .66432 t .12733 .14208 MRT .17533 .61390 a .19346 H20 .55803 * .06250 .07939 H20 .01019 .53839 - .10851 DAT .84430 * .16831 .13948 DAT .27028 .81539 F .17379 MAP .10028 .06146 .74103 a MAP .13106 .11881 .6834? ' GOODLM .22643 .1453? .5256? 0 GOODLM .19481 .1807? .45303 * POORLM .07710 .09688 .6367? * POORLM .05739 .06636 .70083 * TENBEST -.04009 .12056 .15184 * TENBEST .15804 t .01920 .05312 ENGLISH .24800 .60251 a .15180 ENGLISH .65963 ' .34703 .13914 MATH .45570 .49462 a .18483 MATH .60472 ' .39453 .16436 soc .12844 .8046? * .14142 soc .78617.. .15352 .06342 NATSCI ,34062 .74212 a .14162 NATSCI ,82069" ,21224 .11124 Eigenvalue 3.92056 1.09264 .93184 Eigenvalue 4.17344 1.01852 .92523 FOR MEN FACTOR 1 FACTOR 2 FACTOR 3 FACTOR 4 GEFT .57754 a .14742 -.00348 .15919 BMT .51003 * .02935 -.04037 .10364 MRT .64923 * .08225 .04735 .04731 H20 .55041 * .17123 -.02440 -.O4239 DAT .83690 * .03769 .10292 .15269 MAP -.03444 —.01880 .73178 * .13150 GOODLM .30333 .14082 .68554 F -.11231 POORLM .02120 .16790 .58789 a .2138? TENBEST -.08196 .09602 .25191 a -.06414 ENGLISH .17088 .57548" .19195 .17939 MATH .30778 .25958 .09830 .78573 ' soc .02133 .93135" .17017 -.02332 NATSCI ,36540 .65281" .08855 ,20204 Eigenvalue 3.38757 1.52417 1.04136 .58343 .indicates highest factor loading 37 TABLE 9 COBRELATIONS BETWEEN SPATIAL TASKS AND ACT SCORES FOR HEE AED WQMEE COMP ENG MATH SOC NATSCI N GEFT Men 35... 29... 35... 01 35... 122 women 1‘6... 46... m... 35... 38... 171+ BMT Men 20" 11 26" 05 22M 118 women 38... w... 35... 29... 33... 172 MRT Men 28*" 18' 27"' 11 35"' 118 women 37... 36... 37... 24". 31... 172 H20 Men 26H 19" 15* 16’ 29". 121 Women 17H 17H 23"“ O9 10 173 DAT Men 35'" 17' 41'" 10 38-" 118 women 51... 1‘5... 53... 36... 1+0... 17‘" MAP Men 11 O? 11 09 07 120 Women 25H 25H 25" 14' 21H 172 Women 20" 14' 17" 15' 19H 175 GOODLM ”e“ 37'" 28'" 09 23" 26" 120 Women 32... 20.. 29... 29... 27... 172 POORLM Men 31"* 27". 25H 22" 21H 120 Women 18H 19H 19" O9 13' 172 ' P<¢05 I. p‘.o1 "' p‘<.001 38 ACT Scores as Covariates Because the ACT Composite and English, Math, Social Studies and Natural Science subscores all intercorrelated to some degree with the spatial tasks (see Table 9), and men and women's intercorrelations ap- peared to be significantly different on both ACT and spatial tasks, par- tial correlations were computed for men and women separately. Tables 10, 11, 12, 13 and 14 (Appendix A) show the results of par- tially out the ACT Composite and subscores on the correlations among spatial tasks for men and women. Comparing men and women's original correlations (Table 6) with each of the partialled matrices (Tables 10- 14), only slight decreases in degree of correlation appear in the par- tialled matrices. Differences among the partialled matrices are small as well. Partialled Comparisons of Men and Women's Correlations The degree to which men and women's correlations differed after par- tialling was of interest. Intercorrelations were again compared using Fisher's Z (Tables 15-19, Appedix B). Comparing the results in Tables 15 through 19 (comparisons of partialled correlations) with those in Table 7 (comparison of raw correlations), it appears that when the in- fluence of any of the ACT subtests or Composite scores is controlled, no significant difference in intercorrelations can be found between men and women. It should be noted that where Ray et al. (1979) and Liben (1978) found correlation differences, neither had controlled for math or scholastic ability. Since the number of significant sex-related intercorrelation dif- 39 ferences were reduced when ACT scores were controlled, it seemed reason- able that point-biserial correlations would also change if ACT scores were controlled. Point-biserial Comparisons of Partialled and Original Scores Thus, sex-related differences in test performance were again ex- plored using point-biserial correlations, but with ACT Composite and subscores held constant (Table 20). of association between gender and raw scores (same as Table 2). The first column shows the degree When comparing the degree of association across rows (noting asterisks), ACT Math and English partials reveal different patterns of relationships among men and women's scores, with the exception of the MRT and H20 TABLE 20 COMPARISONS OF POINT-BISERIAL CORRELATIOES 0N SPATIAL TASKS CONTROLLING FOR ACT SCORES W OR G NA POINT-BISERIAL CORR TIONS b c NO COM- SOCIAL NATURAL PAWTlAL POSITE ENGLISH MATH STUDIES SCIENCE GEEI -O6 0 -14" 03 -05 01 BMT -01 02 —08 05 -02 03 Mgr -31+... -29... -37... -25... -32... -26... H20 -36... -31... -37... -28... -33... -30... EAT -13" -08 -22'*' -O1 -13' -07 MA? -10' -10' -17" -08 -12' -10* TENBEST 03 04 O1 04 04 04 GOODLM -11* -O5 -13" -05 -07 -04 POORLM -06 -03 -10- -01 -05 -03 aNegative values indicate higher scores for men. .: p<:.gf bN's range from 344 to 353 (See Table 1). ... B: :001 C": 272 for all ACT scores 40 tasks. Controlling for Math and English separately appears to have Opposite effects in regard to sex-related differences in performance. In comparison to the original point-biserial, partialling out the Eng- lish ACT score affects the correlation in the direction favoring men's performance. A significant correlation appears for the GEFT (p <.01) when English is controlled, with men appearing to perform better than women. Further gender differences increase in this direction for all other tasks when English is controlled. When Math is controlled, the sex-related differences are reduced, with the MRT and H20 tasks the only significant gender/task correlations remaining, again with men excelling. Considering the influence of ACT score partialling on correlation comparisons, factor analyses were repeated, but with ACT Math and English scores partialled out. Because the partialling of Math and English did not differentially affect factoring for either men or women, "MATH OR ENGLISH" partialling is indicated in Table 21. (This may be due in part to the equal N's created by partialling.) As is shown, three factors for women and two factors for men emerge when ACT Math or English is partialled, with these numbers of factors consistent with the nonpartialled factors. Similarly, men's scores fall on the same factors here as the nonpartialled factors (see Table 5). And again, women's TENBEST score falls on a third factor, but is now joined by the GEFT and waterblevel scores in the partialled factors. Although GEFT/water-level correlations are higher for men than for women in both raw and partialled correlations, it is interesting to note that when ACT Math/English is controlled, women appear to be using similar 41 TABLE 21 ACE MATH OR ENGLISH PARTIALLED FACTORS N AND W0 FACTORS FOR WOMEN FACTORS FOR MEN FACTOR FACTOR FACTOR FACTOR 1 FACTOR 2 GEFT .10934 -.13384 .35503 GEFT -16746 ' -.01647 BMT .07412 .02079 -.02404 BMT .10776 * -.0507? MRT .08881 * .03054 .05072 MRT .17480 * -.01860 H20 -.01783 -.03755 .44222 “20 o16318 * -.02220 DAT .7961? .02086 -.33629 DAT -52835 * ..00339 TENBEST -.01985 .01190 .07294 TENBEST -.02508 .08086 * MAP -.09358 .41138 . .08175 MAP '-°6626 .45559 * GOODLM —.0113? .18044 a .04099 GOODLM -05524 .32340 . POORLM -.06324 .4281; * .01126 POORLM :121611 .26128 * Eigenvalue 2.61841 .97516 .19420 Eigenvalue 2028375 1-32351 ‘indicates highest factor loading skills to perform the TENBEST, GEFT and water-level tasks. This would not be expected in light of Liben's (1978) findings of a higher relation between the GEFT and water-level task for high school boys than for girls. Thus it appears that even when ACT Math or English is controlled, gender differences are still apparent where relations among all measures (i.e. factor loadings) are concerned. The fact that TEN- BEST loads on a separate factor for women suggests that different strategies may have been used by men and women to complete the task. Futher, this sex-related differences may have contributed to 42 the low interrater reliability among the pilot judges determinations of good and poor landmarks. Summary In sum, the initial analyses indicated that there were sex-related differences in ACT scores, and significant correlations among ACT scores and spatial measures. Thus subsequent analyses of correlation compari- sons and factor analyses controlled for ACT performance. The most marked sex-related difference in spatial tasks appeared when English achievement was controlled. The fewest number of gender differences appeared when Math achievement was controlled. CHAPTER IV DISCUSSION The three major questions investigated in this study were: 1)what is the degree of relation between spatial thought tests--the Group Embedded Figures Test (GEFT), the Building Memory Test (BMT), the Mental Rotation Test (MRT), and the Differential Aptitude spatial subtest (DAT)--and spatial products of spatial cognition--Landmark potential (TENBEST), Map drawing (MAP), Landmark location tasks (GOODLM, POORLM) and the Piagetian water-level task; 2)what is the degree of sex-related dif- ferences in performance on these tasks; and 3) what is the degree to which scholastic achievement, as measured by the ACT, can account for the sex-related differences that do appear. Because of the scholasitc achievement (ACT) differences between men and women and the strong associations between the ACT scores and spatial measures (hypothesis 3, above), the results of the t-tests, point- biserial correlations, simple correlations, and factor analyses on the raw scores for spatial measures may be potentially misleading. There- fore, the only results discussed will be those that have ACT scores controlled for. Relations Between Spatial Thought and Spatial Cognition Regarding the first hypothesis, partialled correlations between spa- tial thought and spatial cognition tasks varied with the ACT score that was partialled. The most consistent relation between the two types of spatial tasks was the DAT/GOODLM correlation, significant for both men 43 44 and women in all partialled correlations. Specific gender differ- ences in the slide/paper and pencil correlations will be discussed below. In related investigations of map reading skills, significant cor- relations among spatial-visual measures and map interpretation and map recall are also lacking. Underwood (1981) failed to find a strong re- lation between a map interpretation task and a visual-spatial task, es- pecially with experienced map readers. Among three groups of 15- to 17-year-old girls representing three levels of experience in reading maps (inexperienced, modestly experienced and highly experienced), only the scores of the least experienced group were significantly correlated with performance on the visual—spatial task. Underwood attributed the result to the way the map task was approached. She hypothesized that the unskilled group approached the (tOpographiC) map task as a visuo- spatial task (i.e., recognizing geometrical configurations, discerning patterns among contour lines). The skilled map reading group, on the other hand, had greater experience with maps which might compensate for any spatial-visual component deficit, according to Underwood. Alter- natively, the skilled map readers may have been better at component skills to begin with. In the present study, however, Moore's (1979) suggestion that rota- tion is a specific cognitive ability in cognitive mapping is partially supported by the significant relations between the DAT spatial subtest (considered a mental rotation-like task) and the slide measures. But the lack of association between the MRT and slide tasks could also sug- gest otherwise. 45 Thorndyke and Stasz (1980) also assumed that visual-Spatial memory would be a necessary skill in reproducing maps from memory. Using the original BMT (Note 7) to measure visual memory ability, they found no difference in performance between experienced and novice map readers. In their main effort to identify procedures for knowledge acquisition from maps, they found that the BMT also correlated significantly with their identified categories of element recall, spatial attribute recall, and verbal attribute recall to a similar degree for both novice and ex- perienced map readers. However, a difference in visual memory (BMT) appeared in a regression analysis after implementation of training in map "element" learning, where high visual memory subjects (experienced map readers) benefitted the most from training. In contrast to Under- wood's interpretation, Thorndyke and Stasz suggest that the high visual subjects benefitted because the effectiveness of spatial learning pro- cedures probably depends on their well-developed spatial memory ability. In the present study, the revised BMT did not relate significantly to the MAP task, nor with the other slide tasks, contrary to what was pre- dicted. However, further computations such as regression analyses with good and poor performers on the BMT and spatial cognition tasks might reveal some significant relations between these measures. Internal Consistency of Slide Tasks. It can be argued that since the TENBEST task did not relate to the MAP and GOOD/POORLM tasks in any par- tialled analysis, it may not be as environmentally valid as the other three slide tasks. The reliability (in the sense of internal consistency) of the slide tasks is not as great as it is on spatial thought measures, which all intercorrelate similarly. If task procedures are considered, a plausible 46 eXplanation for the difference between the TENBEST task and the other slide tasks suggests itself. The MAP and GOOD/POORLM tasks require that the sub- ject recall a route and scene, and designate the route and landmarks. The TENBEST task requires only that a judgment be made on each scene, where location may or may not be considered crucial by the subject. However, this judgment is used in the GOOD/ POORLM tasks, since significantly more good landmarks were correctly located. Therefore, the TENBEST task can be considered an environmentally valid measure that is a prerequisite to land- mark location accuracy. Specific Predictions. Specific predictions about interrelations of the two types of spatial tasks (the first hypothesis) were also made. The hy- pothesis that the perceptual disembedding essential for the GEFT would also be an ability necessary for the water-level task (H20) and landmark poten- tial (TENBEST) was supported in the women's partialled factor analysis, where these three tasks loaded on a separate and distinct factor. The strong association between the GEFT and water-level task in the raw and partialled correlations parallels the findings of Liben (1978). The rela- tion between the DAT and MRT as a nonanalytic pair is supported to some degree in the partialled correlations. As two measures of nonanalytic spa- tial ability, the DAT and MRT do correlate to a higher, but not significant degree, than the nonanalytic/ analytic pairs of the MRT/GEFT and DAT/GEFT. Sex-related Differences As was discussed above, the significant relation between the water- level task the the GEFT supported the findings of Liben (1978). The direction and the lack of significance of the gender differences found 47 here, however, do not support her findings in regard to sex-related dif- ferences. In the partialled analysis where Natural Science is con- trolled, women's scores correlate higher (but not significantly) than men's. And, the partialled factor analysis resulting in the TENBEST, GEFT and water-level tasks loading on a same factor (discussed above) for women, suggests that women used similar abilities to solve all three tasks. This would also contradict Liben's results. However, it should be noted that Liben did not control for scholastic achievement in her analyses. In regard to the interpretation of the sex-related differences in partialled factor loadings for the H20, GEFT and TENBEST, further information on the GEFT and related Rod and Frame Test (RFT) is rele- vant. In their sample of college students, Hyde, Geiringer and Yen (1975) found no significant difference in the GEFT between men and women's scores, as is the case in this study. Futhermore, the sex- related difference that was found in the EFT did not exist when spatial ability was controlled. This would suggest that the gender difference that appeared on the BET in Hyde et a1.'s study was not due to disembedding skills, but to spatial ability (as measured by the Identical Blocks Test). Implications of this finding for the present study would suggest that the H20/GEFT/TENBEST partialled cluster is not a "disembedding" factor, but rather a "second compo- nent" (or "analytic" component?) of spatial ability, at least for women. Again, analysis of gender differences on the water-level task, controlling for various spatial ability measures, could provide support for this hypothesis. 48 ACT and Sex-related Differences As was noted in a previous discussion, when various ACT scores are controlled for, the number of tests where men and women's scores differ significantly dr0ps from four in the original correlations, to two in partial correlations. With partialling, men still consistently and sig- nificantly (p <.001) out-score women on the MRT and H20 tasks. Studies reporting men outperforming women on the DAT (Note 3; Note 4; Hartlage, 1970) were not supported here except in the nonpartialled and English and Social Studies partialling analyses. When Math, Composite, and Natural Science scores were held constant, no significant gender dif- ference appeared. The number of significant differences between correlations drops from 13 (Table 6) to zero (Tables 13-17) when the ACT Composite, or any ACT subscore is partialled. This should not come as a surprise, since men and women differed significantly on all ACT scores (p (.02 to p (.0005); Table 3), and correlations between most of the ACT scores and spatial tasks were significant for both men and women (Table 9). Unpublished ACT data for the larger University population from which the present study's participants were recruited also show a similar pat- tern of sex-related differences (NCte 5). In a combined sample of 5,765, women scored higher than men on the English subtest, and men scored higher than women on the remaining subtests and Composite scores (all p <.001). An ACT technical report (1973) also gives separate pre- diction equations for men and women's college GPA's. According to the report, when one prediction equation was used for men and women com- 49 bined, overpredictions for men and underpredictions for women often resulted. Math and English Achievement. Examining the specific effects of ACT partialling on men and women's Spatial measures, English and Math achieve- ment appear to be related to spatial tasks quite differently. When English achievement is controlled for, men score significantly higher on the GEFT and DAT (p <.001), higher on MAP and GOODLM (differences increase to p <.01), and higher on POORLM (p (.05). In attempting to interpret why men score relatively higher on these tasks when English achievement is con- trolled, one could consider that it has the opposite effect from the Math partialling. Except for the MRT and H20, Math achievement appears to be equally related to paper and pencil tasks and MAP and GOODLM environmental tasks for both men and women (i.e., no significant gender differences), whereas English is not so equally related. Women who do well on spatial ability tasks also do well in Math, English and Natural Science ACT sub- tests. Further support for this interpretation can be seen in the inter- correlations between spatial tasks and ACT scores (Table 9). Women's in- tercorrelations tend to be higher than men's on nearly all of the measures. Except for the DAT and MAP tasks, the results of partialling Social Studies and Natural Science scores for women do not greatly differ from the effects of the Math partialling. The factor analyses of both spatial tasks and ACT scores also showed sex-related differences (Table 8). Both men and women's separate an- alyses revealed different factors from the combined sample. For women, TENBEST loaded on the factor containing ACT scores, which might have 50 been predicted from the previous factor analyses (Table 5) where TENBEST loaded on a third and separate factor from the slide and paper and pencil factors. In the men's ACT/spatial task factoring, Math appeared to be the outlier, falling on a separate and fourth factor (Table 8). This seems to support the interpretation that the men in this sample can do well on ACT Math, but not necessarily in the other ACT subtests. Women, on the other hand, do well in all ACT subtests if they do well in Math. Relevant here are the findings of Thompson, Mann and Harris (1981) on cognitive complexity (measures of ambiguity & complexity), gender and spatial (water-level) task performance. Significant relations between cognitive complexity and the water-level task appeared for men, but com- parable correlations for women were nonsignificant. The several explan- ations prOposed by Thompson et al. could be applicable here as well: 1)Men may depend on a differentiated cognitive style to a greater extent than do women; 2)Women may use a greater variety of strategies in ap- proaching a spatial problem; and 3)Men are more likely than women to use a spatial strategy to deal with non-spatial tasks. Further inquiry, as to what strategies are used by men and women to solve spatial tasks, is needed to resolve such issues of gender differences in cognitive differentiation. Conclusion Taken altogether, the many analyses of these data lead to two main conclusions. First, the general lack of significant relations between measures of spatial thought and spatial (environmental) cognition, sug- gest that either different strategies or different spatial skills are 51 used to solve the two types of tasks. However, the significant rela- tions between the DAT spatial subtest (considered a mental rotation-like task) and slide measures partially support Moore's hypothesis that rota- tion is a specific cognitive ability in cognitive mapping. It was sug- gested that further analyses, examining differences between high and low performers on cognitive mapping tasks, might show different patterns of relations among spatial ability and spatial cognition. With the limited relations between the landmark potential tasks and other slide tasks-~map drawing and landmark locations-~questions can be raised regarding the validity of slide tasks themselves. Making land- mark potential judgments appeared not to be directly related to drawing a map of the route and the later location of good and poor landmarks. However, since more good landmarks were accurately placed compared to poor landmarks, landmark potential is a necessary prerequisite for the correct placement of landmarks on a hand-drawn map. A second major conclusion is that gender differences appear, or do not appear, depending on whether Math or English achievement is taken into consideration. With the exception of the water-line task and the Mental Rotations Test, the controlling of ACT scores affected sex- related differences in mean scores and degrees of correlation among Spa- tial tasks. When comparing men and women's performance on each of the spatial tasks, controlling for ACT English resulted in men scoring sig- nificantly higher on the GEFT, MAP, GOODLM, POORLM, and MRT and H20 tasks. When ACT Math was controlled for, only differences in the MRT and H20 tasks remained, again with men outscoring women. Furthermore, if any ACT subscore is partialled, no significant sex-related difference 52 in degree of relation among any of the spatial tasks is found. The relationship of Math ability to the other ACT subtests also dif- fered between men and women. It was suggested that the men in this sample who did well on Math, did not always do well on the other ACT subtests. For the women in this sample, however, it appears that if Math was mastered, abilities tested by the remaining ACT subtests were mastered as well. Several interpretations were offered, based on the hypotheses of Thompson et al. (1981), suggesting differential cogni- tive styles, or greater use of different strategies among women. As for the question of gender differences in spatial cognition, non- partialled point-biserial correlations showed that men outperformed wo- men on the Map drawing and good landmark location tasks. When Math achievement was controlled, however, no differences in these spatial cognition (slide) tasks appeared. 0f the correlations between the spa- tial thought and slide tasks, significant relations were limited to the DAT/slide tasks pairs. As was noted above, sex-related differences on the DAT were not apparent when ACT Math was controlled. The presence of gender differences in two of the four slide tasks, and their relation to Math achievement and the DAT spatial subtest suggest that the spatial cognition measures used here are somewhat related to spatial thought tasks. Alternatively, it may be that gender differences exist to some degree in the environmentally oriented tasks used here, just as gender differences have been noted in other spatial behavior and representation tasks (see Harris, 1981, Chap. 4). And, these differences could be due to differences in cognitive strategies, or for other reasons not explained or suggested here. 53 Summary In sum, if one's math achievement is known, gender seems to make little difference in the spatial cognition procedures used here, or in the Embedded Figures Test, the Building Memory Test, or the Dif- ferential Aptitude Spatial subtest. Further analysis of the water- line task, controlling for various spatial ability and spatial cog- nition tasks, were proposed to account for the sex-related difference found in this task. The sex-related difference found in the Mental Rotations task, however, remains unaccounted for by any of the ACT scores in this study. The implications of these findings for future research suggest that Math achievement and spatial performance may tap similar skills for both men and women. However, the need to assess men and women separately may be mandated where various components of spatial ability (or math ability) are sought. And, since the nature of the component differences may lie in strategy differences, inquiry into various strategies used by men and women may prove helpful to further explain gender differences and similarities in spatial and math performance. REFERENCE NOTES Krauss, I., Awad, Z. & McCormick, D. Spatial cognition in older adults: From lab to life; learning, remembering, and using Spatial information as an older adult. Paper presented at the Biennial meeting of the Society for Research in Child Development, Boston, March 1981. Wolf, V. A study of the Embedded Figures Test in relation to sex differences and some primary mental abilities. Cited in M. A. Wittig and A. Petersen (Eds.), Sex-relgted differences in cognitive functioning: Developmentgl issues. New York: Academic Press, 1979. Harris, L. J. and Wagner, N. Performance by young adults on the spatial subtest of the Differential Aptitude Test: Sex dif- ferences. Unpublished data, Michigan State University, 1977. Flannagan, J. C., Dailey, J. T., Shaycroft, J. F., Gorham, W. A., Orr, D. B., Goldberg. I. & Neyman, C. A. Jr. Counseler's tech- nical manual for interpreting test scores. (Project Talent) Palo Alto, CA, 1961. Cited in E. Maccoby and C. Jacklin, The, ch 10 0 sex differences. New York: University Press, 1974. Allen, T. & Talor, C. Michigan State University ACT, GPA and major preferences. Unpublished data, Michigan State University, 1982. Ekstrom, R. B., French, W. & Harmon, H. Kit of factor—referenced cognitive tests. Princeton, NJ: Educational Testing Service, 1976. 54 LIST OF REFERENCES LIST OF REFERENCES Acredolo, L. P., Pick, H. L. & Olsen, M. G. Environmental differenti- ation and familiarity as determinants of children's memory for spatial location. Develo ental s ho , 1975. 11, 495-501. Allen, G. L., Kirasic, K. C., Siegel, A. W., & Herman, J. F. De- veIOpmental issues in cognitive mapping; the selection and util- ization of environmental landmarks. Child Development, 1979, .29, 1062-1070. Allen, G. L., Siegel, A. W. & Rosinski, K. K. The role of perceptual context in structuring spatial knowledge. Journal of Experimental Psychology: Human Learning and Memory, 1978, 4, 617-630. Bennett, G. K., Seashore, H. G. & Wesman, A. G. Diffigygntial_Aptitnde Tests. New York: Psychological Corporation, 1974. Bettis, N. C. An assessment of the geographical knowledge and under- standing of fifth-grade students in Michigan. Unpublished doctoral dissertation, Michigan State Univerisity, 1974. Cited in L. J. Harris, Sex-related variations in spatial skills, in L. Liben, A. Patterson & N. Newcombe (Eds.), Spatial representgtion and avior across the life span, 83-125, New York: Academic Press, 1981. Coates, S. Field independence and intellectual functioning in pre- school children. Perceptual and Motor Skills, 1978, 51, 251-254. Cohen, J. The factorial structure of the WISC at age 7-6, 10-6, and 13-6. Journal of Consulting Psychology, 1959, 21, 285-290. Cohen, J. The factorial structure of the WAIS between early adulthood and old age. Journal of Consulting Psychology, 1957, glJ 283-290. Fennema, E. & Sherman, J. Sex-related differences in mathematics achievement, spatial visualization and affective factors. Ameri- can Educational Research Journal, 1977,_g, 51-71. Gibson, E. J. Principles ofgperceptual learnigg and develgpgen . New York: Appleton, 1969. Gibson, E. J. Gibson, J. J., Pick, A. D. & Osser, H. A developmental study of the discrimination of letter~like forms. Journal 9f Com- parative and Physiological Psychology, 1962, 55, 897-906. 55 56 Gildemeister, J. & Friedman, P. Cognitive style and visual analysis in first graders of high and low ability. Perceptual and M9191 SKILLS. 1978’ .419 759-766. Goldstein, K. & Blackman, S. Cognitive style: Five gpproaches and relevant regearch. New York: Wiley, 1978. Goodenough, D. R. & Karp, S. A. Field dependence and intellectual functioning. Journal of Abnormal and Social Psychology, 1961, 241-246. Harris, L. J. Sex related variations in spatial skill. In L. Liben, A. Patterson & N. Newcombe (Eds.), Spgtial representation and behavior acrosggthe life span, 83-125, New York: Academic Press, 1981. Harris, L. J., Manley, C. & Best, C. T. Conservation of horizontality: Sex differences in sixth-graders and college students. In R. 0. Smart & M. S. Smart (Eds.), Readings in Child Develonment gnd Rela- tionships, 2nd Ed., 375-387. New York: McMillan, 1977. Hart, R. A. & Moore, C. T. The development of spatial cognition: A re- view. In R. M. Downs and D. Stea (Eds.) Image and Environment: Cognitive Mappigg and Spatial Behavior. Chicago: Adine, 1973. Hartlage, L. C. Sex-linked inheritance of spatial ability. Perceptual gnd Motor Skills, 1970, 31, 610. Herman, J., Kail, R. & Siegel, A. Cognitive maps of a college campus: A new look at freshman orientation. Bulletin of the Psychonomig Society, 1979,_13, 183-186. Herman, J. F. & Siegel, A. The development of cognitive mapping of the large-scale environment. Journal of Experimental Child Psychology, 1978, gé, 389-406. Hyde, J. S. How large are cognitive gender differences? A meta- analysis using eta square and "d". American Psychologist, 1981, 36, 892-901. Hyde, J. S. Geiringer, E. R. & Yen, W. M. On the empirical relation be- tween spatial ability and sex differences in other aspects of cog- nitive performance. Multivariate Behgviorgl Reggarch, 1975, 19, 289-310. Karp, S. A. Field dependence and overcoming embeddedness. Perceptual and Motor Skills, 1963, 24, 603-609. Liben, L. Spatial representation and behavior: Multiple perspectives. In L. Liben, A. Patterson & N. Newcombe (Eds.) Spatial representa- tion and behavior across the life span, 3-32, New York: Academic Press, 1981. 57 Liben, L. Performance on Piagetian spatial tasks as a function of sex, field dependence, and training. Merril-Palmer Quarterly, 1978, 24, 99-110. Lynch, K. The Image of the city. Cambridge, Mass: MIT Press, 1960. Maccoby, E. & Jacklin, C. The psychology of sex differences. New York: University Press, 1974. Moore, G. T. Knowing about environmental knowing; the current state of theory and research on environmental cognition. Environment and Behavior, 1979.,11. 33-70. Piaget, J. & Inhelder, B. The child's conception of space. New York: Norton, 1967. Ray, W. J., Georgiou, S. & Ravizza, R. Spatial abilities, sex dif- ferences, and lateral eye movements. Developmental Ps cholo , 1979. 12. 455-457. Research and DeveIOpment Division; The American College Testing Program. Assessing students on the way to college: Technicgl report for the ACT assessment program, Vol. 1. Iowa City, Iowa: ACT Publications, 1973. Shephard, R. N. & Metzler, J. Mental rotation of three-dimensional objects. Science, 1971, 111, 701-703. Sherman, J. Sex-relgted cognitive differences. Springfield, Ill: Charles C. Thomas, 1978. Shields, S. A. Functionalism, Darwinism, and the psychology of women: A study in social myth. Americgn Psychologist, 1975..§Q. 759-754. Siegel, A. W. The externalizing of cognitive maps by children and adults. In L. Liben, A. Patterson & N. Newcombe (Eds.) Spatial rspresentgtion gnd behgviorggcrosggthe lifeigpan: Theory sag a lication,l67-194. New York: Academic Press, 1981. New York: Academic Press, 1981. Siegel, A. & Schadler, M. The development of young children's spatial representations of their classrooms. Child DeveIOpmgnt, Siegel, A. & Schadler, M. Young children's cognitive maps of their classroom. Child Develo ment, 1977, 58, 386-394. Siegel, A. & White, S. The development of spatial representation of large scale environments. In H. Reese (Ed.) Advances in child development gnd behgvior, Vol. 10. New York: Academic Press, 1975. 58 Subkoviak, M. J. The use of multidimensional scaling in educational research. Review of Education Research, 1975, 45, 387-423. Thomson, E. G., Mann, I. T. & Harris, L. J. Relationships among cog- nitive complexity, sex, and spatial task performance in college students. British Journal of Ps cholo , 1981, 12, 249-256. Thorndyke, P. & Stasz, C. Individual differences in procedures for knowledge acquisition from maps. Cognitive Ppychology, 1980, Underwood, J. D. M. Skilled map interpretation and visual-spatial ablity. Journal of Geography, 1981, §Q. 55-58. Vandenberg, S. & Kuse, A. Mental rotations: A group test of three dimensional spatial visualization. Perceptual and Motor Skills, 1978. Al, 599-604. Witkin, H. A. & Berry, J. W. Psychological differentiation in cross- cultural perspective. Journal of Cross Cultural Psychology, 1975, .6, 4-87. Witkin, H., Oltman, P., Raskin, E. & Karp, s. A Manual for the Embedded Figures Test. Consulting Psychologists Press, Inc., Palo Alto, CA. 1971. Witkin, H., Goodenough, D. & Karp, S. Stability of cognitive style from childhood to young adulthood. Journal of Personality and Sogigl Psychology, 1967, 1, 291-300. Witkin, H., Lewis, H., Hertzmer, M., Machover, K., Meissner, P. & Wapner, S. Personality through perception. New York: Harper, 1954. Wittig, M. A. & Petersen, A. C. (Eds.) Sex-related differencepfiin cogni- tive functionipg: Developmental issues. New York: Academic Press, 1979. APPENDIX A BMT Men Women MRT Men Women H20 Men Women DAT Men Women MAP Men women TENBEST "9" Women GOODLM “w Women POORLM "°" Women BMT Men Women MRT Men Women H20 “9“ Women DAT Men Women MAP Men Women TENBEST "9“ Women GOODLM “‘m Women POORLM '4“ Women 23.. 22.. 28.. 32... a]... 38... 44". “A... -09 -02 -09 -03 06 03 05 -O7 GEFT 26.. 25... 32... 33... #3... 38... “8... “5.*' -06 -02 -07 OO 08 08 08 -07 GEFT 289.. 23.. 21.. 19.. ““99. no... -05 02 -17. OO 11 12 -1O 11 BMT 31... 25... 2a.. 20.. 46". 44"’ -03 04 -15 03 14 17* -06 12 BMT 33... so... 53... 53... -01 17* -O4 -06 19' 12 -08 13‘ MRT 36... 30... 56... 5h... 01 17.. -O3 -03 22}. 16‘ -03 13‘ MRT 59 TABLE 10 EARTIAL CORRELATIONS ON SPATIAL TASKS; CONTROLLING FOR Ag COMPOSITE “000' “O... -08 1O 02 05 05 13* -03 08 320 TABLE 11 PARTIAL CORRELATIONS ON SPATIAL TASKS: 41... 39... -O6 10 03 O6 07 15' OO 08 H20 01 14' -07 -04 25.. 16.. O1 10 DAT CONTROLLING FOR ACT ENGLISH # O4 16' -05 O1 29.. 23.. 08 1O DAT MEN N=107 WOMEN N=162 ' p (.05 1" p< .01 "* p< .001 14 OO 46... 32... “55" 47!!! MAP MEN WOMEN O .5 CG. 15 02 “7.“. 34§*§ “6.. 47... MAP 16' ~01 16. 33... on 31... TENBEST GOODLM N=107 N=162 P‘<005 P‘(o01 16‘ O2 17. 3h... 05 32... TENBEST GOODLM BMT "’n Women MRT Men Women 320 “°“ women DAT "9“ Women MAP Men Women TENBEST "e“ Women GOODLM "‘3 Women POORLM "'n Women BMT "°” Women MRT Men Women H20 Men Women DAT Men Women MAP Men Women TENBEST "'n Women GOODLM “'n Women POORLM "‘n Women 21.. 26... 28" 3h... 43... 36... 43... 46... ~10 O1 -06 OO 13 06 O7 -06 GEFT 27.. 309.1. 34... 39... L15!" h10.‘ so... 52... -05 06 -07 02 13 08 13 OO GEFT 6() TABLE 12 PARTIAL CORRELATIONS ON SPATIAL TASKS: CONTROLLING FOR ACT MATH MEN N=107 WOMEN N=162 27" * p<.05 24... .. P<001 22.. 36... ... p<.oo1 .13.. 28... 42... 52... 41... 42... 53... 37... -06 ~02 ~07 ~01 04 18" 09 15* ~15 ~02 04 ~05 15' 01 ~Ou 05 ~02 02 14 24.. 11 31... 47... 17. 1h 13‘ 12 18 32*'* O1 -10 -06 01 01 45... 18. 38... 12 13' 07 1O 47... o4 31... BMT MRT 820 DAT MA? TENBEST GOODLM TABLE 13 PARTIAL CORRELATIONS 0N SPATIAL TASKS: CONTROLLING FOR ACT SOCIAL STUDIES EN N=107 WOMEN N=162 32... . P<005 29... .. 134.01 25.. 37... "‘ p<.001 22.. 32... 47... 57... 42... 46... 58... 42... ~02 02 ~06 05 08 22.. 13. 22.. -15 -03 02 -06 1h 03 ~02 O7 02 03 15 24.. 09 31... 47... 14 16‘ 16' 16' 21.. 3u**‘ 01 -ou -01 02 1o 46*** 15 35««« 15. 17. 1O 16. 48... O6 33... BMT MRT B20 DAT MAP TENBEST GOODLM 61 TABLE 114» PARTIAL CORRELATIONS ON SPATIAL TASKS: CONTROLLING FOR ACT NATURAL SQIENQE Women 26‘*' WOMEN N=162 MRT "an 26w». 274w . P < .05 Women 36"” 26*” ... p < .01 320 Men noun 20‘ 32." "' p < .001 Women 1+1". 22M 32." DAT Hen hS‘” 16‘" 51*" 37*" Women 49“. ggeee 56"“ #2... MAP Men -06 -03 00 -O7 03 Women 02 Oh 19" 12 18" TENBEST Men -06 '15 -02 0‘} -05 1 5 Women -02 OO -05 O6 -O2 O1 GOODLM ”e” 05 11 19* 06 26H :43". 17. Women 06 1h 14‘ 15* 20" 33"* -O1 POORLM "an 08 -08 “'06 00 05 (+61»!!- 18. 36§u§ Women '03 13 15* 1Q ‘ 11+. 1+8." 05 32w». GEFT BMT MRT HBO DAT MAP TE?! BE ST GOO DLP-i APPENDIX B 62 TABLE 15 COMPARISONS OF MEN & WOMEN'S CORRELATIONS USING FISCHER'S Z CONTROLLING FOR ACT COMPOSITE ' p (.05 " p<.01 m .08 p (.001 Negative values indicate higher positive MRT -.23 .43 coefficients for women. Positive values indicate higher positive coefficients for men. H20 .24 .18 .33 DAT 0 .35 .05 --39 MAP .59 .56 -1.h7 1.h6 -1.1O TENBEST .91 -1.37 -.46 -.23 .28 1.13 GOODLM .32 -.09' .65 -.65 .71 1.55 1.33 POORLM .93 -1.72 -1.65 -.87 -.69 -.22 .96 .20 GEFT BMT MRT H20 DAT MAP TENBEST GOODLM TABLE 16 COMPARISONS OF MEN & WOMEN'S CORRELATIONS USING FISCHER'S Z CONTROLLING FOR ACT ENGLISH ' p<.05 " p<.01 BMT 006 .’. p<.001 Negative values indicate higher positive MRT -.O9 .52 coefficients for women. Positive values indicate higher positive coefficients for men. H20 .A6 .35 .59 DAT .29 .25 .29 .15 MAP .36 .52 -1.26 -1.28 -.91 TENBEST .58 1.43 -.02 -.25 .48 1.02 GOODLM -.02 -.26 ' .54 -.63 .52 1.30 1.13 POORLM 1.17 1.45 -1.25 -.60 -.21 -.07 .95 .55 GEFT BMT MRT H20 DAT MAP TENBEST GOODLM 63 TABLE 17 COMPARISONS OF MEN & WOMEN'S CORRELATIONS USING FISCHER'S Z CONTROLLING FOR ACT MATH ' p«<.05 .. p<001 -, 3 BMT 3 ... p<.001 “RT -.52 .24 Negative values indicate higher positive “ . coefficients for women. Positive values indicate higher positive N20 .09 .AO .70 coefficients for men. DAT '03} O ’006 037 MP .65 -075 -1058 -102“ '1030 TENBEST -.51 -1.10 -.20 -.O6 .23 1.07 GOODLM .07 O 010 -011 1011 1039 1033 POORLM .95 -1075 '1056 -0195 “068 '025 1.10 .62 GEFT ENE MRT H20 DAT MAP TENBEST GOODLM TABLE 18 COHPARISONS OF MEN & WOMEN'S CORPBLATIONS USING FISCHER'S z CONTROLLING FOR ACT SOCIAL STUDIES * p<.05 ” p<.01 BMT "21 0" p<.oo1 Negative values indicate higher positive MRT -.g7 .25 coefficients for women. Positive values indicate higher positive coefficients for men. H20 .39 .21 .u1 DAT “.20 007 -312 “.02 MAP ’087 079 -1063 -1051 -1037 TENBEST 067 ”10,48 .11 .0160 -060 081 GOODLM 0‘4“ -002' 070 -0514 082 1017 1.05 POORLM 1008 -1052 10,40 .57 -.LbLt -031 .73 .21 GEFT BMT MRT H20 DAT MAP TENBEST GOODLM 64 TABLE 19 COMPARISONS OF MEN & WOMEN'S;CO?RELATIONS USING FISCHER'S Z CONTROLLING FOR ACT NATURAL SCIENCE " p<-05 " p<.O1 BMT -.37 ’“ p<.OO1 . Negative values indicate higher positive MRT -.75 .10 coefficients for women. Positive values indicate higher positive coefficients for men. H20 -.07 -.13 -.OS DAT -.57 -.1O -.h8 -.56 MAP -.37 -.O7 -1.h8 1.h8 -1.20 TENBEST .33 -1.20 -.22 -.17 .27 1.11 GOODLM .15 -.2h ‘ .41 -.79 .A5 1.h0 1.h5 POORLM .13 -1.71 -.76 -.83 -.73 -.17 1.06 .32 GEFT BMT MRT H20 DAT MAP TENBEST GOODLM "11111111111111117111“