SELF‘CONGEPT 0F ACADEMICABILUYEOE; '« } _i _ WEST GERMAN EIGHTH~GRADE STUDENTS; '_f;' ‘ Thesis for: the Degree ofPhQD.‘ V . a MlCHIGANSTA-TE UNIVERSITY 2 ’ HOWARD. JOSEPHMICHAEL AUER, Jr; , 1~197'1“ LIBRARY \\\\ 1‘ Michigan 5m University J This is topertify that the thesis entitled SELF-CONCEPT OF ACADEMIC ABILITY OF WEST GERMAN EIGHTH-GRADE STUDENTS presented by Howard Joseph Michael Auer, Jr. has been accepted towards fulfillment _ of the requirements for Ph .1). degree in Education Major professor Date March 1, 1971 0-7639 ABSTRACT SELF-CONCEPT OF ACADEMIC ABILITY OF WEST GERMAN EIGHTH-GRADE STUDENTS BY Howard Joseph Michael Auer, Jr. This study is designed to examine the differences in self-concept of academic ability among eighth-grade students in West German Gymnasien, Mittelschulen, and Volksschulen. While it cannot demonstrate causal rela- tionships, it can point out differences among students of each of these three branches which can be used as bases for inferences and for future studies. The study was conducted in Giessen, a city of 60,000 in Hesse, West Germany. At the time of the study, the Gymnasien enrolled 463 students, the Mittelschulen, 247 students, and the Volksschulen, 321 students; 1031 were included in the study. Usable data were collected from 412 gymnasium students, 169 Mittelschule students, and 175 Volksschule students. Attitudinal data gathered from each student included responses to the Michigan State University Self-Concept of Academic Ability Scale, Perceived Parent Evaluation Scale, and Perceived Teacher Evaluation Scale, Howard Joseph Michael Auer, Jr. each of which was translated into German. Additionally, each student's school grades in social studies, mathematics, science, and German for the preceding year were obtained, as was parental occupation, which was used as the index of socio—economic status. All data were collected in each eighth-grade classroom by classroom administration of the attitude questionnaires and the cover sheet. Analysis of variance and Scheffé post-hoc compari- sons revealed an inconsistent pattern of Gymnasium students having superior self-concepts of academic ability and perceived evaluations, with Volksschule students having inferior self-concepts of academic ability and perceived evaluations; while Gymnasium data were consistently superior and Volksschule data were consistently inferior, the differences did not consistently exceed the level of significance. Multiple regression analysis indicated that branch attended by a student was a poor predictor of self-concept of academic ability and that branch did not appreciably improve the predictive power of other combinations. Perceived evaluations were the most potent predictors of self-concept of academic ability; this, however, may be due to statistical artifact. SELF-CONCEPT OF ACADEMIC ABILITY OF WEST GERMAN EIGHTH-GRADE STUDENTS BY Howard Joseph Michael Auer, Jr. A THESIS Submitted to Michigan State University in partial fulfillment of the requirements for the degree of DOCTOR OF PHILOSOPHY Department of Secondary Education and Curriculum 1971 07/ 7 5’ 7 TABLE OF CONTENTS LIST OF TABLES . . . . . . . LIST OF APPENDICES . . . . . . . Chapter I. II. III. PROBLEM . O O C O O Q 0 Need . . . . . . . . Purpose of the Study . . . Hypotheses and Questions . . ‘Theory . . . . . . . . Overview . . . . . . . REVIEW OF LITERATURE . . The West German Schools Since Selection . . . . . . Volksschule . . . . . Mittelschule . . . . . Gymnasium . . . . . . Self-Concept . . . . . . Theory . . . . . . . Definition . . . Accuracy of Self— -Concept . Correlates of Self- -Concept Change in Self— -Concept . Ability Grouping . . . Grouping and Student Social Teacher Effect . . . . Summary . . . . . . . DESIGN OF THE STUDY . . . . Site . . . . . . . . Sample . . . . . . . . Missing Data . . . . . . Major Variables and Instrumentation Self— -Concept . . . . Reliability of SCA . . . Perceived Evaluations . . ii Page vii H OGDWUJH 11 l3 l4 14 15 l7 19 21 27 29 32 4O 45 46 48 50 50 50 51 52 52 52 52 Chapter Page Reliability of PPEv and PTEV . . . . . 53 Grade-Point Average . . . . . . . . 53 Socio-Economic Status . . . . . . . 54 Design . . . . . . . . . . . . . 55 Testable Hypotheses . . . . . . . . 55 SCA . . . . . . . . . . . . . 55 PPEv . . . . . . . . . . . . . 56 PTEV . . . . . . . . . . . . . S6 SES . . . . . . . . . . . . . 56 GPA . . . . . . . . . . . . . 57 SCA-PPEV . . . . . . . . . . . 57 SCA-PTEV . . . . . . . . . . . 57 Analysis . . . . . . . . . . . . 57 Limitations . . . . . . . . . . . 63 IV. FINDINGS . . . . . . . . . . . . . 67 SCA . . . . . . . . . .~ . . . . 67 PPEV . . . . . . . . . . . . . 70 PTEv . . . . . . . . . . . . . 72 SES . . . . . . . . . . . . . . 74 GPA . . . . . . . . . . . . . . 75 Family SES, SCA, PPEV, PTEV, GPA Correlations . . . . . . . . . . 77 SES . . . . . . . . . . . . . 78 SCA . . . . . . . . . . . . . 81 PPEv . . . . . . . . . . . . . 82 PTEv . . . . . . . . . . . . . 83 SCA As An Intervening Variable . . . . 84 Correlations by Branch . . . . . . . 85 SES . . . . . . . . . . . . . 85 SCA . . . . . . . . . . . . . 85 PPEv . . . . . . . . . . . . . 86 PTEV . . . . . . . . . . . . . 86 Comparisons Between Branches . . . . . 86 Partial Correlations Within Branches . . . 88 Branch Attended . . . . . . . . . . 90 Predicting SCA from GPA, SES, Branch, PPEV, and PTEV . . . . . . . . . . . 92 Predicting SCA from GPA, SES, PPEV, PTEv, within Each Branch . . . . . . . . 96 V. SUMMARY AND CONCLUSION . . . . . . . . 100 Introduction . . . . . . . . . . . 100 Summary . 101 SCA . . . . . . . . . . . . . 101 PPEV . . . . . . . . . . . . . 101 PTEV . . . . . . . . . . . . . 102 iii Chapter SES . . . . . . . . . . . . GPA . . . . . . . . . . . Discussion of Major Hypotheses . Correlational Analyses . . . . . . Differences in Correlations Between Branches . . . . . . . . . Regression Analysis . . . . Differences in Regression Coefficients Between Branches . . . . . . . Conclusions . . . . . Suggestions for Further Study . . . . Implications for Education . . . . . BIBLIOGRAPHY . . . . . . . . . . . . APPENDICES iv Page 102 102 102 103 104 105 105 106 107 108 110 121 10. 11. 12. LIST OF TABLES ANOVA: SCA by branch . . . . . . . . . ANOVA: PPEv by branch . . . . . . . . ANOVA: PTEv by branch . . . . . . . . ANOVA: SES by branch . . . . . . . . . ANOVA: GPA by branch . . . . . . . . . Zero-order correlations between family SES, SCA, PPEV, PTEv, and GPA for all students . Correlation coefficients between SCA, GPA, PPEV, and PTEv with SES partialed out for all students . . . . . . . . . . . Correlation coefficients between SES, GPA, PPEV, and PTEv with SCA partialed out for all students . . . . . . . . . . . . Correlation coefficients between SCA, GPA, PTEv, and SES with PPEv partialed out for all students . . . . . . . . . . . Correlation coefficients between SCA, GPA, PPEV, and SES with PTEv partialed out for all students . . . . . . . . . . . Correlation coefficients between SES, SCA, PPEV, PTEv, and GPA for Gymnasium, Mittelschule, and Volksschule samples . . . . . . . Correlation coefficients between SES, SCA, PPEV, PTEv, and GPA for Gymnasium, Mittel- schule, and Volksschule sampIes and partial correlation coefficients between SES, SCA, PPEV, PTEv, and GPA with each variable partialed out for Gymnasium, Mittelschule, and Volksschule samples . . . . . . . Page 68 71 73 74 76 78 81 82 83 84 87 89 Table Page 13. Kendall's Tau between SCA, PPEV, PTEV, GPA, and branch and partial correlations between SCA, PPEV, PTEV, and GPA with branch partialed out . . . . . . . . . . . 91 14. Multiple regression coefficients: using GPA, SES, Branch, PPEV, PTEv to predict SCA within total sample . . . . . . . . . 93 15. Multiple regression coefficients: using GPA, SES, PPEV, PTEV, to predict SCA within each branch . . . . . . . . . . . . . 96 vi LIST OF APPENDICES APPENDIX Page A. Self-Concept of Academid Ability Scale in German . . . . . . . . . . . 122 B. Self-Concept of Ability--Genera1 (Form A) . 125 C. Perceived Parental Evaluation Scale in German . . . . . . . . . . . 128 D. Perceived Parental Evaluation Scale in English . . . . . . . . . . . 131 E. Perceived Teacher Evaluation Scale in German . . . . . . . . . . . 134 F. Perceived Teacher Evaluation Scale in English . . . . . . . . . . . 137 G. Scheffé Post-Hoc Contrasts . . . . . . 140 H. Standard Deviations of A11 Variables for Total Sample and for Each Branch . . . . 143 I. Means of All Variables for Total Sample and for Each Branch . . . . . . . . 145 vii CHAPTER I PROBLEM Need The manifest purpose of schooling is to help students learn. While what is learned is a function of cultural desirability and may therefore vary from school to school or from nation to nation, schools are nevertheless ostensibly organized to maximize the likelihood of students' acquiring certain modes of behavior. It is therefore impera- tive that schools be aware of any practices which might depress or impede student learning. Some school practices may depress student learning by depressing students' perceptions of their ability to learn; students are more likely to attempt tasks that they believe will result in success than those they believe will result in failure, and they are more likely to succeed at tasks at which they believe they can succeed. Students may be falsely labeled as unable to perform at higher levels when in fact they are able to do so. The successes of minority group students in American colleges indicate that groups of persons previously deemed unable to succeed at academic tasks, if subjected to expectations of success and exposed to others who succeed at those tasks, can in 1 fact perform quite satisfactorily in areas previously deemed inappr0priate to the academic abilities of these students. Similarly, the United States Army has success— fully provided recruits identified as poor school achievers with reading and computational skills (75); these recruits completed advanced training with as much success as normal recruits, thereby casting some doubt upon the validity of their supposed inability to learn. Labeling students as unable to succeed academically can initiate a self-fulfilling prOphecy (82) which becomes reified when students apply it to themselves; if students are identified as possessing low ability to learn and are treated as low ability learners, they are quite likely to see themselves as able to perform as high-ability learners. If, however, they see themselves as being expected to per- form at higher levels, they are more likely to do so than if they see themselves as being expected to perform at low levels (3;28). Using this perspective, it can be seen that such practices of school segregation as ability grouping and tracking become traps for children denied initial access to the higher curricular strata. These children take on the false labels of low achievers thrust upon them by the schools and so validate what may be an invalid premise-- that ability to learn is fundamentally a static phenomenon which is impervious to change. Given the goal of educating all students to the limits of their potential, it follows that practices which place artificial limits upon potential are dysfunctional in at least two respects. First, by limiting students' lati- tude of choice, inhibitory practices diminish human liberties. Second, Operating in an economy which requires large numbers of highly trained personnel, school practices which create false notions of inability to learn to perform certain tasks are in conflict with the needs of that economy. It is imperative, therefore, that the relation- ship between students' perceptions of their abilities as learners and school segregation practices be examined. Purpose of the StuGy . The study is designed to examine the differences in self-concept of academic ability among eighth-grade students in three types of schools in West Germany. The West German educational system was selected because its students are formally segregated by academic achievement upon completion of the fourth grade, and this formal segregation provides more nearly discrete groups than do American segregation patterns in which students of different academic strata may share teachers, classes, or extra-curricular activities. Where the West German branching system provides three fairly autonomous branches, each with its own staff, curriculum, grading criteria, and buildings, the American system does not. It is possible for an American student in a college .preparatory track to have extensive in-school contact with students in other tracks, and this contact complicates the study of the correlates of academic segregation. This study will compare the students of the three branches of the West German school system in an attempt to gain some insight into the nature of possible differences in student perceptions of academic ability. The study is essentially a cross-cultural replica- tion of previous work done by Brookover and associates in the United States (109;110;lll), and so it will use the same variables employed in these earlier studies to see if the earlier relationships found among these variables hold true for those among West German students. Therefore, this study will examine differences in students' perceptions of the evaluations of their academic ability held by parents and teachers, and it will examine the relationships among self-concept of academic ability, perceived evaluations, socio-economic status, academic achievement, and branch attended by students. Finally, the study will employ per- ceived evaluations, academic achievement, and branch to attempt to predict self-concept of academic ability. It should be noted that the study is not designed to prove the cause of any difference among students in the three branches of the West German schools. Rather, it is designed to identify the extent of any such attitudinal differences. It is an exploratory study intended to provide a basis for further quasi-experimental and experimental inquiry. Hypotheses and Questions The fundamental purpose of the study is examination of differences in self-concept of academic ability among eighth-grade students in three types of schools in west Germany. Students are selected for these schools on the basis of their academic achievement in grades one through four, with the t0p twelve per cent entering the Gymnasium, the middle eight per cent entering the Mittelschule, and the remaining eighty per cent entering the Volksschule. Thus, Gymnasium students represent the most successful students in the first four years of the system; further, they are the only students generally admitted into higher education. These two factors, combined with the generally higher status afforded Gymnasium students, might induce Gymnasium students to form and maintain higher self-concepts of academic ability than students in the other two branches. Similarly, Volksschule students are identified as the lowest achievers, and they usually enter jobs requiring little training which confer little status upon the worker. These factors might Operate upon the Volksschule student and reinforce a low self-concept of academic ability. Thus, The relationship between branch and student self-concept Of academic ability is direct; the higher the branch, the higher the self-concept Of academic ability. Students do not Operate in social vacuums. They are aware that their parents and teachers hold evaluations Of their academic performance, and the German students are especially aware Of those evaluations because the students are assigned to branches after parents and teachers have discussed assignment. Branch assignment thus provides one bit Of evidence of the evaluations held by parents and teachers. Further, students might perceive parents and teachers as expecting them to perform at a level appro- priate to their branch, and since the most able students enter the Gymnasium and the least able to into the Volks- schule, it might be hypothesized that H2: The relationship between branch and student perception of parental evaluation is direct; the higher the branch, the higher the perceived parental evaluation. H3: The relationship between branch and student perception Of teacher evaluation is direct; the higher the branch, the higher the perceived teacher evaluation. Not all students are educated for the same future. Gymnasium graduates, upon passing an examination, may enter the universities and prepare for tOp-level jobs, while Mittelschule graduates generally move into middle- level occupations, with Volksschule graduates occupying blue-collar jobs requiring little training. In this respect, the West German schools serve the same sorting and selecting function provided by American schools. Further, since the most accurate predictor Of future socio-economic status Of children is parental socio-economic status, and since children of Gymnasium graduates are more likely to have been exposed tO evaluations and experiences which encourage high achievement while children Of Volksschule graduates are less likely to have had these experiences, and since exposure to a set Of experiences and evaluations is likely to induce an appropriate level Of academic per- formance which will result in a student's being placed within a given branch, H4: The relationship between branch and family Socio-economic status is direct; the higher the branch, the higher the family socio-economic status. Students are assigned to a branch on the basis of academic achievement, and the curriculum Of that branch is ostensibly designed for the abilities and interests Of students attending that branch. That is, the Gymnasium curriculum is intended to be as difficult for Gymnasium students as the Mittelschule and Volksschule curricula are for Mittelschule and Volksschule students. If these intentions are realized, one might hypothesize that H5: There are no differences in academic achievement between the three branches. One Of the factors employed in evolving and main- taining self-concept of academic ability is perception Of the evaluations Of others; persons who perceive that others important to them hold high evaluations of their academic ability are more likely to hold high evaluations Of them- selves than are persons who perceive that others hold low evaluations Of the academic ability. Thus, H ° The relationShip between self-concept 6 Of academic ability and perceived parental evaluations is positive. The relationship between self-concept of academic ability and perceived teacher evaluations is positive. The relationships examined in these seven hypotheses, however, are Of little predictive utility, so the contribue tions of perceived evaluations, socio-economic status, and academic achievement tO prediction Of self-concept Of academic ability will be examined. Student self-concept of academic ability depends to some extent upon each of the former variables; the question is Q1: What is the most economical combination Of variables for prediction Of student self-concept of academic ability? Theory The theoretical work of the symbolic interactionist school Of social psychology, including W. I. Thomas, John Dewey, and George Herbert Mead, provides the background for this study. From this point of view, human behavior is the juncture Of the dynamic relationship between the human organism and its environment; through behavior, the organism modifies its environment, and perceiving the modifications, further alters its behavior tO accommodate the changes. This interaction is facilitated by the individual's symbolic representation Of the environment which makes possible his exchanging meanings and evaluations Of Objects with others in his environment. It is through these exchanges that the self emerges; the individual comes to see his self in terms of his perceptions of the expectations and evaluations others hold Of him. His self, then, is socially determined, and his behavior at any point in time is a function Of his past experiences and his perceptions Of the expectations which others hold Of him. As exPeriences and perceived expectations vary, so will behavior, and changes in behavior generate changes in experience and expectations. Thus, human behavior is a function Of the view one holds Of oneself, that self-concept being a function Of the expectations imputed by the person tO others. It follows that the imputed expectations can limit the range Of individual behavior. Examples Of these limitations are found in such behaviors as dress and language; American males simply do not wear grass skirts to the Office, nor do they generally speak Swahili. Further, one is unlikely to attempt a behavior if it is not evoked by his self- concept. A person who views himself as polite would be unlikely to behave crudely; similarly, a student who views himself as unable tO excel academically might find it 10 difficult tO excel (3:469;49). Self-concept, then, limits the range Of behaviors Of the individual. Overview This study is designed to examine the correlates of branch membership in the West German schools. In Chapter II, relevant literature will be reviewed in three main areas, the organization of West German schools, correlates of various schemes of ability grouping, and the influence of the social environment upon self-concept and behavior. In Chapter III, the design Of the study will be explained; characteristics of the sample will be presented, the instrument will be described, and the hypotheses elaborated upon. In addition, the modes Of analysis to be employed will be explicated and justified. In Chapter IV, the research findings will be presented and examined. CHAPTER II REVIEW OF LITERATURE In this chapter literature concerning the organiza- tion of the West German schools since 1945, self-concept theory, self-concept accuracy, self-concept change, behav- ioral correlates Of self-concept, and various schemes Of academic segregation will be reviewed. The first section of the chapter is an analysis Of the organization of the West German schools since 1945. The West German Schools Since 1945 The organization of the West German schools has remained essentially unchanged since the issuance Of Allied Control Council Directive 54, June 25, 1947, which attempted to standardize the educational programs Of the several Lander (states) by establishing nationwide com- pulsory education tO age 18; full-time attendance was required from ages 6 to 15, and part-time attendance was required from ages 16 to 18. Despite this directive, there is still some variation in school organization among the. Lander. For example, some require primary education to age 14, while others require attendance tO age 15 (19:75; 86:20). However, the general scheme, which is that 11 12 employed in Hesse, the Land within which Giessen, the site of this study, is located, is to require full-time primary education to age 15. In general, West German education is organized along the lines Of this outline (36:29): Grundsschule Grades 1-4 Ages 6-9 Gymna51um _ - Mittelschule ‘ _ _VOlksschu1e Grades 5:13 Grades 5-10 Grades 5-10 Ages 10-18 Ages 10-15 Ages 10-15 Universitat Eachschule Beru sschule Full-time Part-time to age 18 to age 18 The striking difference between the West German schools and our own is the West German practice Of formally segregating students after the fourth grade. While many American schools provide homogeneous ability grouping, few systems have established distinct sub-systems for various ability groups. It is, Of course, this aspect Of West German schOOl organization which makes it especially relevant to a study Of variations in student self-concept Of academic ability, for while nearly all West German children attend the common Grundsschule (foundational school), they are sorted after the fourth grade into a branch which is unique from the others in terms Of 13 curriculum, Objectives, grading criteria, and teacher qualifications. One Of the consequences of the branching scheme, then, is complete separation Of children in the three branches after a common four-year primary education, which provides an Opportunity for comparison Of character- istics Of children in each branch; since separation is complete, contamination is minimized. Selection Children are selected for one Of the three branches during their fourth grade year by their teachers, with their fourth grade teacher holding the greatest influence. Teachers consider the childrens' academic achievement and the likelihood Of their fitting into a given branch; the teachers then consult with the childrens' parents, who can overrule the teachers' decisions. This is rarely done. In addition to academic achievement and perceived likelihood Of success, some Lander require candidates to pass entrance examinations (19:78). Hesse does not require these examinations. It should be noted that these are not intelligence tests; the West German schools have resisted adopting them. Trial classes are held for borderline children who might attend either Of two branches. These classes, lasting one or two weeks, provide teachers additional 14 Opportunities tO Observe children in simulated environments and tO make what is hopefully a more appropriate decision. Volksschule The Volksschule (common school) is an extension Of Grundsschule and the eventual destination Of 80 per cent Of the Grundsschule students. In Hesse, it enrolls children between the ages of 10 and 15 in grades 5 through 10 in a curriculum which includes German, arithmetic, geometry, history, geography, social studies, nature study, drawing, music, physical education, wood working for boys, and needlework for girls (36:28). Emphasis is placed upon practical and manual activities, and Volksschule leavers generally enter the Bergfsschule (occupational school) for part-time occupational education until age 18 (16:32). Mittelschule The Mittelschule (middle school) typically enrolls eight per cent Of the 10 to 15 year-Olds in grades 5 through 10. The curriculum, consisting Of German, social studies, natural sciences, and mathematics, also includes foreign languages (36:30) and is intended tO prepare students for middle-range positions in business and industry (16:53-54). Upon successful completion Of the course Of study, students generally attend the Fachschule (business school) full-time to age 18, after which the more successful may enter the Technische Hochschule for further study. il“ll‘l“l. 15 Gymnasium The best-performing 12 per cent Of the fourth grades enter the Gymnasium for a course of study which continues full-time until grade 13. The Gymnasien are Of three types, either classical, modern language, or mathematics-natural science, and each presents a unique curriculum (36:30). Nevertheless, each type Gymnasium prepares its students for matriculation at a university; only by passing the Abitur can students enter universities. Thus, students who attend either the Mittelschule or the Volksschule are effectively barred from university education, since the Abitur is Offered only to Gymnasium graduates. While all teachers must pass the Abitur, Volksschule teachers attend a Padogogische Akademie for three years before teaching; Mittelschule and Gymnasium teachers, by contrast, must study four to six years at a university. Thus, Volksschule teachers may appear to have inferior training. In general, only the Gymnasium is viewed by the people as giving a "genuine" education, the other two branches being somewhat diluted. Parents and employers tend to require completion Of Mittelschule or Gymnasium for their children and employees (21:170), and some critics suggest that the branching scheme maintains social privilege (16:146). (I? II ll'llll' '[‘U.l [I’ll vl‘l’ll lli 16 Another problem is wastage. Some sources charge that as many as 10 per cent of the branch selections are inapprOpriate (68;115), a criticism lent credance by the reversion of Mittelschule students to the Volksschule and Gymnasium students to the Mittelschule after the first year in the branch (19:78). Further, about one-half the students who enter the Gymnasium will either drop back to the Mittelschule or fail the Abitur (20:88). The ultimate result of this wastage is that only six per cent Of West German youth enter higher education in comparison with over 35 per cent in the United States. The West German educational system may be under-educating West German youth. In an attempt tO provide greater Opportunity for students, the schools provide lateral transfers. For example, high-achieving Volksschule graduates may enter the Mittelschule, enter the third year Of a Gymnasium, or attend the Kurzform (short form), a six-year Gymnasium. However, the lock-step curriculum Of the Volksschule makes transfers very difficult, and they are rarely effected (19:80). A similar situation exists with Mittelschule students (16:54). A further problem is the possibility Of students' adOpting artificially deflated views Of their ability to succeed academically as a function Of the branch they attend. Thus, Volksschule students might hold 17 unrealistically low estimates Of their ability to succeed in the more academic Mittelschule or Gymnasium because they aren't expected to; rather, they are eXpected to succeed in a relatively "practical" curriculum. This study investi- gates the extent tO which these estimates of ability vary among the three branches. In order to gain an understand- ing of the phenomenon Of self-evaluation, theories and research relating to self-concept will be examined in the next section Of this chapter. Self-Concept We all use some kind Of self-concept in daily interaction with each other and within ourselves. Respond- ing to a simple ritualistic question such as "How are you?" may require fleeting reference to you by you and is there- fore impossible in the absence of thoughts about the self. Self-reflection is necessary for one to be able to say, "I am not myself today;" clearly, one must have a conception Of what that self usually is in order tO ascertain its not being that. However, use Of a term is not synonymous with agreement upon meaning Of that term. Thus, self-concept may be viewed as being real, imaginary, inferred, con- sciously referred to, unconsciously referred to, or some- thing revealed only through behavior. Probably the only safe statement to make about usage Of the term is that it is a verbal symbol associated with a construct. This does 18 not mean that the construct exists or can be defined so as to have Operational validity, but rather that we have invented a symbol which we associate with that construct and then proceed as though that construct had Operational or empirical validity (37:18-20). A rather drastic convention is prOposed by MacCorquOdale and Meehl, who define hypothetical constructs as "terms which are not wholly reducible to empirical terms; they refer tO processes that are not directly Observed (although they need not be in principle unobserv- able)" (79:104). They suggest that an intervening variable be defined as "a quantity Obtained by a specified manipula- tion of the values Of empirical variables" which "will involve no hypothesis as to the existence Of nonobserved entities or the occurrence of unobserved processes" (79: 103). This convention, however, seems prohibitively stringent and would in effect reduce much social- psychological research tO mere conjecture, since so much Of that research involves making inferences about unob- served phenomena. A less restrictive convention is that suggested by Marx, who defines an intervening variable as any inter— vening construct with a maximum amount of Operational validity and hypothetical construct as any construct with a relatively low degree Of Operational validity (81:236). The net impact Of these two prOposals upon this researcher, 19 then, might be to strive for a high degree of Operational validity while remembering that the data describe not the variables but the observable correlates of those variables and that any inferences are, of necessity, frail. Given these limitations, the notion Of self-concept will be developed. Theory ' . The work of C. H. Cooley (6) and George Herbert Mead (23;33) provides the theoretical framework within which this study is located. Kinch (73) has suggested a formalized theory of the self-concept which, with one modi- fication, accurately describes what Cooley and Mead devised. Kinch defines the self-concept as "that organiza- tion of qualities that the individual attributes to him- self," and develops this into the general theory that "the individual's conception of himself emerges from social interaction and, in turn, guides or influences the behavior of that individual" (73:481). The theory and the definition yield four prOposi- tions. First, self-concept is a reflexive phenomenon; that is, a person can attribute qualities--either roles or adjectives-~to himself. This is analogous to Mead's notion that the self can be an Object as well as a subject, that one can reflect upon himself (23:138). Second, self-concept emerges from social experience. Third, a synthesis Of ;propositions one and two, one's self-concept is a function 20 of his perception of how others perceive him. Cooley suggests a looking-glass self. As we see our face, figure, and dress in the glass . . . and [are] pleased or otherwise in them . . . as in imagination we perceive in another's mind some thought of our appearance, manners, aims, deeds, character, friends, and so on, and are variously affected by it . . . (6:184). Similarly, Mead suggests The organization of the self is simply the organization, by the individual organism, of the set of attitudes toward its social environ- ment--and toward itself from the standpoint of that environment . . . which it is able to take (23:91). In other words, one's self-concept is at least partially a function of how he imagines those persons he interacts with evaluate him; research evidence, to be reviewed later in this study, indicates that a change in perceived evalua— tions of others is accompanied by a change in self-concept. The third proposition is that the self-concept guides or influences behavior. Mead suggests that the attitude of the actor is part of the act in that it determines which behaviors he will present (23:11). The perceived evaluations of others, when internalized by the actor, serve as a censor upon his behavior; this is Mead's "me," a store of experience (23:175,210). As "me" changes, so does the nature of the censor, and so does the range of possible behavior of the actor. 21 Kinch presents his propositions in this model: Perceived response of others Actual reSponse Se1€;concept Of others , '\ Behavior ‘6 of self’ I The dotted line was added by this writer to indicate that the actor takes note of his own behavior and may use it to maintain or to modify his self-concept. The salient points of self-concept as employed in this study are: 1. Self-concept is seen as emerging through social interaction. 2. Self-concept is seen as a reflexive phenomenon. ‘ 3. Self-concept is seen as guiding behavior. These points will be used in examining various conceptions of self presented by several authors. Definition Self-concept appears in the literature as self- esteem, as central self-perceptions, as all self-perceptions, and as self-perceptions compared to others. COOpersmith, writing on self-esteem, agrees that the self-concept develOps through social experience (7:20), but chosh to concentrate upon self-esteem in his study. Self-esteem, "a personal judgment Of worthiness that is expressed in the attitudes 22 the individual holds toward himself" (7:4-5), seems too diffuse to have much economy in predicting behavior; while COOpersmith found a correlation of .30 between his self- esteem inventory and academic achievement among 85 fifth, sixth, and seventh grade children (7:129), the instrument used in this study consistently yields correlations with academic achievement in excess of .50. While Lowther (76) can argue that self-esteem is a function of all activities of a child and that it therefore reflects academic activities, he fails to realize that a measure which concentrates upon school activities may be more economical. Trowbridge (96), assessing correlates of socio-economic status with self-concept, used COOpersmith's Self Esteem Inventory, failing to realize that she might be measuring but one aspect of self-concept. Garvey (65), studying self-concept and success in student teaching, used a global instrument tapping self- satisfaction, perception of physical self, Openness to criticism, perception of social self, and perception of religious self. Her instrument seems unnecessarily broad and does not tap individual perception of the evaluations of others. Williams and Cole (102), using the same instru- ment, use the term "self-concept" in their title, but employ self-esteem and self-concept interchangeably in the body of their article. 23 Adopting a different perspective, Combs and Snygg (5) define self-concept as the individual's most inportant self-perceptions (5:127) and emphasize the point of view of the individual (5:11). While they agree that the self is a social product, they concentrate upon the individual's entire perceptual field, thereby becoming vulnerable to a crippling Operational problem: how can the entire percep— tual field be measured accurately? As wylie (39:24) notes, the response of an individual to a measure may be a function not only of his perceptual field, but also of his selecting only some items from it for public disclosure, of his including items which are not part of his field, and by his unique use of language. Another perspective is suggested by Helper (67), who defines self-concept as "whatever symbolic responses are associated with the individual's identity symbols." While he does not relate his study to classroom learning, he does indicate that self-concept may be learned through social interaction. Similarly, Davidson and Lang (54), using an adjective checklist Of favorable and unfavorable words on 203 fourth, fifth, and sixth graders, did not con- centrate upon self-concept with reference to a specific task but instead employed a rather global measure, using only descriptive adjectives with no reference to social roles. Another example of an idiosyncratic use of self- concept is that Of Soares (94), who used a global instrument 24 which did not ask the individual to compare himself to others. In view of this, Soares' inferences about the reference groups of his subjects are puzzling. Roth (89) also employed a global measure, asking subjects to describe themselves using a Q-sort technique. However, his technique, like those of Soares and Davidson, allowed the subject to demonstrate agreement or disagreement with a descriptive term and did not consider the degree of agreement or disagreement. This characteristic may limit the predictive power of these instruments. Rosenberg (27:5) defined self-image as an attitude toward an Object, and employed attitude to refer to facts, Opinions, and values regarding the self. However, he also referred to a favorable or unfavorable attitude toward the self and thereby confounded self-concept with self-esteem. In contrast to the somewhat idiosyncratic approach of the previous authors, Brandt, Coleman, and Mann produced data which describe students' perceptions of their abilities in comparison to the abilities of others. Brandt (48) asked 139 sixth and eleventh graders to estimate whether they would do better or worse than each of their classmates on various intellectual and physical tasks. Coleman (112: 281) asked students how bright they thought they were in comparison with other students in their grade. However, he included student perception of teaching style in his self-concept battery and may have diluted its predictive 25 power. Mann (80), asking fifth grade students why they thought they had been placed in ability groups, purported to measure self-concept, although her instrument was phrased in such a way as to tap a multitude of responses. Nevertheless, studentsiJ1the tOp and bottom groups indi- cated that they were "smarter" or "not as smart" as other students, while students in the middle two ability groups gave mixed responses, which could be expected, given the ambiguous phrasing of the questions. Despite the flaws in these three studies, their conception of self-concept seems most consistent with that of Mead (23:91), whose work pro- vides the theoretical orientation of this study. Mead's theory is adapted by Brookover, who defines self-cOncept as "symbolic behavior in which the individual articulates a program of action for himself as an Object in relation to others" (109:8). Self-concept is thus not something inferred from behavior but is symbolic indication; this definition eliminates the problems of inference which many phenomenal theorists find so vexing (39:7-8,l9). Mead theorized that the self is faceted and that the social situation determines which facets are brought forth (23:142). Brookover adapted this by positing a self-concept Of academic ability, reasoning that a global self-concept might be too broad to have significant power to predict any one facet Of behavior. He therefore defines self-concept of academic ability as "behavior in which one indicates to himself 26 (publically or privately) his ability to achieve in academic tasks with others engaged in the same task" (109:8). It is this self-concept Of academic ability which is employed in this study. In an earlier work, Brookover presented four hypotheses of social psychology of learning: 1. 2. People learn to behave in ways that each considers appropriate to himself. ApprOpriateness Of behavior is defined by each person through the internalization of the expectations which others whom he considers important to him hold. The functional limits of one's ability to learn are determined by his self-conception or self-image as acquired in social inter- action. The individual learns what he believes others who are important to him eXpect him to learn in a given situation (3:34-35). Brookover, emphasizing the social-psychological aspects of learning, suggests that what and how much one learns are functions of what and how much one believes one can and ought to learn. He does not deny possible organic limita- tions upon learning ability but does suggest that a student will learn to the limits of his organically-defined limits only if his self-concept of academic ability is consistent with such a level of learning. "Functional limit" is the term used to emphasize that we are speaking not of genetic organic limits on learning but rather Of those perceptions Of what is apprOpriate, desirable, and possible for the individual to learn. We postulate the 27 latter as the limits that actually Operate, within broader organic limits, in determining the nature or extent of the particular behavior learned (3:469). Thus, a person with a high IQ but a low self-concept of academic ability usually has low academic achievement, while a person with a low IQ and a high self-concept of academic ability usually has low academic achievement, too (110); a high self-concept of academic ability is a necessary but not sufficient condition for high academic achievement. This finding suggests that a person is unlikely to actualize his potential unless he believes that he is able to succeed. The problem of changing that belief is dealt with below; a prior concern is the accuracy of that belief. Accuragy of Self-Concept One method Of assessing the accuracy of self-concept is to ask subjects to estimate their level of performance on a given task, have them perform the task, and compare the estimate with the performance. However, since self- concept is a function of past performance as well as an estimate of present ability, the accuracy of recall must be assessed. Froelich and Moser (63) gave the Differential Aptitude Tests to 150 ninth-grade students and, after scoring the tests, explained the test profiles to each student and let him keep a profile. After 15 months, students were asked to draw their test profiles from memory. 28 Every sub-test score was reported higher by the group than the group real score. However, not every student increased his scores; a high percentage of high-scoring students remembered their scores as being lower than they actually were. The authors suggested that students compared test scores and were influenced by their perceptions of the group mean; both high and low groups remembered their scores as closer to the group mean than they actually were. Rather than study the accuracy of memory, Brandt (48) tested the accuracy of self-estimate, asking 139 students in five classrooms to estimate how well they expected to do in comparison with each classmate on tests of arithmetic, spelling, vocabulary, and assorted physical tasks. He found a fairly consistent tendency to overrate one's performance in comparison with the relationship between one's expected performance and classmates' expected performances. However, he made no attempt to correlate self-estimate with actual performance. Torrance (95) asked college freshmen how well they thought they would perform on tests of scholastic ability and aptitude; students were asked to estimate the quarter within which their scores would fall. Like Brandt, Torrance found a tendency for students to overestimate their performance; 65 per cent of the group placed themselves in the upper quarter, and 95 per cent of the group placed themselves in the upper half. Of those students actually in the lowest quarter, 62 per 29 cent had placed themselves in the top quarter and 92 per cent, in the upper half. These estimates, however, might have been inflated as a reaction to the anxiety connected with being a college freshman. Neither Froelich, Brandt, or Torrance tested the tendencies Of self-estimates and behavior to vary together. Thus, if estimates were systematically inflated, their correlations with behavior might remain unchanged. Similarly, if inflation were systematic, comparison of means would not be affected significantly. Correlates of Self-Concept Since the chief value of self-concept of academic ability is the extent to which it correlates with academic achievement, studies examining this relationship were reviewed. Lowther (76), although dealing with self-esteem, found that children with high academic achievement were twice as likely to have high scores on a COOpersmith Self- Esteem Inventory as were children with low academic achieve- ment, regardless of IQ; conversely, children with low academic achievement were twice as likely to have low scores on the COOpersmith instrument as were children with high academic achievement. However, since he did not find what percentage of children with high self-esteem had high or low academic achievement, it is difficult to draw any causal inferences from his study. 30 Trowbridge (96) used COOpersmith's inventory to compare self-esteem of disadvantaged elementary school children with self-esteem Of nondisadvantaged elementary school children. She found that disadvantaged children indicated higher self-esteem than nondisadvantaged children, but since she did not attempt to correlate self-esteem with academic achievement, her study is of limited utility. It does indicate, however, that disadvantaged groups may be beginning to utilize group members rather than all age- mates as a reference group. Her findings are consistent with those Of Soares (94) who found that disadvantaged students scored higher on his instrument than did advantaged students. Soares did not report any relationship between scores on his instrument and academic achievement, nor did he or Trowbridge attempt to describe the reference groups used by their students. COOpersmith (7:188), examining self-esteem, found a relationship between self-esteem and parental expressions of attention, concern, and value. Rosenberg (88) found a similar relationship, as did Helper (67). Each of these researchers, though using slightly different instruments, found a relationship between parental evaluations and self- evaluations of their children. Brookover (109:104) found strong relationships between childrens' self-concepts of academic ability and their perceptions of their parents evaluations of that ability. Brookover notes that 31 self-esteem does not correlate as highly with academic achievement as does self-concept of academic ability (109: 102); Rosenberg's (27) self-esteem measure produces a cor— relation Of .31 with academic achievement among 534 twelfth-grade students, while Brookover's self-concept Of academic ability scale produces a correlation of .49 with academic achievement. Furthermore, partialing the variance of self-esteem from the self-concept correlation depresses that correlation from .49 to .46, while partialing self- concept from the self-esteem correlation drOp that cor- relation from .31 to .06. Self-esteem apparently asso- ciates with academic achievement indirectly, through self-concept of academic ability. Williams and Cole (102) found a correlation of .31 between the Tennessee Self-Concept Scale and reading achievement measured by the California Test Of Mental Maturity. It is noteworthy that the Tennessee instrument has a greater range of scores than Brookover's scale, and the California Test has a greater range of scores than grade-point average. Despite the greater ranges of the variables involved in Williams' correlation, that correla- tion is lower than any of those of Brookover. Perhaps the Tennessee Self-Concept Scale is of less utility in pre- dicting achievement than the Brookover scale. Davidson and Lang (54) used an adjective checklist of favorable and unfavorable words on 203 fourth, fifth, 32 and sixth grade students. They found significant high correlations between self-concept and academic achievement and between self-concept and student perceptions of teacher feelings toward them. Their instrument was quite crude, tapping only the content of children's beliefs about them- selves; it did not measure the strength of those beliefs, nor did it directly probe children's beliefs about their academic ability. However, Davidson and Lang did relate academic achievement, a crude self-concept, and the imputed perceptions of teachers. In a series of studies involving seventh through twelfth-grade students, Brookover (110:51;lll:97) found correlations between self-concept of academic ability and grade-point average which ranged between .49 and .58, indicating a fairly strong relationship between the two variables. He also found that the correlations between self-concept Of academic ability and perceived evaluations of parents varied between .58 and .76 (109:104); the corre- lations between self-concept of academic ability and per- ceived evaluations of teachers varied between .59 and .63 (109:105). Change in Self-Concept If learning behavior, self-concept of academic ability, and perceived evaluations are functionally related, a change in one can be expected to produce a change in one or more of the others. This section of the review of 33 literature contains an examination of research in support of this proposition. Festinger (61) posits a drive to evaluate one's abilities. In the absence of objective physical standards, such as strength or speed, one may accept the evaluation of other persons, and his satisfaction of that evaluation varies with the extent of agreement among those persons. In other words, social reality becomes the measuring stick, and social reality can be as much in error as individual reality. Thus, a group assessment can be, if not erroneous, at least Open to change. Merton (8&195) defines the self-fulfilling prOphecy as "a false definition of the situation evoking new behavior which makes the originally false conception come true," and cites the example of examination neurosis, wherein a student, fearful of failing the examination, worries so much about failure that his studying is ineffective, and so he fails the examination. Festinger and Merton suggest that a false social reality can become real in its consequences; Rosenthal and Jacobsen (28), in a comprehensive review of the self-fulfilling prophecy, apply it to learning, arguing that children may place artificially low ceilings on their learning ability because their teachers erroneously believe them to be unable to learn. Students who believe that others perceive them to be unable to learn may accept that definition, regardless of whether or not they may be able 34 to learn. Atkinson (42) found that one's estimate Of success or failure in a risk-taking situation reflected the subject's perceived ability Of his ability to deal with that situation, and it seems plausible that his findings might generalize to students who could be expected to avoid learning situations in which they saw little chance for success. Roth (89), studying college students enrolled in a voluntary reading improvement course, administered a real self-concept Q-sort and an ideal self-concept Q-sort before and after the course. He found that the correlation of pre-course real self-concept with post-course real self- concept and the correlation of pre-course ideal self- concept with post-course ideal self-concept were higher for non—improvers than for improvers. On this basis, he concluded that the real and ideal self-concepts of the improvers changed more than those of the non-improvers. However, he did not establish pre-treatment equivalency, and so his findings are Open to question. Further, the role of self-concept was never made clear; if it changed, he failed to clarify whether that change followed, preceded, or accompanied any change in reading achievement. Videbeck (97) asked students to rate themselves on a task before and after attempting it and being evaluated on it; the evaluators were stooges instructed to give randomly laudatory or critical evaluations. He found that 35 self-ratings improved after approval and declined after (iisapproval, and since he controlled for initial self- rating, the change seems to have been a consequence of the evaluations. In a study of the perceptual consequences Of fail- ure, Postman and Brown (85) found that students who experienced artificially-induced failure were more prone to perceive deprivation words flashed briefly upon a screen than were students who experienced artificially-induced success; conversely, students who experienced artificially- induced success perceived more success words flashed before them than did students who had experienced artificially-induced failure. The authors concluded that success or failure eXperiences induce a set in the indi- vidual which sensitizes him to success or failure, respectively. However, the groups were not equivalent, and the study is vulnerable to invalidity by selection. Nevertheless, the objective of this study is plausible. Postman and Weingartner (26:95) write that, "what we 'see' is a product of what we believe to be 'Out there.‘ We see things not as 'they' are but as we are." The metaphors of the mind limit the environment, and it is possible that a child accustomed to failure might not per- ceive success; indeed, as Mead suggests, success might not be part of his environment (23:128). Coleman (112:320) found that minority children's sense Of control of the 36 environment had the strongest relationship to achievement of minority achievement of any of the cluster of variables he studied; perhaps these children have been socialized in a milieu of failure and powerlessness. Examining the effects of discouragement upon test performance, Gordon and Durea (66) randomly assigned 40 seventh, eighth, and ninth-grade students to either a con- trol group or an experimental group. Each group was given Form L of the 1937 Revision of the Stanford-Binet scale as a pre-test; two weeks later, both groups took Form M of the same scale, but each person in the experi- mental group was told that he had done poorly on Form L just prior to his taking Form M. Using analysis of co- variance to control for any differences among scores on Form L, the authors found that the experimental group scored significantly lower on the post-test. They con- cluded that perception of failure is a learned phenomenon. Sacks (90) found that showing interest in and playing with a group of nursery-school students was followed by significant gains in their IQ post-test, while no change resulted in a group she was uninterested in or in another group she never came into contact with. After finding no regression effect, she concluded that a child will perform better for a person he likes than for a person he doesn't know or doesn't like. Her conclusion seems unusually simplistic, especially in view of Rosenberg's findings on 37 the childhood self-esteem correlates of parental interest (27,88) and COOpersmith's findings on parental disciplinary practices and development of self-esteem in children (7). Nevertheless, the quality of the interaction does seem to have impact upon the way the child views himself as well as impact upon his test performance. Lecky (22) reported that a group of poor spellers misspelled approximately the same percentage of words, regardless of the length or the degree of difficulty of the list. They did not follow this pattern with lists of foreign words. After exposure to‘a lengthy counseling program designed to develop more positive self-images, the students misspelled fewer words. He posited a drive for self-consistency and suggested that successful spelling behavior might threaten the self-consistency of children who identified themselves as poor learners (22:144). In another experiment involving counseling, Dolan (56) found that individual counseling for a semester was accompanied by significantly greater paragraph and word comprehension among counselees than among those receiving no counseling. He concluded that "the learning process can be limited by attitudes toward it which can be altered in a positive and favorable direction through a program of effective counsel- ing" (56:919). These results are contrary to those pro- duced by Brookover's counselor eXperiment (109:136). Perhaps his counselor was ineffective. 38 In a related study, Flook and Saggar (62) randomly divided a group Of first-year college engineering students into two groups at the beginning of the academic year and administered tests of intelligence, engineering aptitude, and personality to them. The authors then disclosed the test scores to each member of one group, which they designated K, and told each member how his scores compared with those of the other students in both groups. The second group, designated NK, received no information. Group K did better on its year-end examinations than NK, prompting the authors to conclude that knowledge of test scores gave each member Of Group K a clearer and more realistic picture of his abilities in relation to the other students, hereby motivating him to seek additional help in the form of tutoring in study habits and counseling on personal problems. The counseling, however, was conducted by the junior author, so the possibility of experimenter effect cannot be discounted; she may have expressed expecta- tions of higher achievement to her counselees, unwittingly converting a study of the impact of knowledge of test scores into a study of the effects of counseling upon test performance. Studying the effects of group pressure upon per- formance, Waterman (99) gave physical fitness tests to 102 fourth-grade students from two schools, then asked them to predict their performance on the next physical fitness test. 39 He reported the scores to the children of one school in group totals, summing the totals of all students to yield a group score; at the other school, he reported individual scores to individuals. He found that the group children's performance improved by a greater increment than that of the individual children's and concluded that the group disclosure encouraged group members to assume responsibility for each. other's performance. However, he did not establish pre- treatment equivalence. Brookover found that changes in self-concept of ability were associated with changes in grade-point average over two-year periods in a longitudinal study of a group Of junior high school and senior high school students (109: 89-90) and that changes in self-concept of academic ability were also accompanied by changes in perceived evaluations of parents and teachers (109:109-110). In a related series of experiments involving students whose initial grade-point averages were below their school averages, Brookover found that involving parents in discussions designed to increase their evaluations and expectations Of their children's academic ability yielded increased academic achievement and improved self-concept of academic ability during the period of the program (109:132-136). Similar programs involving students and either a counselor or an "expert" yielded no significant results. One year after the end of the treat- ment, the improved children had regressed to their earlier 40 levels, possibly in response to the lack of support for their newly acquired self-concepts of academic ability. This research indicates that self-concept of academic ability is a learned phenomenon, that it is associated with academic achievement, that it can be improved, and that an improvement in self-concept of academic ability may be accompanied by an improvement in academic achievement. Since the branching system of the West German schools is a form of ability grouping, research on the effects of ability grouping will be examined in the next section Of this chapter. Ability Grogping Children in public schools are segregated on the basis of a variety Of criteria, among them age, place of residence, language, socio-economic status, and special needs. Many schools also segregate students on the basis of ability and attainment, reasoning that by decreasing the range of abilities within a classroom, a program can be designed to take advantage of that narrower range, thereby resulting in greater learning. Two other premises of ability grouping are that the presence of slower children impedes the progress of more advanced children and that the successes of the more advanced children heightens the dis- couragement of the slower children. However, the extant research on grouping is less than decisive on the effects 41 of ability grouping upon pupil achievement because of the confounding of selection and treatment (4); Yates (41:51) has suggested that what is needed to resolve the issue is random assignment of children and teachers to classrooms of varying types. Since this is an unlikely event, the research must be reviewed with qualification. In a review of 33 studies of ability grouping, Eckstrom (59) found that 13 studies claimed that ability grouping increases achievement, 15 studies yielded no sig- nificant effects, and five were mixed. She attributed this confusion to the wide variety of experimental conditions, methods, and samples, and noted that many studies lasted only one semester, a period of time she deemed too short to allow significant differences to emerge. She did note, however, that if the experiment provided for differentiation of methods and materials and if an effort was made to push the bright homogeneous classes, the results favored homo- geneous grouping. Unfortunately, she reviewed no studies in which teachers pushed lower-ability homogeneous classes; indeed, such studies are conspicuous by their absence from research on ability grouping. Eash (58), reviewing 26 studies, concluded that ability grouping was ineffective in increasing learning unless accompanied by curriculum adaptation and altered teaching methods. In addition, he found that ability group- ing impeded the progress of slow and average learners and 42 that ability grouping reflected a strong social class bias, with the upper classes overrepresented in high ability groups. Della-Dora's (55) findings were generally consistent with those of Eckstrom and Eash. Writing on ability grouping, or streaming, in England, Douglas noted that "Once allocated, the children tend to take on the characteristics expected of them and the forecasts Of ability made at the point Of streaming are to this extent self-fulfilling" (8:115). This assertion appears warranted in view of other evidence. Berenda (1), using an Asch technique (43), found that only seven per cent of seven to ten-year old children and 20 per cent of 10 to 13 year olds remained independent; since the Asch technique calls for judgments involving concrete phenomena, one might expect these children to be even less independent when dealing with academic ability, an abstract phenomenon. Festinger (61) suggests that this may be true. Daniels (53) found that 72 per cent of the 173 primary school teachers he studied believed that the anecdotal information and test scores which accompanied children were adequate evidence for streaming the children; only 28 per cent Of the teachers deemed the data insufficient. The studies by Berenda and Daniels suggest that children may be trapped in an ability level which is inappropriate to their abilities and which may, indeed, stifle them. Rist's (87) study of the manner in which teachers confirm their self-fulfilling prOphecies 43 about children's learning potentials is chilling evidence that not all children receive equal treatment in classrooms and that teachers' styles and expectations need to be studied more closely. Jackson (17) found that once a child enters a stream, he'll stay there. He noted that given normal shifts in IQ scores, 40 per cent of the children might move, but that only one to five per cent do. This inequity seems at variance with the importance attached to IQ scores in English schools. Millman and Johnson (83), in a study of 327 class sections enrolling 4000 seventh and eighth-grade students in 28 schools, found no consistent gains between grades seven and eight on the Iowa Tests of Basic Skills as a function of within-class variability. They used scores on the Iowa Tests as their criterion variable and concluded that decreasing within-class variability will not, Of itself, effect greater learning. Their findings are consistent with those of Goldberg, Passow, and Justman (11:34-71) and Svensson (32:124) who concluded that achievement showed no consistent variation as a function of the ability range within classrooms or as a function of the subject matter. Goldberg, Passow, and Justman noted, however, that narrow— range high-ability groups showed the least increase in learning and suggest that test ceilings and regression might cause this. However, since narrow-range low-ability groups 44 also showed little learning increase, regression can be dis— counted. They also found that the presence of bright students tended to yield higher self-attitudes for bright students and lower self-attitudes for slow students (11:79) and that slow students in broad range classrooms reported somewhat lower self-attitudes than did comparable students in narrow range classrooms. Unfortunately, the authors were unable to control for teacher effects and noted that: It is difficult to come to any firm conclusions as to the desirability of one kind of pattern over another . . . . What the children experi- ence within the classroom makes more difference in how they view themselves than does the organizational climate of the class (11:105). Husen and Svensson (71) gave tests of intelligence, arithmetic and reading comprehension to 2,755 Swedish fourth- grade students in one city before they were segregated into one Of three milieux; children from the southern half of the city remained undifferentiated during grades five and six, while children in the northern half were divided into high or low groups on the basis of their grades in the first four years of school. While they found that the high select groups showed greater gains and the low select groups showed least gains in standardized tests, they were unable to control for teacher expectations. However, they did find that between-group differences disappeared when parent social class was controlled. Eash, Rist, Husen and Svensson, and Goldberg, Passow, and Justman found a relationship between grouping 45 practices and student social class. This relationship will be examined in the next section. Grogping and Student Social Class Warner, in a study of a medium-sized New England city, noted that most of the students in the highest academic track were of upper-middle class families and that most upper-middle class students were in the highest academic track (35:194-196). In another study, he found that the high grades usually went tO children of the upper classes (34:95). Hollingshead, writing about teachers in a town Of 20,000, said that they believe that the college preparatory students have more ability . . . [and] look upon students in the general course as persons who . . . are mediocre in ability . . . . Students in the commercial course are believed to be lower in ability than those in the general course (15:171). He noted that entrance into a track was determined largely by student social class, as was grading distribution, with students Of the upper classes being over-represented in the upper tracks and receiving a disprOportionate share of high grades (15:169-172). The studies of Warner and Hollingshead touch upon the extent to which the schools reflect ascribed rather than achieved social class and raise questions about the extent to which schools thus inhibit social mobility. Later research by Wilson (102, 103) produced similar findings and similar concerns. Douglas, studying streaming in England, found that 46 middle-class students were over-represented in the upper stream and under-represented in the lower stream when standardized test scores were controlled; the converse was true for working-class students. He hypothesized that teacher judgments favored middle-class students. The problem of teacher judgment and teacher expectation has been a recurrent one in this review. The next section will deal with it. Teacher Effect In a well-known but little understood study, Lewin, Lippitt, and White (74) found that variations in leader- ship style were accompanied by variations in group behavior. Rosenthal and Jacobson (28), in a more relevant work, inferred that teacher expectations could improve student performance for some children. After gathering IQ scores for all students, the authors administered a test which ostensibly identified children who were due to experience an IQ growth spurt, and they identified those children to their teachers at the beginning of the school year. In reality, the children represented a random 20 per cent Of each class room. Testing IQ at the end of the year, the authors found no significant differences between the tagged children and others for the school as a whole, but they did find that tagged children in grades one and two showed greater IQ gains over the year than did other first and second-grade children (28:77). There were no significant 47 differences in tracks by grade level, although the middle of the three tracks showed the greatest increase. The authors suggested that since the students were not informed of being tagged, any impetus for the changes in their test scores was a consequence of changes in teacher behavior. Unfortunately, they were unable to document the dynamics Of this hypothesis. They did note, however, that the teachers in the classrooms where sig- nificant increases occurred were judged more effective than other teachers by their building principals. They also suggested that the younger children might have had more freedom to change because they were more malleable, being younger, and because their reputations as learners had not been established. The malleability theory, however, is undercut by Little's report of a United States Army experiment involv- ing men of low academic achievement (75). Under Project 100,000, men who previously would have been rejected for service because of poor school records were accepted into basic training; entrance criteria were drastically lowered but not abolished. Little reported that more than 96 per cent of these men completed basic training; the "normal" rate is 98 per cent. Further, Of those men Who went on to advanced training in communications, intelligence, medical and dental schools, electrical and mechanical equipment repair, and electronics, 87.7 per cent successfully completed the course Of study; the "normal" rate is 92 per cent. 48 While these men received compensatory basic educa- tion apart from other recruits, they underwent all other Army basic training with other recruits. Little suggested that this created an Opportunity for group support. Commenting upon the program, Little wrote that If the task were assigned to a civilian educa- tional system, it would have most likely segre- gated these men, devised a special curriculum, and engendered in them the feeling that they were clearly inferior manpower resources (75:871). It is ironic that a program devised to educate exceptional persons treated them as unexceptional. It is also encour- aging that the program demonstrated that the ceiling on ability to learn can be lifted. Summary In the first section of this chapter, the organiza- tion of the West German schools was presented. In the second section, self-concept was examined. In the first part of this section, an analysis of the theory of George Herbert Mead, it was suggested that persons come to see themselves as they believe others come to see them. In the next part of the section, various definitions of self- concept were examined, and that of Brookover was adOpted. Accuracy of self-concept was examined next, and while self- concepts may be inflated, it was suggested that since this inflation tends to be systematic, it may be of no Opera- tional consequence. In the next part, correlates of self- concept were discussed, and it was found that self-concept 49 may serve as a guide or limit upon behavior. Change in self-concept was then discussed, and it was noted that changes in self-concept are accompanied by changes in behavior. Ability grouping was the topic of the next section, with the finding that children of low ability do not profit from ability grouping when achievement is considered and that they do experience a depression of self-concept when placed in classrooms with brighter children. Grouping was found to rest on a strong social class bias. Finally, in the section on teacher effect, it was found that a change in teacher expectations could produce a change in learner behavior. In the next chapter the research design will be explained. CHAPTER III DESIGN OF THE STUDY In this chapter, the research site and sampling population will be described, the measures employed will be defined, and the research design presented. In addition, the hypotheses will be stated in testable form, and data analysis procedures will be explained and justified. Finally, limitations to the design and instrumentation will be stated. Site Data were collected during the Fall Of 1968 in Giessen, a light manufacturing city of 70,000 persons, located in Hesse, West Germany. Giessen was selected because its pOpulation and economy are similar to those of the site of previous research conducted by Brookover and associates (109). Data were collected by Harro Kahler, who also translated the instrument into German. Sample The sample was drawn from all children attending eighth grade Of any Gymnasium, Mittelschule, or Volksschule in Giessen. At the time of the study, there were 463 50 51 students enrolled in the Gymnasium, 247 students enrolled in the Mittelschule, and 321 students enroleld in the Volksschule. Enrollments in the Gymnasium and in the Mittelschule are higher than the national average because the communities near Giessen send students into Giessen's Gymnasien and Mittelschulen; however, since these communities maintain their own Volksschulen, the Giessen Volksschule enrollment is not similarly inflated. Questionnaires were administered to every student in the eighth grade of these schools. Seventeen schools were involved in the study; four Gymnasien, six Mittelschulen, and seven Volksschulen. Not all questionnaires were suitable for coding; of the 941 received, 756 were usable. A questionnaire was discarded if the student omitted family socio-economic status information, if two or more of the four school grades from his seventh grade year were missing, or if he failed to respond to more than half of the items on any of the three attitude inventories. Missing Data If three of the four school grades were available, the mean grade-point average was computed from these three. Similarly, if a student responded to more than half of the items on an attitude inventory, his mean response to the completed items was computed and assigned to the uncompleted items. These techniques made possible inclusion of 43 students, yielding a total sample of 756 students, 412 of 52 whom attended the Gymnasium, 169 of whom attended the Mittelschule, and 175 of whom attended the Volksschule. Major Variables and Instrumentation Self-Concept The basic instrument was the Michigan State General Self-Concept of Ability Scale, translated into German (Appendix A). This instrument is designed to measure general self-concept of academic ability, defined as "evaluating definitions which an individual holds of himself in respect to his ability to achieve in academic tasks in general as compared with others in his school class" (109:59). The Operational definition Of general self-concept of ability is the sum of scored responses to the Michigan State General Self-Concept Of Ability Scale (SCA). The scale consists of eight multiple-choice items, each item scored from one to five; high self-concept choice is-indicated by a high score. Reliabilityfof SCA A reliability coefficient of .86 was derived using Hoyt's analysis of variance (69); this compares with a coefficient of .865 for an American sample Of similar characteristics (109:60). Perceived Evaluations Student perceptions of the evaluations of his parents and of his favorite teacher were measured by two 53 five item multiple-choice scales, with each item scored in the same manner as an SCA item. Perceived evaluation is defined as evaluating definitions which an individual perceives another person holds of him in respect to his ability to achieve in academic tasks in general as compared with others in his school class. Perceived parental evaluation is operationally defined as the sum of scored responses to the Perceived Parental Evaluation Scale (Appendix C) and will hereafter be referred to as PPEv. Perceived teacher evaluation is operationally defined as the sum of scored responses to the Perceived Teacher Evalua- tion Scale (Appendix E), hereafter referred to as PTEv. Reliabilityiof PPEv and PTEv Hoyt's analysis of variance produced a reliability coefficient of .82 on the PPEv scale and .85 on the PTEv scale. Respective figures on the American sample were .838 and .918. Grade-Point Average Grade-point average, used as the measure of academic achievement, was derived from the grades earned by students in their seventh-grade science, social studies, German, and mathematics classes. Grades are represented on a six point scale, one being the highest and six being the lowest grade. Grade-point average is Operationally defined as the mean score of the student's grades received during his 54 seventh-grade science, social studies, German and mathe- matics classes. Socio-Economic Status Socio-economic status was measured by coding responses to two items: "What does your father (or whoever earns the most money in your family) do for a living?" and "Describe what your father (or whoever earns the most money in your family) does on the job." Harro Kahler used criteria set forth in Moore and Kleining's Soziale Selbst- Einstufung Instrument (84) to code responses. Their scheme identifies seven discrete strate and uses occupation as the identifying criterion. The upper class consists of big business men, financiers, and poli- ticians, and it represents one per cent of the West German pOpulation. The upper-middle class, about five per cent Of the pOpulation, contains upper management, professors, physicians, and attorneys. The middle-middle class is composed of middle management, engineers, teachers, and owners of middle-sized companies; it contains about 15 per cent of the pOpulation. The lower-middle class, 30 per cent of the pOpulace, consists of lower management, highly skilled workers, and owners of small businesses. Approxi- mately 28 per cent of the pOpulation falls in the upper- 1ower class, which contains such occupations as clerks, waiters, and skilled workers. The lower-lower class con- sists of unskilled manual workers and encompasses roughly 55 17 per cent of the population. The bottom four per cent, the underdogs, engage in seasonal labor or especially dirty, difficult jobs; however, only the top six strata are included in this study. A subject's socio-economic status is Operationally defined as the rating of the occu- pation of his father or head of household on the Soziale - Selbst-Einstufung Instrument and will hereafter be referred to as SES. Design The study design was a three-groups static-group comparison using one-way fixed-effects analysis of variance (4:12); this design afforded cross-sectional comparison of data from the three groups. Testable Hypotheses Hypotheses will be stated in testable form in this section. Since the study is exploratory rather than definitive, only null hypotheses will be examined. §_C_§_ The first hypothesis, stated in Chapter I (supra, p. 6), is "The relationship between student self-concept is direct; the higher the branch, the higher the self-concept of students attending that branch." Stated in testable form, this becomes, "No difference will be found between mean SCA of the Gymnasium sample, mean SCA of the Mittel— schule sample, and mean SCA of the Volksschule sample." 56 PPEv The second hypothesis (supra, page 6) was, "The relationship between perceived parental evaluation and branch is direct; the higher the branch, the higher the perceived parental evaluation of students attending that branch." Stated in testable form, this becomes, "NO difference will be found between mean PPEv of the Gymnasium sample, mean PPEv of the Mittelschule sample, and mean PPEV of the Volksschule sample." PTEV The third hypothesis (supra, page 6), "The rela- tionship between branch and perceived teacher evaluation is direct; the higher the branch, the higher the perceived teacher evaluation of students attending that branch," is tested as "NO difference will be found between the mean PTEV Of the Gymnasium sample, the mean PTEV of the Mittel- schule sample, and the mean PTEV of the Volksschule sample." §_E_S_ The fourth hypothesis was stated in Chapter I (supra, page 7) as, "The relationship between branch and family socio-economic status is direct; the higher the branch, the higher the family socio-economic status of students attending that branch." This is stated as the null hypothe- sis, "No difference will be found between the mean SES of the Gymnasium sample, the mean SES of the Mittelschule sample, and the mean SES of the Volksschule sample." 57 GPA The fifth hypothesis (supra, page 7), "There are no differences in grade-point averages between the three branches," is restated as "NO difference will be found between the mean GPA of the Gymnasium sample, the mean GPA of the Mittelschule sample, and the mean GPA of the Volks- schule sample." SCA-PPEV "The relationship between SCA and PPEv is positive for the entire sample and within the Gymnasium sample, the Mittelschule sample, and the Volksschule sample." SCA-PTEV "The relationship between SCA and PTEV is positive for the entire sample and within the Gymnasium sample, the Mittelschule sample, and the Volksschule sample." Analysis This study utilized six analytic procedures: one- way fixed-effects analysis of variance (ANOVA), Scheffé post-hoe contrasts, Pearson product-moment correlation, partial correlation, Kendall's Tau, and multiple regression. The rationale for selection of these procedures, the func- tions Of these procedures, and the assumptions underlying their use will be explained in this section. 58 An ANOVA with a = .05 was used to detect any systematic differences among the three branches. Self- Concept of Ability (SCA), Perceived Parental Evaluation (PPEv), Perceived Teacher Evaluation (PTEV), Grade-Point Average (GPA), and Socio-Economic Status (SES) were each used as a dependent variable in an ANOVA, branch serving as the independent variable in each instance. Three assumptions regarding the data must be met for ANOVA to be valid (29). The variance Of all groups must be homogeneous; all group distributions must be normal; and all error factors must be independent. Homogeneity of variance was tested by the Hartley technique (24), which refers the ratio of the largest sample variance to the smallest sample variance to a table; ratios computed for the variances of these data were within the limits defined by Hartley, and the variances were therefore treated as homogeneous. Normality has almost no importance as an assumption underlying ANOVA; the procedure is especially robust (47). Independence was inferred from the academic segre- gation of the students. Gymnasium students attend class in separate buildings. Mittelschule students and Volks- schule students, while attending classes in the same building complex, at no time attend the same classrooms, nor do they share teachers or curricula. Moreover, no score on any measure was in any way dependent upon a score 59 on any other measure. Independence was therefore assumed to be preserved. However, this ANOVA procedure does nto indicate which group means vary from which other group means; rather, it merely indicates the presence of a difference among group means. Thus, in those instances where the omnibus F was significant, it was necessary to employ post- hoc comparisons in order to identify which groups were different from which other groups. Possible procedures were those of Dunn (57), Dunnett (38:90-91), Scheffé (10:388-395; 29:68-73), and Tukey (10:383-388; 12:54-57; 29:73-75). The Dunn method was rejected because it allowed only three comparisons, and this investigator pre- ferred to retain the Option of making more than three comparisons if the data seemed to warrant them. The Dunnett method was rejected because it required a control group which this design does not provide. Since the Tukey method generally required equal-sized groups, it was deemed inapprOpriate to the discrepant sample sizes of this study. The Scheffé contrast, while somewhat more conservative than the others and therefore more likely to lead to a Type II error, nevertheless afforded greater Opportunities for data investigation and so was employed. The primary function Of the Scheffé post-hoc contrast is to construct a confidence interval around each contrast and to Observe if zero lies within this interval. If zero falls within that interval, the contrast is not 60 significant, while zero falling outside the interval implies the contrast's significance; that is, a difference between the two groups. With a = .05, Scheffé post-hoc contrasts were used to contrast each branch with the other two branches on all variables: SCA, PPEV, PTEV, SES, and GPA. Since the Scheffé contrasts Operate under the same assumptions as ANOVA, it was assumed that the conditions of variance and independence were met. Additional analyses were required to examine rela- tionships among variables, since both ANOVA and the Scheffé contrasts investigated relationships among groups. TO this end, Pearson product-moment correlations were used to compute all possible simple combinations of SCA, PPEV, PTEV, SES, and GPA. Four correlations were computed: one for the entire sample of 756 and one for each Of the three branches so that comparisons of relationships between branches could be made. Since these correlations were used only to describe the extent of relationships of variables for these groups, it was not necessary to make any assumptions about the nature of the distributions. These assumptions need to be examined only if correlations are to be used as bases for inferences about the nature Of a population (22:510). However, visual inspection of scatterplots of each set of data indicated apparent linearity and homoscedasticity. 61 Since some of the correlations might be affected by a third variable, partial correlations were computed for all the simple correlations, partialing out the effect of every other variable. For example, the SES-SCA simple correlations yielded three partial correlations: SES-SCA with PPEv partialed out; SES-SCA with PTEV partialed out; and SES-SCA with GPA partialed out. Partial correlation is an attempt to derive a degree of association between two variables exclusive of any confounding effects due to the influence of a third variable which has a relationship with each of the two variables (9:150). This measure calculates the arrays of errors of estimate for predicting variable X from variable Z and predicting variably Y from variable Z and correlates these errors (10:184). That is, the score on variable X for a given case has a degree of association with variable Z which provides a basis for prediction of X from Z, but since the association is imperfect, the prediction contains an inevitable error. It is this error between the actual score and the predicted score which is used in computing partial correlation. In order to examine the relationships between branch and SCA, PPEV, PTEV, and GPA, Kendall's Tau was used. Pro-l duct moment correlation could not be used because it requires interval data, and branch does not meet this requirement. The procedure is a measure of the extent of 62 disagreement between rankings of two sets of data (10:176). One set of data is arranged in ascending order, and the dif- ference in rank for each subject between the ordered set of data and the second set of data is computed. Tau is the ratio between the extent of similarity of the two ranks and the maximum possible similarity between the two ranks (30:215). While Tau is generally smaller than a Pearson correlation derived from the same data, it is of equivalent power (31:195). Only one assumption, ordinal data, underlies Tau. While branch is strictly nominal data, the status of the branches are not equivalent; Gymnasium is highest and Volks- schule is lowest. Further, students selected for the Gymnasium had the highest Grundsschule academic records and those students selected for the Volksschule, the lowest. Because Of these characteristics associated with it, branch was treated as an ordinal measure. Again, since the tendency of two variables to change together could be influenced by a third variable, Kendall partial rank correlation was used to examine the extent to which branch contributes to the correlations among SCA, PPEV, PTEV, and GPA. This procedure compares the agreement of rankings of each case on the retained variables with rankings on the partialed out variable, then compares the disagreement of these rankings, and finally produces an aggregate measure of agreement between the retained 63 variables which is independent of the partialed out variable (30:224-226). In addition to the correlational measures used, a multiple regression computor program develOped by the Agricultural Experiment Station of Michigan State University was employed (114). This program utilizes a least squares technique to determine whether or not selected independent variables are necessary to significantly improve prediction of a single dependent variable. That is, it asks which independent variables are the best predictors of that dependent variable. In this study, multiple regression was used on the entire sample of 756 subjects to attempt to predict SCA; independent variables employed were branch, SES, PPEV, PTEV, and GPA. Multiple regression equations were also computed within each branch to attempt to predict SCA; independent variables employed were SES, PPEV, PTEV, and GPA. Limitations Despite attempts to control for effects of extraneous factors, the possibility of their influencing the findings remains. While their possible effects may not invalidate the study, they at least point out the need for qualification of any conclusions which might emerge. One group of factors affect the internal validity of the study, questioning the extent to.which the data are 64 "real;" that is, the extent to which a discovered difference between groups is a result of factors other than those investigated in the study. The first of these factors is the validity of the instrument; the fundamental assumption is that the student reveals information which he includes in describing himself to himself, that the public disclosure equates with the private disclosure which, in turn, repre- sents an accurate statement of the student's thoughts about himself. Pencil and paper tests may or may not depict what goes on within the little black box. A second limitation is the questionable validity of the subjects' responses to the occupational questions. Subjects were asked, "What does your father (or whoever earns the most money in your family) do for a living?" According to Mr. Kahler, who translated and administered the questionnaire, many subjects were confused by the question and responded with the occupation of a relative who earned more money than the subjects' father, although the father was living at home. Mr. Kahler therefore cautioned against placing much trust in any analyses involving socio-economic status. A third limitation is selection (4:5).1 Students were not randomly assigned to the groups, thus one cannot conclude that differences among groups are due only to differential group experiences. Those differences may result largely from differences between students when they entered the groups. 65 A fourth limitation is mortality (4:5). Since the design does not control for differential drop-out rates between branches, it is possible that one branch might have a higher drOp-out rate than others and that this mortality might be concentrated within a limited stratum. Thus, Gymnasium SCA might be inflated artificially by the loss of a group of low SCA students between the fifth and eighth grades. The final factor limiting the internal validity is the exceeding the degrees Of freedom allowed by the design. While the design allowed but two degrees of freedom, the investigator decided to examine as many relationships as possible in order to provide questions for further research; the study, then, is hypothesis-generating rather than hypothesis-testing. A second group of limitations affect the generaliza- bility, or external validity, Of the study. First, since the design employs a fixed-effects data model, findings are applicable only to those effects included in the model. It is inapprOpriate, then, to generalize to any situation in which students are segregated by criteria other than their academic achievement during grades one through four. A second factor is a possible interaction between selection and treatment (4:5). Students were not randomly assigned to groups, and it is possible that the qualities which led to selection to a branch might also facilitate 66 differential growth rates. That is, there might be some- thing in Volksschule students which makes them especially vulnerable to depressive influences of the Volksschule upon SCA. Finally, the study is fundamentally correlational and examines the existence of relationships between groups and variables. Any statements about cause are based upon logical inference, not fact. Despite these apparently staggering limitations, the study nevertheless has utility. It is intended to explore the nature and extent of differences in SCA among students in a formally segregated school system. While statements regarding the possible existence of those dif- ferences must be qualified, they can provide impetus for future studies. Furthermore, the possible interference of an extraneous factor with internal or external validity does not denote invalidity but only its possible presence. CHAPTER IV FINDINGS In this chapter, the findings will be presented and discussed. Each hypothesis will be restated and followed by the relevant findings and a brief discussion. In the first section of the chapter, the three branches will be compared with respect to the other variables: students' self-concepts of academic ability (SCA), students' perceptions of parental evaluation of their academic ability (PPEv), students' perceptions of teacher evalua- tion of their academic ability (PTEV), socio-economic status Of the students' families (SES), and students' grade-point average (GPA). The next section of the chapter contains an examination of the correlations between each Of the variables, both for the entire sample and for each branch. Finally, the power of various combinations of branch, GPA, SES, PPEV, and TPEv to predict SCA will be examined. SCA The first hypothesis, "NO difference will be found between mean SCA of the Gymnasium sample, mean SCA of the 67 68 Mittelschule sample, and mean SCA of the Volksschule sample," was tested by an ANOVA with the following result: TABLE l.--ANOVA: SCA by branch. Gymnasium Mittelschule Volksschule Mean SCA 26.626 24.789 23.923 Standard Deviation 3.916 3.123 4.652 Source Sum of Degrees of Mean F P Squares Freedom Square Between 1024.337 2 512.169 33.145 <.0005 Within 11635.614 753 15.452 TOTAL 12659.951 755 These data indicate that the mean SCA of the branches are dissimilar, but do not precisely define the nature of these dissimilarities. Scheffé comparisons were utilized in order to examine the relationships between mean Gymnasium SCA and mean Mittelschule SCA and between mean Mittelschule SCA and mean Volksschule SCA (Table l, Appendix G). With the level Of rejection set at .05, the mean SCA Of Gymnasium students was found to be greater than that of Mittelschule students, but the mean SCA of Mittel- schule students was not found to be significantly greater than the mean SCA of Volksschule students. Since mean 69 Mittelschule SCA is greater than mean Volksschule SCA, and since mean Gymnasium SCA is significantly greater than mean Mittelschule SCA, it follows that mean Gymnasium SCA is significantly greater than mean Volksschule SCA. The superior mean SCA of Gymnasium students could be a result Of a number of factors. First, children are selected for Gymnasium by virtue Of having been highest achievers in their Grundsschule classes and thus may have formed high self-concepts of ability since they have had early academic success. Second, the Gymnasium has the reputation of providing the most difficult curriculum of the three branches, and inclusion in a group expected to succeed in the most difficult curriculum might support a self-conception of high academic ability. Third, the Gymnasium is the only branch preparing its leavers for university entrance, and anticipation of a future predi- cated upon academic success could produce high self- perceptions of ability among its students. These three factors might Operate singly or in concert. Conversely, Mittelschule and Volksschule students were identified by their Grundsschule teachers as unlikely to succeed in Gymnasium on the basis of less than excellent academic records. Furthermore, neither Mittelschule nor Volksschule students normally attend universities and so need not anticipate future activities predicated upon a high level of academic ability. These factors might 70 induce students in the two lower branches to hold self- concepts of academic ability somewhat lower than those of Gymnasium students. The relative similarity of mean SCAs of Mittelschule and Volksschule is puzzling when one considers that these branches embrace approximately 87 per cent of the eighth- year student pOpulation. Perhaps the fact that all but one of the six Mittelschulen and all but one of the five Volksschulen share a building with the other branch creates an environment of similarity; physical commonality might diminish self-conceptions distinguishing one branch from the other. PPEV The second hypothesis is, "NO difference will be found between mean PPEV of the Gymnasium sample, mean PPEV of the Mittelschule sample, and mean PPEV of the Volks- schule sample." An ANOVA was used to examine the variance of the sample means, employing the data found in Table 2. The same pattern held for PPEV as did for SCA, the omnibus ANOVA rejecting the null hypotheses at the .0005 level. The Scheffé comparisons produced findings similar to those from SCA data (Table 2, Appendix G). While mean PPEV of the Gymnasium sample was greater than that of either the Mittelschule sample or of the Volksschule sample, there was no significant difference between the mean PPEV of 71 TABLE 2.--ANOVA: PPEV by branch. Gymnasium Mittelschule Volksschule Mean SCA 18.255 16.069 15.337 Standard Deviation 3.059 2.650 3.440 Sum Of Degrees of Mean Source Squares Freedom Square F P Between 1260.758 2 630.379 67.052 <.0005 Within 7079.193 753 9.401 TOTAL 8339.951 755 the Mittelschule sample and the mean PPEV Of the Volks- schule sample. Parents are involved in selecting the appropriate branch for their children; teachers and parents make the selection together after children complete the fourth year of the Grundsschule, and the parents have the Option Of overruling the teachers' decisions. Thus, a teacher's decision to direct a child into the Volksschule could be negated by parents who wanted the child to attend either the Mittelschule or the Gymnasium; similarly, a parent could insist upon Volksschule or Mittelschule attendance for a child whom the teacher had designated as a potential Gymnasium student. In other words, a child would not be attending a branch his parents deemed inapprOpriate for 72 his abilities, and the child might infer a parental evaluation of his academic abilities from their having final approval of his branch assignment. While this might explain the superiority Of the mean Gymnasium PPEV to both the Mittelschule and Volks- schule mean PPEv's, it does not explain the insignificant difference between mean Mittelschule PPEV and mean Volks- schule PPEV. It is possible that students perceive no distinction in parental expectation between students of the two lower branches; perhaps the students believe that parents see two sets of students, those who attend Gymnasium and those who do not, ignoring distinctions within the latter group. PTEV The third hypothesis is, "NO difference will be found between mean PTEV of the Gymnasium sample, mean PTEv of the Mittelschule sample, and mean PTEV of the Volks— schule sample." It should be noted that PTEV is the students' perception of evaluation of him held by his favorite teacher and is not an aggregate measure of per- ceived evaluations of all his teachers. ANOVA was used to examine variance among these sample means, and the null hypothesis was rejected. However, the Scheffé comparisons produced a pattern different from those of SCA and PPEV (Table 3, Appendix G). The mean PTEV Of the Gymnasium sample was greater than 73 TABLE 3.--ANOVA: PTEV by branch. Gymnasium Mittelschule Volksschule Mean PTEV 18.138 16.211 15.302 Standard Deviation 3.031 2.572 3.512 Sum of Degrees Of Mean Source Squares Freedom Square F P Between 1127.539 2 563.769 60.646 <.0005 Within 6999.901 753 9.296 TOTAL 8127.439 755 those of both Mittelschule and Volksschule samples, and the mean PTEV of the Mittelschule sample was greater than the mean PTEV of the Volksschule; SCA and PPEV data indi- cated that while the mean of the Gymnasium sample was greater than those of both the Mittelschule and Volksschule samples, there was no difference between the means of Mittelschule and the Volksschule sample. The difference between PTEV data and both PPEV and SCA data might be rooted in the different referents the students are asked to assume; since teachers are perceived solely as func- tionaries of a specific branch, where students and parents are not, it is possible that asking students to assume the perspective of teachers heightens the salience of branch stratification and makes students more sensitive to 74 gradations of expectations. That is, a teacher is not merely a teacher, but is more accurately a teacher-at-a- Gymnasium and might be seen as such by a student who sees himself as a student without reference to branch attendance. SES The fourth hypothesis is, "NO difference will be found between mean SES of the Gymnasium sample, mean SES Of the Mittelschule sample, or mean SES of the Volksschule sample." The ANOVA data indicate that a difference does exist, and so a series of Scheffé comparisons was computed to locate the difference (Table 4, Appendix G). TABLE 4.--ANOVA: SES by branch. Gymnasium Mittelschule Volksschule Mean SES 3.67 4.50 4.81 Standard Deviation .99 .76 .64 Sum of Degrees of Mean Source Squares Freedom Square F P Between 182.727 2 94.863 125.584 <.0005 Within 568.796 753 .755 TOTAL 758.522 755 75 The Scheffé comparisons indicate that mean SES of the Gymnasium sample is greater than mean SES of the Mittelschule sample, and that mean SES of the Mittelschule sample is greater than mean SES of the Volksschule sample; it follows, then, that mean SES of the Gymnasium sample is greater than mean SES of the Volksschule sample. It appears that the branching scheme may inhibit both SCA and social mobility when it concentrates upper— class children in the Gymnasium and lower-class children in the Volksschule. While each branch draws students from all socio-economic statuses, no one branch draws exclusively from any one or two statuses, and so these schools are not creating or perpetuating a caste system. They are, however, employing ascribed status in that a child born into an upper class household is likely to be socialized in ways conducive to his being selected for the Gymnasium rather than for the Mittelschule or the Volksschule. Experiences in school further reinforce his high SCA, facilitating that academic success which enables him to move into high-level jobs, thereby maintaining his privileged position. GPA The fifth hypothesis is, "NO difference will be found between the mean GPA of the Gymnasium sample, the mean GPA of the Mittelschule sample, or the mean GPA of the Volksschule sample." It should be noted that under the West German grading system, one is the highest grade 76 and six is the lowest. Thus, the "lowest" GPA Of the Gymnasium is actually the highest GPA of the three branches. The omnibus ANOVA indicated that there are in fact dif- ferences between sample means. TABLE 5.--ANOVA: GPA by branch. Gymnasium Mittelschule Volksschule Mean GPA 3.078 3.417 3.280 Standard Deviation .634 .554 .709 Sum of Degrees of Mean D Source Squares Freedom Square F ‘ Between 15.424 2 7.712 19.161 <.0005 Within 303.064 753 .402 TOTAL 318.488 755 Using Scheffé comparisons to clarify the nature of this difference, it is apparent that the mean GPA of the Gymnasium sample is greater than those of the Mittelschule and Volksschule samples, but that the mean GPA of the Volksschule sample is not greater than the mean GPA of Mittelschule sample (Table 5, Appendix G). Further, it should be noted that since the three curricula are dissimilar, grades earned within one curricu- lum are not strictly comparable with grades earned within 77 the other two. Nevertheless, the separate curricula are ostensibly designed to provide experiences apprOpriate to the interests and abilities of the students enrolled within each branch; GPA should not vary between branches if this aim is met. With these students, however, the Gymnasium sample has the highest GPA, 3.078; the Mittelschule sample the lowest GPA, 3.417; the Volksschule sample has the middle GPA, 3.280, although Volksschule GPA is not significantly higher than Mittelschule GPA. It is possible that Gymnasium teachers expect their students to meet superior academic standards and so validate this expectation by granting them the highest grades. Con- versely, Mittelschule teachers might employ stringent evaluative criteria in an attempt to justify their students' not being placed in the Gymnasium. Such a policy of ”high standards" might also serve to justify claims of a higher academic status for the Mittelschule curriculum than for the Volksschule curriculum. Family SES, SCA, PPEV, PTEV, GPA CorrelatiOns The relationships among family SES, SCA, PPEV, PTEV, and GPA were tested by Pearson product-moment correla- tions. Data will be presented first for the entire sample, then for each branch within the sample. Zero-order and first-order partial correlations will be utilized. 78 TABLE 6.--Zero-order correlations between family SES, SCA, PPEV, PTEV, and GPA for all students. SES SCA PPEV PTEV GPA SES 1.000 .2795 .3330 .3138 .1705 SCA 1.000 .8130 .7879 .5622 PPEV 1.000 .7929 .5208 PTEV 1.000 .5003 GPA 1.000 All r's significant beyond .01 level. The correlation between SES and GPA, .1705, is negligible, and those between SES and SCA, .2795, SES and PPEV, .3330, and SES and SCA, .3138, are slight. More substantial are the correlations of GPA and SCA, .5622, GPA and PPEV, .5208, and GPA and PTEV, .5003. The strongest correlations are those involving SCA and PPEV, .8130, SCA and PTEV, .7879, and PPEV and PTEV, .7929. §§§_ The low SES-GPA correlation might be depressed by a statistical artifact; SES is measured on a six-point scale, as is GPA, while PPEV and PTEV employ 20-point scales, and SCA employs a 32-point scale. The relatively depressed ranges Of the SES and GPA scales could produce a correlation which is Speciously low. Similarly, all correlations involving SES might be artificially lessened. 79 However, the correlations involving GPA are not similarly depressed despite the use of a six-point scale to present grades. Scaling artifact, then, does not seem to explain satisfactorily the low correlations involving SES. A poSsible explanation might involve the criterion used to define family SES, the occupation of head of household. It is possible that occupation is too global a measure to have any sizeable relationship with other vari- ables. One might be stretching a point to argue that an out-of-home activity engaged in by one family member could be expected to exhibit a strong correlation with another family member. In fact, the SES-GPA correlation, .1705, indicates that variation in family SES accounts for less than three per cent (2.89%) of the variance in student GPA. The SES-GPA correlation of .17 is lower than that of .25 found in a related study by Brookover, who used an American sample of the same age (109). This difference could be a result of scaling, for where Brookover used a 100-point SES scale, this study uses a six-point scale in the correlational analysis. The decrease in the range of the SES scores might depress the magnitude Of the correla- tion. Another factor which might depress the correlations in this study is the tendency of the branching system to spread children of various socio-economic levels more equitably throughout each branch than does the American system, which tends to concentrate children of high 80 socio-economic status in college preparatory curricula and children of lower socio—economic status in general or vocational curricula. Finally, it is possible that many persons in the American schools believe that children of low socio-economic status are not capable of earning high grades (104), or that high grades are inapprOpriate in the lower tracks of a tracked curriculum; the German branching system avoids the latter problem by granting each branch virtual autonomy in establishing criteria of evaluation and measurement. While family SES has a negligible association with GPA, it does have a slight association with SCA (.2795), PPEV (.3330), and PTEV (.3138). Children might infer some expectation of academic ability appropriate to their parents' occupational level. However, these correlations are so low that they have little significance outside a table of data. Variance in family SES accounts for less than eight per cent (7.78%) of SCA variance, and less than ten per cent (9.86%) of PTEV variance. Further doubt regarding the importance of family SES in describing SCA, PPEV, PTEV, or GPA is supported by the minimal effect of partialing out SES from other corre- lations (see Table 7). Since none of the zero-order correlations is depressed appreciably by partialing out SES, one can conclude that SES exerts little independent influence upon the magnitude of these relationships. 81 TABLE 7.--Correlation coefficients between SCA, GPA, PPEV, and PTEV with SES partialed out for all students. Correlation Zero-Order SES Partialed Out SCA-GPA .5622 .5439 PPEv-GPA .5208 .4994 PTEv-GPA .5003 .4775 SCA-PPEV .8130 .7951 SCA-PTEV .7879 .7680 PPEv-PTEV .7929 .7689 All r's significant beyond .01 level. s_cp: Correlations between SCA and the other variables, with the exception of SES, are positive and strong. The correlations between SCA and PPEV and between SCA and PTEV are very high, probably because all are eXpressions of expectations or Of perceived expectations of student academic ability; the perceived evaluations, ostensibly originating in students' significant others, might be used as data for formation and maintenance of an SCA. The relationship between SCA and GPA is strong (.5622), although not as strong as those between SCA and perceived evaluations. However, GPA is a behavioral vari- able and is probably influenced by more intervening variables than the perceived evaluations, which are 82 attitudinal variables. Variance in SCA accounts for more than 31 per cent (31.61%) of the variance in GPA, and vice- versa. TABLE 8.--Correlation coefficients between SES, GPA, PPEV, and PTEV with SCA partialed out for all students. Correlation Zero-Order SCA Partialed Out SES-GPA ** .1705 - .0168 PPEv-GPA ** .5208 ** .1322 PTEv-GPA ** .5003 ** .1125 PTEv-PPEV ** .7929 ** .4249 * Significant beyond .01. SCA also influences other relationships. The SES- GPA correlation becomes virtually nonexistent when the influence of SCA is controlled, and the associations of PPEV and PTEV are appreciably diminished when SCA is par- tialed out. The PPEv-PTEv correlation, although depressed greatly by partialing out SCA, remains fairly strong. PPEV Correlations between PPEV and other variables, with the exception of SES, are positive and strong. The asso- ciations Of PPEV with PTEV and SCA are marked. 83 TABLE 9.--Correlation coefficients between SCA, GPA, PTEV, and SES with PPEV partialed out for all students. Correlation Zero-Order PPEV Partialed Out SCA-GPA ** .5622 ** .2794 SES—GPA ** .1705 - .0036 PTEv-GPA ** .5003 ** .1679 SCA-PTEV ** .7879 ** .4038 * Significant beyond .01. The effects of PPEV upon other correlations are demon- strated by partialing out PPEv. SES and GPA lose all association, and the PTEv-GPA correlation decreases to a level which, while statistically significant, holds little operational significance. SCA and PTEV retain a strong association. PTEV The pattern of relationships of PTEV and the other variables is similar to that of PPEv. Correlations with SCA and PPEV are marked, the GPA-PTEV correlation is .strong, and the SES-PTEV correlation is slight (see Table 10). Partialing out PTEV from the SES-GPA correlation weakens that correlation almost to zero. Both the SCA-GPA and the PPEv-GPA correlations, though depressed appreciably, retain a slight relationship, while the SCA-PPEV correla- tion is quite high after the partialing out of PTEV. 84 TABLE 10.--Correlation coefficients between SCA, GPA, PPEV, and SES with PTEV partialed out for all students. Correlation Zero-Order PTEV Partialed Out SCA-GPA ** .5622 ** .3152 SES-GPA ** .1705 .0164 PPEv-GPA ** .5208 ** .2352 SCA-PPEV ** .8130 ** .5017 '1: Significant beyond .01. SCA As An Intervening Variable The data indicate that PPEV and PTEV influence GPA by Operating through SCA, which thus Operates as an inter- vening variable. Students possibly utilize data from the perceived evaluations to establish and maintain SCA. SCA is interpreted as an intervening variable by virtue of its exerting a greater depressive effect upon the PPEv-GPA and PTEv-GPA correlations than do PPEV or PTEV upon the SCA- GPA correlation. Variables: l SCA 2 PPEV 3 PTEV 4 GPA r14 = .5622 r24 = .5208 r34 = .5003 r14.2 = .2794 r24.1 = .1322 r34.l .1125 r14.3 = .3152 r24.3 = .2352 r34.2 — .1679 All correlations significant beyond .01. 85 Correlations by Branch Separate correlations were computed within each branch to see if any branch varied from the others. For example, might Volksschule SCA correlate higher with Volks- schule GPA than Gymnasium SCA with Gymnasium GPA? The patterns of correlations involving the variables of the entire sample are also found among the correlations among variables within the three branches. SES Correlations involving SES are weak, with none of them exceeding .22 and only six of the twelve being signifi- cantly different from zero. While all of the Gymnasium correlations are significant, only two of the four Eglksf schule correlations and none of the Mittelschule correla- tions are significant. The only consistent pattern for correlations involving SES seems to be smallness; SES does not demonstrate a strong tendency to change with any of the other variables. SCA SCA has strong associations with PPEV, PTEV, and GPA and a weak association with SES; the correlations with PPEV range between .69 and .82; those with PTEV, between .66 and .81; those with GPA, between .51 and .57. All of these correlations are significant at the .01 level. 86 PPEV Correlations between PPEV and SCA, PPEV and PTEV, and GPA and PPEV are strong, with the PPEv-GPA correlations for each branch being lower than those between PPEV and PTEV and between PPEV and SCA. Correlations between PPEV and PTEV, GPA, and SCA are significant at the .01 level. PTEV Correlations involving PTEV follow the pattern of those involving PPEV, with the PTEv-SES correlations being weak, ranging from -.02 to .19; the Gymnasium correlation, .19, is significant at the .01 level, and the Volksschule correlation, .18, is significant at the .05 level. How- ever, these correlations are not sufficiently large to have Operational utility. PTEV demonstrates a strong relationship with GPA in all branches; the correlations, which range from .45 to .49, are significant at the .01 level. Comparisons Between Branches NO consistent pattern emerges from comparing correlations of each branch; that is, no branch demon- strates consistently high or low correlations. While the SCA-PPEV, SCA-PTEV, PPEv-PTEV, and PPEv-GPA correlations of the Mittelschule are lower than those of the Gymnasium and the Volksschule, the PTEv-GPA and SCA-GPA Mittelschule correlations are not appreciably different from those of 87 .oasnommxao> > amassomamuuflz u z «Esflmmcawm u 0 so. be bemoaowqmnm.. mo. um unmowoncunm. mv.a« mv.aa mv.«« >m9m om.%« ov.«s mm.«« mm.«« wm.s¥ mh.«« >mmm mm.aa Hm.«« hm.«« Hm.«s mm.«« hh.«« Hm.«« am.a« Nm.¥% ¢Um mo. no.| va.«e mH.« mo.| mH.«« ma. mo.| NN.«« ha.« vo. ma.ae mmm > E U > S U > 2 U > 2 G #50 >m9m Emma. ”mum .mmHmEmw mascommxao> one .masnomawuuflz .Esflmmceww now «do one .>mem .>mmm .aom .mmm consumn mucmaoammmoo coaumamuuoouu.aa mamas 88 the other branches. This inconsistency is puzzling in light of the consistently smaller variances of the Mittelschule SCA, PPEV, PTEV, and GPA data in comparison to those of the other branches. Partial Correlations Within Branches Partial correlations within each branch indicate that SES exerts little independent influence upon any correlation in any branch; no correlation drops more than one point as a result of partialing out SES. As with the entire sample, SCA appears to operate within each sub- sample as an intervening variable between Perceived Evaluations and GPA; partialing out SCA from the PPEv-GPA and PTEv-GPA correlations for each branch depresses those correlations more than partialing out PPEV from PTEv-GPA or SCA-GPA and partialing out PTEV from PPEv-GPA from PPEv-GPA or SCA-GPA. PPev exerts less influence upon correlations within the Mittelschule than within the other branches; for example, the Mittelschule SCA-GPA correlation of .51 drOps to .36 when PPEV is partialed out, while the Gymnasium SCA-GPA correlation of .57 declines to .29 with the partialing out of PPEV, and the Volksschule correla- tion is reduced from 152 to .22 after PPEV is partialed out. While the Mittelschule SCA-GPA correlation is slightly lower than those of the other two branches, the Mittelschule SCA-GPA correlation with PPEV partialed out .xaamofluozucoumm “magnumamuuflz u z “Esflmmcfimw n 0 so. be bcmonmnamnm.. UODMOHOQH manmflum> 0:0 Umamfluumm .OHsnommxHo> u > 89 mo. um bemonmncmnm. :: u: n: Laney 50. Hm.«« 0H.«« A>mmmv 00. 0H.¥ 0H.a Adumv mv.«¥ w¢.«a mv.«« Ammmv mv.«s 0¢.«¥ mv.a« >m9m :: II I: mb.«« mv.«¥ 0b.«« Admuv 0N.«« 0H.« 0N.¥« I: I: :: A>memv 0H.« 00. NH.« mv.«a mm.¥a ov.«« Adumv 0m.t« 00.4% Hm.¥« mm.«« mm.s« hh.«« Ammmv 0m.«¥ 0v.«« mm.«« mm.«« mm.«a mh.a« >mmm :: I: I: mh.«« mm.«a 00.«« mn.«« Hm.«a mh.«« Admov 0m.«« Hm.a« wm.a« I: I: I: mv.a« 0m.«a mm.«« A>m9mv NN.«« 0m.*a 0N.«¥ Nv.¥« v¢.a« mm.«* :: :I :: A>mmmv mm.«a mm.«« 0m.«a ow.a« 00.«« hh.«s Hm.e« 00.«a Hm.«« Ammmv Nm.«« Hm.«« hm.a# Hm.«« mm.«« hh.¥« Hm.«« mm.a« mm.*« «Um I: :I I: ha.« N0. va.a 0H. N0.: hH.«« ma. 00. NH.« A¢mwv H0. 50.: m0. I: :I :: v0. m0.: HH.« mo. 50. m0. A>mamv H0. 00.: m0. 0H. H0. MO. I: I: I: Ha. 0H. 00. A>mmmv H0. HH. 00. mo. 00.: mo. mo. HH.: NH.« I: :I :: Adomv 00. 50.: va.«« mH.« N0.: 0H.«« ma. m0.: NN.¥« NH.“ 00. ma.s« mmm > E O > 2 U > S U > 2 U fimO >MBm >mmm 40m .monEmm mascommxao> 0cm .mascomamuonz .ssnmmcsmw nob oso omflmnuumd manmwum> comm nuns ado mmm .>mem .>mmm .aom .mmm comBumn mucwfloflmwmoo cofiumaouuoo Hoauumm 0cm mmHmEom wasnommxao> 0cm .masnomamuuflz .anmmceaw How dam 0cm .>mem .>mmm .gom .mmm c003uon mDCOAOAMMOOO GOADMHOHHOU::.NH mqmde 90 is larger. This minimal tendency of PPEV to affect Mittelschule correlations is also reflected in correlations between SES and all other variables, between PTEV and SCA, and between PTEV and GPA. This tendency may indicate a deprecation of parental expectations by Mittelschule students; perhaps these students perceive their parents as being unsure about an appropriate level of academic performance. This possibility is not supported by data, however, and is intended merely as con- jecture. It can be seen that while partialing either PPEV or PTEV decreases the SCA-GPA correlation, that decrease is not as great as either of those which occur in the PPEv-GPA or PTEv-GPA relationships when SCA is partialed out. Again, SCA appears to Operate as an intervening variable between Perceived Evaluations and GPA. Branch Attended In order to examine the relationships of branch attendance with SCA, PPEV, PTEV, and GPA, two analyses were employed. First, Kendall's Tau was used to correlate the five variables with each other; the effects of branch variance were then partialed out of the other correlations. The correlation between Branch and GPA is low, possibly due in part to the low range of the three point scale used to define branch. This possible artifact, however, does not negate a conceivable relationship between 91 TABLE l3.--Kendall's Tau between SCA, PPEV, PTEV, GPA, and branch and partial correlations between SCA, PPEV, PTEV, and GPA with branch partialed out. SCA PPEV PTEV GPA Branch .21* .32* .30* .14* SCA .66* .63* .42* Branch partialed out .64 .61 .40 PPEV .64* .39* Branch partialed out .61 .36 PTEV .37* Branch partialed out .35 * Correlation significant at the .01 level. No test extant for significance of partial correlation. Branch and GPA. In fact, the Scheffé comparisons indicated that GPA did vary as a function of Branch. The apparent discrepancy between the results of Kendall's Tau and the Scheffé comparisons can be reconciled by considering the difference in the natures of the statistical procedures. The correlations examine the tendency Of two variables to change together; the Scheffé comparisons test the differ- ences among sample means. The failure of the two variables to exhibit a correlational effect does not imply the absence Of other differences in the data. The correlation between Branch and SCA is lower than the Branch-PPEV and the Branch-PTEV correlations. This difference indicates that Branch and SCA do not 92 increase or decrease with a high degree of similarity, but again, Scheffé contrasts indicate differences between the SCA of the branches irrespective of the rates of change of SCA within each branch. It is noteworthy, however, that the Branch-SCA correlation is lower than either the Branch-PPEV or the Branch PTEV cOrrelation. This does not seem to be due to statistical artifact; the standard deviation of SCA is greater than that of either PPEV or PTEV, and the range of SCA is also greater than the range of PPEV or PTEV. However, the reason for this pattern is unknown. Partialing the variance of Branch from the other correlations produces little change, in no case greater than three points. It is apparent, then, that the effects Of branch membership exert little, if any, independent influence upon the relationships among the other variables. This does not mean that branch membership is not related to change in other variables; those relationships were demonstrated by the Scheffé contrasts. Predicting GGA from GPA, SES, Branch, PPEV, and PTEV Multiple regression was used to see how effective GPA, SES, Branch, PPEV, and PTEV were in predicting SCA. The data indicate that Branch, relative to the other independent variables, adds little to prediction of SCA. The total set of five variables accounts for 74 per cent of the variance of SCA; deleting Branch from this set 93 TABLE 14.--Multiple regression coefficients: using GPA, SES, Branch, PPEV, PTEV to predict SCA within total sample. Delete Use R2 None All .74 2 1345 .74 4 1235 .66 5 1234 .69 1 2345 .72 3 1245 .73 24 135 .66 25 134 .69 12 345 .72 23 e 145 .73 45 123 .37 14 235 .62 34 125 .66 15 234 .66 35 124 .69 13 245 .72 245 13 .35 124 35 .62 234 15 .66 125 34 .66 235 14 .69 123 45 .72 145 23 .11 345 12 .35 134 25 .62 135 24 .66 1245 3 .08 2345 1 .31 1235 4 .66 1345 2 .08 1234 5 .62 Legend: 1 = GPA; 2 = SES; 3 = Branch; 4 = PPEV; 5 = PTEv. R2 is the square of the correlation coefficient of the dependent variable and the independent variable or variables and represents the percentage of variance of the dependent variable predicted by variance of the independent variable or variables. 94 depresses predicted variance to .73. Further, it adds or detracts little or nothing from other predictors. GPA accounts for 31 per cent of SCA variance; GPA and Branch together predict 35 per cent. SES accounts for eight per cent of SCA variance; SES and branch predict 11 per cent. PPEV accounts for 66 per cent Of SCA variance, while the inclusion of branch does nothing to R2. Similarly, PTEV accounts for 62 per cent of SCA variance, but using Branch as an additional predictor does not change R2. Finally, Branch alone accounts for but eight per cent of the variance in SCA. These results cast doubt upon the influence of branch attendance upon SCA of these students; it is possible that after four years in a branch, these students have adOpted their classmates as their reference group and no longer compare themselves to agemates in other branches. Similarly, SES adds or detracts little from the pre- dictive power of the other independent variables. GPA predicts 31 per cent of SCA variance, but GPA and SES together predict but 35 per cent Of SCA variance. Branch predicts eight per cent of SCA variance, while Branch and SES Operating in concert predict only 11 per cent. While GPA and Branch together predict 35 per cent of SCA variance, GPA, Branch, and SES predict 37 per cent. SES adds nothing to the predictive powers of PPEV, PTEV, PPEV and PTEV, GPA and PPEV, or GPA and PTEV. Finally, SES predicts only eight per cent of SCA variance. In sum, 95 then, SES is of little utility in predicting SCA in this study. GPA, used alone, accounts for 31 per cent of the SCA variance, a finding accounted for by the theoretical postulation Of a dynamic between self-concept and the environment; GPA is part of the environment used to evolve and maintain SCA, and it is a tangible consequence of one's self-conception. However, while it increases the resultant R2 of both SES and Branch from .08 to .35, these increases seem to be largely a function of GPA variance and not of any interaction between GPA and SES or between GPA and Branch. Further inclusion of GPA with PPEV, PTEV, or PPEV and PTEV produces little appreciable increase in any R2. The most effective predictive combination seems to be PPEV and PTEV, which accounts for 72 per cent of a possible 74 per cent of the SCA variance. Addition of other variables adds little to this combination's predictive power; GPA increases R2 to .73, and neither SES nor Branch changes R2. The combined power of PPEV and PTEV, however, is only slightly greater than the power of either operating alone; each accounts for 66 per cent of the variance of SCA. These relationships indicate that while the most effective predictor of SCA is the combination of PPEV and PTEV, either PPEV or PTEV is adequate operating alone. The high regression coefficients derived from PPEV, PTEV, and PPEV and PTEV are no doubt spuriously high because of the inextricable confounding of PPEV, PTEV, and SCA. 96 TABLE 15.--Mu1tiple regression coefficients: using GPA, SES, PPEV, PTEV, to predict SCA within each branch. Delete Use Gymnasium R2 Mittelschule R2 Volksschule R2 None All .74 .61 .73 2 145 .73 .60 .73 4 125 .65 .49 .69 5 124 .70 .55 .68 l 245 .70 .58 .72 24 15 .65 .49 .69 25 14 .69 .54 .68 12 45 .72 .57 .72 45 12 .35 .27 .28 14 25 .60 .43 .66 15 24 .67 .47 .66 245 l .32 .26 .27 124 5 .60 .43 .66 125 4 .67 .47 .66 145 2 .05 .00 .02 PredictingySCA from_GPA, SES, PPEV, PTEV, within Each Branch The most Obvious discrepancy among branches is the systematically lower R2 for Mittelschule; this seems to be a consequence of the lower Mittelschule variances. Using GPA, SES, PPEV, and PTEV to predict SCA within each branch yields the same patterns evident in the entire sample. SES is relatively ineffective, accounting for five per cent of SCA variance in the Gymnasium, two per cent Of SCA variance in the Volksschule, and none in the Mittelschule. Deleting SES from the four-predictor 97 set does not affect the accounted-for variance in the Volksschule and depresses the Gymnasium R2 and the Mittel- schule R2 but one point. Addition of SES to other pre- dictors produces a maximum increase in prediction of one point, and deletion Of SES from other combinations pro- duces similar negligible results. SES, then, is ineffective as a predictor of SCA within each branch, whether operating alone or in combination with other variables. GPA has some utility, predicting 32 per cent Of the Gymnasium variance, 26 per cent of the Mittelschule variance, and 27 per cent Of the Volksschule variance. It does not, however, appreciably increase the percentage of SCA variance accounted for by PPEV and PTEV Operating together. PPEV and PTEV account for 72 per cent of the possible 74 per cent of Gymnasium variance, 57 per cent of the possible 61 per cent of the Mittelschule variance, and 72 per cent of the possible 73 per cent of the Volksschule variance. Including GPA with PPEV produces little increase in R2 in any branch; from .72 to .73 in the Gymnasium, from .57 to .60 in the Mittelschule, and from .72 to .73 in the Volksschule. However, little gain in R2 could be expected because of the high correlations between PPEV, PTEv, and SCA. It is possible, however, that GPA Operates through the perceived evaluations to affect SCA. This possibility gains credibility when the relationships between perceived evaluations and SCA are 98 examined. PPEV accounts for 67 per cent of Gymnasium SCA variance, 47 per cent of Mittelschule SCA variance, and 66 per cent Of Volksschule variance. PTEV accounts for 60 per cent of Gymnasium variance, 43 per cent of Mittelschule SCA variance, and 66 per cent of Volksschule SCA variance. Since these variances are larger than those accounted for by GPA, it is possible that the students place more importance upon perceived evaluations than GPA in forming and maintaining self-concept of academic ability. More accurately, the student's self-concept is a social-psychological construct, and empirical data must be converted into a social-psychological frame of reference to have an effect upon self-concept. It is possible that GPA operates upon SCA through perceived evaluations and that PPEV and PTEV Operate as intervening variables. Differences exist in relative importance of PPEV and PTEV among branches. PPEV accounts for 67 per cent of the Gymnasium SCA variance, and PTEV accounts for 60 per cent. In the Mittelschule, PPEV accounts for 47 per cent Of the SCA variance and PTEV, 43 per cent. PPEV and PTEV each account for 66 per cent of SCA variance in the Mittelschule. Deleting PPEV drops R2 more than deleting PTEV in both Gymnasium, Mittelschule, but deleting PTEV results in a slightly larger decrease than drOpping PPEV in the Gymnasium. These results indicate that both Gymnasium students and Mittelschule students place slightly 99 more importance upon the perceived evaluations of their parents than they do those of their teachers, while Volksschule students place equal importance upon them. In the next chapter, these findings will be examined in the light of prior findings and theory, and recommendations for further study will be made. CHAPTER V SUMMARY AND CONCLUSION Introduction This study was conceived as an examination of the correlates of the self-concepts of academic ability held by West German eighth-grade students and the branch of the school system they attended. Its primary Objectives were to test the relationships between student self-concept of academic ability, perceived evaluations of others, and academic achievement, and to identify any differences in these relationships among the three branches. The basic theory of the study, drawn from George Herbert Mead and eXplained in greater detail in Chapter II, suggests that most human behavior is a function of one's conception of himself and that his self-concept is largely learned through interaction with others. This theory, when applied to learning behavior, suggests that an indi- vidual may place a ceiling upon his learning ability if he perceives that others believe him to be unable to exceed that ceiling; research evidence cited in Chapter III sup- ports this view but does not confirm it. 100 101 Summary Using this theoretical perspective, a number of hypotheses and questions were devised. These will now be examined. In this examination, SCA signifies self-concept of academic ability; PPEV, perceived parental evaluation; PTEv, perceived teacher evaluation; SES, family socio- economic status; GPA, grade-point average. These variables were Operationally defined in Chapter III. SCA The first hypothesis, "NO difference will be found between mean SCA of the Gymnasium sample, mean SCA of the Mittelschule sample, and mean SCA of the Volksschule sample,‘ was rejected. Gymnasium SCA was found to be significantly greater than Mittelschule SCA and Volksschule SCA, but Mittelschule SCA, while greater than Volksschule SCA, was not significantly greater. PPEV The second hypothesis, "NO difference will be found between mean PPEV of the Gymnasium sample, mean PPEV of the Mittelschule sample, and mean PPEV of the Volksschule sample," was rejected. Gymnasium PPEV was significantly greater than Mittelschule PPEV and Volksschule PPEV, but Mittelschule PPEV, while greater than Volksschule PPEV, was not significantly greater. 102 PTEV The third hypothesis, "No difference will be found between mean PTEV of the Gymnasium sample, mean PTEV of the Mittelschule sample, and mean PTEV of the Volksschule sample," was rejected. Gymnasium PTEV was greater than Mittelschule PTEV and Volksschule PTEV, and Mittelschule PTEV was greater than Volksschule PTEV. SES The fourth hypothesis, "NO difference will be found between mean SES of the Gymnasium sample, mean SES of the Mittelschule sample, and mean SES of the Volksschule sample," was rejected. Gymnasium SES was greater than Mittelschule SES and Volksschule SES, and Mittelschule SES was greater than Volksschule SES. GPA The fifth hypothesis, "No difference will be found between mean GPA of the Gymnasium sample, mean GPA of the Mittelschule sample, and mean GPA of the Volksschule sample," was rejected. Gymnasium GPA was significantly greater than Mittelschule GPA and Volksschule GPA, but Volksschule GPA, while greater than Mittelschule GPA, was not significantly greater. Discussion of Major Hypotheses Although the findings were apparently inconsistent, a pattern of Gymnasium Mittelschule Volksschule was 103 evident for SCA, PPEV, and PTEV. While these relationships were not consistently significant, they hint at something more than chance relationships. Gymnasium students view themselves as being more able students than do either Mittelschule or Volksschule students, and they see this evaluation reflected by their teachers and parents. It is unclear, however, whether students in any branch compared themselves to students in other branches while indicating their self-concepts of academic ability. The differences in GPA among the three branches raised some questions about the extent to which each branch met student abilities; one of the assumptions of branching is that learning activities apprOpriate to a narrow range of abilities can be engineered more easily than if a wide range of abilities must be considered. Were this the case, greater equivalence of GPA among the branches might be expected. SES and branch varied directly, with Gymnasium SES the highest and Volksschule SES, the lowest, indicating an apparent SES bias. However, this bias may be a consequence of other factors being confounded with SES. Correlational Analyses The correlations between major variables were comparable to those found by Brookover on an American sample of similar age (111). The comparisons are: 104 Correlation West German American SCA x GPA .56 .55 SCA x PPEV .81 .71 SCA x PTEV .79 .67 PPEV x GPA .52 .47 PTEV x GPA .50 .41 While the SCA x GPA correlation Of this study is comparable to that of Brookover's, the other correlations are higher. Perhaps West German eighth-grade students are more attuned to the perceptions of their elders than are their American contemporaries. Data from the present study indicated that SCA correlated higher with GPA than perceived evaluations, and further analysis indicated that SCA Operated as an inter- vening variable between GPA and perceived evaluations. SES demonstrated a low correlation with SCA and GPA, and partialing it out from other correlations yielded minimal decreases. However, since students may have mis- understood the SES items on the questionnaire, these findings are tentative. Differences in Correlations Between BranEhes Within all branches, SCA demonstrated the highest correlation with GPA; PPEV and PTEV, while exhibiting strong correlations, were not as strong. Correlations within the Mittelschule were, with the exception of SCA x GPA, systematically lower. While this may have been a 105 consequence of low variances on all variables within the Mittelschule, no definitive explanation can be presented. Rggression Analysis Multiple regression indicated that PPEV and PTEV were the most powerful predictors of SCA. However, since the three variables are confounded operationally and social-psychologically, this might be expected. The SCA, PPEV, and PTEV scales use similar wording, and subjects might develOp a response set because of the similarity. Of course, the theory used to develOp the scales posited a confounding of these factors. Branch was found to be an ineffective predictor of SCA; while Gymnasium SCA was found to be greater than either Mittelschule SCA or Volksschule SCA, branch did not account for any appreciable variance in SCA. A Kendall partial correlation yielded a similar result. This may indicate that students adopt their classmates as a reference group and do not compare their abilities to those of all agemates. Differences in Regression Coefficients Between Branches Regression analyses within each branch also indicated that PPEV and PTEV were the most effective pre- dictors of SCA. Two discrepancies were apparent. Mittel- schule coefficients were systematically lower than 106 Gymnasium and Volksschule coefficients, possibly because of lower Mittelschule variances. Second, Volksschule PTEV accounted for a larger percentage of Volksschule SCA variance than Gymnasium PTEV or Mittelschule PTEV for Gymnasium SCA or Mittelschule SCA, although the correlation coefficients between PTEV and SCA are almost identical for the three branches. While Volksschule PTEV variance is higher than the others, this does not seem an adequate explanation; Volksschule PPEV variance is also higher than the others, and the correlations of PPEV and SCA for Gymnasium and Volksschule are also similar, yet there is no discrepancy between percentage of SCA variance accounted for by Volksschule PPEV or by Gymnasium PPEV. Conclusions The data presented support the principal points of the theory. The high correlations between SCA and GPA indicate a strong relationship between an individual's evaluation of his ability to achieve academically and his academic achievement. Further, the high correlations between SCA and perceived evaluations lend credence to the notion that one's self-concept is in large part a reflec- tion of perceived conceptions held of him by others. However, the correlations are not perfect, indicating that self-concept is something more than a social mirror turned inward. Since partialing SCA from correlations involving perceived evaluations and GPA depresses those 107 correlations more than does partialing perceived evalua- tions from SCA x GPA correlations, SCA seems to Operate as an intervening variable between perceived evaluations and individual behavior. While Gymnasium students held higher SCA, PPEV, and PTEV than Mittelschule and Volksschule students, Mittelschule and Volksschule students held similar SCA and PPEV. This may indicate that the four-year period which these students spent in their respective branches cut them off from their age-mates; they seem to have identified their branch as their reference group. Perhaps they have been "cooled out" of the academic ladder at a relatively lower level than Clark described (51). Suggestions for Further Study The most obvious next step is a carefully con- trolled longitudinal study involving a group Of West German children from first grade through the end of school. Such a study would control for possible effects of selection and treatment and thereby generate more accurate knowledge regarding the relationship between academic segregation and self-concept of academic ability. Research is also needed in the area of student reference groups to investigate the extent to which students might adopt new reference groups in the course of their schooling. Do children at a given academic level 108 use classmates or agemates as reference persons? Does this vary as a function of self-concept of academic ability? Finally, there is an urgent need for research directed at identifying the mechanisms through which students acquire their self-concepts of academic ability in the classroom, at home, with friends, and in other settings. Until these mechanisms are accurately described, little can be done to change student self-concept of ability in a permanent fashion. Implications for Education The chief thrust of this study has been that persons come to View themselves in a large measure as a response to the way that they perceive others evaluate them. An implication of this human characteristic is that we are all, however tenuously, bound to one another in the sense that we can have impact upon one another. The great tragedy is that we seem relatively unaware of our impact and of our fellowship and of our love. While the study has hinted at a rather cold blooded social behaviorism as a technique for improving student self-concept of academic ability, this orientation presupposes that the primary, if not the only, function of schooling is to produce children able to perform well at academic tasks. The author does not deny that academic learning is a crucial function of schooling, but he hopes 109 that the goal of that learning can somehow be exPanded to include more human phenomena or that more human phenomena can be expended.and included in the pursuit of conventional learnings. Most of the research involving change of self- concept or change of behavior cited in this study was couched in non-humane terms. Perhaps there is a better way. Recent research by Aspy (107) and by Aspy and Hadlock (108) indicated that students of teachers who were highly empathic and non- possessively warm achieved greater gains on four of the five subtests of the Stanford Achievement Test over a year than did students of less empathic, warm teachers. Since the students Of each type Of teacher were matched for sex and pre-treatment IQ, it seems plausible that teacher empathy and warmth had something to do with the student gains. Indeed, this may be the mechanism which accounted for the gains reported by Rosenthal and Jacobson (28). Wolfensberger (105), in a discussion of psychiatric institutions, prOposed that the most effective strategy for evoking normal behavior from neurotics was to treat them as normal persons whenever possible, citing clinical research evidence in support of the prOposition that clients tend to respond to new expectations with new behaviors. Might we begin to eXpect empathy and warmth from each other? BIBLIOGRAPHY 110 10. 11. 12. BIBLIOGRAPHY Books Berenda, R. W. The Influence of the_Group on the Judgement of Peers. New York: Kingrs Crown, 1950. Borg, Walter R. Ability Grouping in the Public Schools. Salt Lake City: Utah State University, 1964. Brookover, Wilbur B., and Gottlieb, David. A Sociology of Education. 2nd ed. New York: American Book CO., 1964. Campbell, Donald T., and Stanley, Julian C. Experi- mental and Quasi-Experimental Designs for Research. Chicago: Rand-McNally and CO., 1963. Combs, ARthur W., and Snygg, Donald. Individual Behavior. New York: Harter and Row, 1959. Cooley, C. H. Human Nature and the Social Order. New York: Charles Scribner's Sons, 1902. COOpersmith, Stanley. The Antecedents of Self-Esteem. San Francisco: W. H. Freeman and Company, 1967. Douglas, J. W. B. The Home and the School. London: MacGibbon and Kee, 1963. Draper, N. R., and Smith, H. Applied Regressigp Analysis. New York: John Wiley and Sons, Inc., 1966. Glass, Gene V., and Stanley, Julian C. Statistical Methods in Education and Psychology. Englewood Cliffs, N.J.: Prentice Hall, Inc., 1970. Goldberg, Miriam L., Passow, A. Harry, and Justman,- Joseph. The Effects Of_Ability Grouping. New York: Teachers College Press, I966. Guenther, W. C. Analysis of Variance. Englewood Cliffs, N.J.: Prentice-Hall, Inc., 1964. 111 13. 14. 15. 16. 17. 18. 19. 20. 21. 22. 23. 24. 25. 26. 27. 112 Hays, William L. Statistics. New York: Holt, Rinehart, and Winston, 1963. Herriott, Robert E., and St. John, Nancy Holt. Social Class and the Urban School. New York: John Wiley and'Sons, 1966. Hollingshead, August B. Elmtown's Youth. New York: John Wiley and Sons, 1961. Huebener, Theodore. The Schools of West Germany. New York: New York University Press, I962. Jackson, Brian. Streaming: An Educational System in Miniature. London: RoutIédge andfiegan PauI, 1964. Johnson, David W. The Social Psycholggy of Education. New York: Holt, Rinehart, and Winston, 1970. Kerr, Anthony. Schools of EurOpe. London: Bowes and Bowes, 1960. King, Edmund. Education and Development in Western Euro e. Reading, Mass.: Addison-Wesley Publish- ing CO., 1969. Lawson, Robert F. Reform of the West German School System, 1945-62. Ann AfEOr: University of Michigan Press, 1965. Lecky, Prescott. Self Consistency: A Theory of Personality. New York: Island Press, 1945. Mead, George H. Mind, Self, and Society. Chicago: University of Chicago Press, 1934. Pearson, E. S., and Hartley, H. O., eds. Biometrika Tables for Statisticians. 2nd ed. Vol. I, Table 31: DiStribution of Fmax Statistic, pp. 60-61. New York: Cambridge Press, 1958. Pedley, R. The Comprehensive School. London: Pelican, 1963. Postman, Neil, and Weingartner, Charles. Teaching as a Subversive Activity. New York: Delacorte Press, 1969. Rosenberg, Morris. Society anGthe Adolescent Self- Image. Princeton, N.J.: Princeton University Press, 1965. 28. 29. 30. 31. 32. 33. 34. 35. 36. 37. 38. 39. 40. 41. 42. 113 Rosenthal, Robert, and Jacobson, Lenore. Pygmalion in the Classroom. New York: Holt, Rinehart, and WifiSton, Inc., 1968. Scheffé, Henry. The Analysis of Variance. New York: John Wiley and—Sons, Inc., 1959. Siegel, Sidney. Nonparametric Statistics for the Behavioral Sciences. New York: McGraw-Hill Book CO., 1956. Snedecor, George, and Cochrane, William. Statistical Methods. Ames: Iowa State University Press, 1967. Svensson, Nils-Eric. Ability Grouping and Scholastic Achievement. Stockholm: Almqvist and Wiksell, 1962. Strauss, Anselfm, ed. George Herbert Mead on Social Ps cholog . Chicago: University of Chicago Press, I956. Warner, W. Lloyd, et a1. Democracy in Jonesville. New York: Harper Torchbooks, 1949. Warner, W. Lloyd, et a1. Yankee City. Abridged ed. New Haven: Yale University Press, 1963. Werstler, Richard E. ThGISchools Of Eurgpe. Adrian, Mich.: Swenk-Tuttle Press, 1967? Wilson, John. Lapguage and the Pursuit of Truth. London: Cambridge University Press, 1967. Winer, Ben J. Statistical Principles in Behavioral Design. New York: McGraw-Hill Book CO., 1962. Wylie, Ruth C. The Self-Concept: A Critical Survey of Pertinent Research Literature. Loncoln: UniVersity of Nebraska Press, 1961. X, Malcolm. The Autobiography of Malcolm X. New York: Grove Press, 1964. Yates, Alfred, ed. Grouping in Education. New York: John Wiley and Sons, 1966. Articles and Periodicals Atkinson, J. W. "Motivational Determinants Of Risk- Taking Behavior." Psychological Review, 64:359- 372, 1957. 43. 44. 45. 46. 47. 48. 49. 50. 51. 52. 53. 54. 114 Asch, Solomon E. "Effects of Group Pressure upon the Modification and Distortion of Judgements." In Readings in Social Psyghology. Edited by Eleanor E. Maccoby, TheOdOre M. Newcomb, and Eugene L. Hartley. New York: Holt, Rinehart, and Winston, Inc., 1958. Bales, Robert F. "Comment on Herbert Blumer's Paper." American Journal Of Sociology, 71:545-547, 1966. Blumer, Herbert. "Society as Symbolic Interaction." In Human Behavior and Social Processes. Edited by Arnold Rose. Boston: Houghton-Mifflin CO., 1962. Blumer, Herbert. "Sociological Implications of the Thought of George Herbert Mead." American Journal of Sociology, 71:535-544, 1966. Boneau, C. A. "The Effects of Violations of Assump- tions Underlying the t-Test." Psychological Bulletin, 57:49-64, 1960. Brandt, Richard M. "The Accuracy of Self-Estimate: A Measure of Self-Concept Reality." Genetic Psychology Monographs, 58:55-99, 1958. Brookover, Wilbur B. "A Social-Psychological Con- ception of Classroom Learning." School and Society, 87:84-87, 1959. Campbell, Ernest Q. "The Internalization of Moral Norms." Sociometry, 27:391-412, 1964. Clark, Burton R. "The 'Cooling-Out' Function in Higher Education." American Journal of Sociology, 65:569-576, 1960. Dai, Bingham. "A Socio-Psychiatric Approach to Personality Organization." American Sociological Review, 17:44-49, 1952. Daniels, J. C. "The Effects of Streaming in the Primary Schools: What Teachers Believe." British Journal of Educational Psychology, 31:69-78, 1961. Davidson, Helen, and Lang, Gerhard. "Children's Perceptions of Their Teachers' Feelings Toward Them Related to Self-Perception, School Achieve- ment, and Behavior." Journal of Etperimental Education, 29:107-118, 1960. 55. 56. 57. 58. 59. 60. 61. 62. 63. 64. 65. 66. 67. 115 Dela—Dora, Delmo. "What Research Says about Grouping. Michigan Educational Journal, 16:513-542, 1960. Dolan, G. Keith. "Effects of Individual Counseling on Selected Test Scores for Delayed Readers." Personnel and Guidance Journal, 42:914-919, 1964. Dunn, Olive J. "Multiple Comparisons Among Means." Journal of the American Statistical Association, 56:52-64, 1961. Eash, Maurice J. "Grouping: What Have We Learned?" Educational Leadership, 18:429-434, 1961. Eckstrom, Ruth B. "Experimental Studies of Homo- geneous Grouping-~A Critical Review." School Review, 69:216-226, 1961. Farber, Bernard. "Social Class and Intelligence." Social Forces, 44:215-225, 1965. Festinger, Leon. "A Theory of Social Comparison Processes." Human Relations, 7:117-140, 1954. Flook, Alfred J. M., and Saggar, Usha. "Academic Performance With, and Without, Knowledge of Scores on Tests of Intelligence, Aptitude, and Personality." Journal of Educational Psychology, 59:395-401, 1968. Froelich, C. P., and Moser, W. E. "DO Counselees Remember Test Scores?" Journal of Counseling Psychology, 1:149-152, 1954. Frymier, Jack R. "A Study of Students Motivation to DO Good Work in School." Journal of Educational Research, 57:239-244, 1964. Garvey, Reba. "Self-Concept and Success in Student Teaching." Journal of Teacher Education, 21: 357-361, 1970. Gordon, L. V., and Durea, M. A. "The Effect of Discouragement on the Revised Stanford-Binet Scale." Journal of Genetic Psychology, 73:201- 207, 1948. Helper, Malcolm. "Learning Theory and Self-Concept." Journal of Abnormal and Social Psychology, 51: 184¥194, 1955. 68. 69. 70. 71. 72. 73. 74. 75. 76. 77. 78. 79. 116 Hitpass, J. "Bericht fiber eine 6 jéhrig Bewerbungs- kontrolle von Aufnahmenprfifungen und Test- prfifungen." Psychologie, 10:211-218, 1960. Howard, Robert C., and Berkowitz, Leonard. "Reactions to the Evaluations of One's Performance." Journal of Personality, 26:494-507, 1958. Hoyt, Cyril J. "Test Reliability Estimated by Analysis of Variance." Psychometrika, 6:153-160, 1941. Husen, Torsten, and Svensson, Nils-Eric. "Pedagogic Milieu and Development of Academic Skills." School Review, 68:36-51, 1960. Kahl, Joseph A. "Educational and Occupational Aspira- tions of 'Common Man' Boys." Harvard Educational Review, 23:186-203, 1953. Kinch, John W. "A Formalized Theory of the Self- Concept." American Journal of Sociology, 68: 481-486, 1963. Lewin, Kurt; Lippett, Ronald; and White, Ralph K. "Patterns of Aggressive Behavior in Experi- mentally Created Social Climates." Journal of Social Psychology, 10:271-299, 1939. Little, Roger W. "Basic Education and Youth Social- ization in the Armed Forces." American Journal of OrthOpsychiatty, 38:869-876, 1968. Lowther, Malcolm A. "Academic Achievement and Self— Esteem." The University of Michigan School of Education Bulletin, 35:8-11, 1963. McDavid, John. "Some Relationships Between Social Reinforcement and Scholastic Achievement." Journal of Consultinngsychology, 23:151-154, 1959. McDill, Edward L.; Meyers, Edmund W.; and Rigsby, Leo C. "Institutional Effects upon the Academic Behavior of High School Students." Sociology of Education, 40:181-199, 1967. MacCorquodale, Kenneth, and Meehl, Paul E. "On a Distinction Between Hypothetical Constructs and Intervening Variables." Psychological Review, 55:95-107, 1948. 80. 81. 82. 83. 84. 85. 86. 87. 88. 89. 90. 91. 117 Mann, Maxine. "What Does Ability Grouping DO to the Self-Concept?" Childhood Education, 36:357-360, 1960. Marx, Melvin H. "Intervening Variable or Hypothetical Construct?" Psyghological Review, 58:235-247, 1951. Merton, Robert K. "The Self-Fulfilling PrOphecy." Antioch Review, 8:193-210, 1948. Millman, Jason, and Johnson, Mauritz. "Relation of Section Variance to Achievement Gains in English and Mathematics in Grades 7 and 8." American Educational Research Journal, 1:47-51, 1964. Moore, Harriett, and Kleining, Gerhard. "Das Soziale Selbstbild der Gesellschaftsschichten in Deutsch- land." KOlner Zeitschrift ffir Soziologie und Sozialpsychologte, I2€86-ll9, 1960. SI Postman, Leo, and Brown, Donald R. "The Perceptual Consequences of Success and Failure." Journal rof Abnormal and Social Psycholggy, 47:213-221, 1952. "Problems of the German Schools--the Years up to 1970." Education in Germany, 1:19-25, 1963. Rist, Ray C. "Student Social Class and Teacher Expectations: The Self-Fulfilling Prophecy in Ghetto Education." Harvard Educational Review, 40:411-451, 1970. Rosenberg, Morris. "Parental Interest and Childrens' Self-Conceptions." Sociometry, 26:35-49, 1963. Roth, Robert M. "The Role of Self-Concept in Achievement." Journal Of Experimental Education, 27:265-281, 1959. Rudd, W. G. A. "The Psychological Effect of Streaming by Attainment." British Journal of Educational Psychology, 28:47-60, 1958. Sacks, Elinor L. "Intelligence Scores as a Function of Experimentally Established Social Relation- ships Between Child and Examiner." Journal of Abnormal and Social Psychology, 47:354-358,’l952. 118 92. Smith, M. Brewster. "The Phenomenological Approach in Personality Theory:' Some Critical Remarks." Journal of Abnormal and Social Psychology, 45: 516-522,’1950. 93. Snygg, Donald, and Combs, Arthur W. “The Phenomeno- logical Approach and the Problem Of Unconscious Behavior: A Reply to Dr. Smith." Journal of Abnormal and Social ngchology, 45:523-528, 1950. 94. Soares, Anthony T., and Soares, Louise M. "Self- Perceptions of Culturally Disadvantaged Children.‘ American Educational Research Journal, 6:31-45, 1969. 95. Torrance, E. Paul. "Rationalizations About Test Performance as a Function of Self-Concepts." Journal of Social ngchology, 39:211-217, 1954. 96. Trowbridge, Norma. "Effects of Socio-Economic ClaSs on Self-Concept of Children." Psychology in the Classroom, 7:302-306, 1970. 97. Videbeck, Richard. "Self-Conceptions and the Reactions of Others." Sociometty, 23:351-359, 1960. 98. "The Vocational Desires of Grammer School Leavers." Education in Germany, 1:4D-4E, 1964. 99. Waterman, Albert D.; NorthrOp, Gary W.; and Olson, Lloyd D. "Motivation and Achievement in the Elementary School." Elementary School Journal, 67:375-380, 1967. 100. Wattenberg, William W., and Clifford, Claire. "Relation Of Self-Concepts to Beginning Achieve- ment in Reading." Child Development, 35:461- 467, 1964. 101. Weiss, Robert F. "Aspirations and EXpectations: A Dimensional Analysis." Journal of Social Psychology, 53:249-254, 1961. 102. Williams, Robert L., and Cole, Spurgeon. "Self- Concept and School Adjustment." Personnel and Guidance Journal, 46:478-481, 1968. 103. Wilson, Alan B. "Residential Segregation and Aspira- tions of High School Boys." American Socio- logical Review, 14:836-845, 1959. 104. 105. 106. 107. 108. 109. 110. 111. 119 Wilson, Alan B. "Social Stratification and Academic Achievement." In Education in Dopressed Areas. Edited by Harry A. Passow. New York: Teachers College Press, 1963. Wolfensberger, Wolf. "The Concept of Normalization and Its Implications to Psychiatric Services." American Journal of Psychiatry, 127:291-297, 1970. Wrong, Dennis. "The Oversocialized Concept of Man in Modern Sociology." American Sociological Review, 26:183-192, 1961. Reports and Unpublished Material Aspy, Donald N. "A Study of Three Facilitative Conditions and Their Relationships to the Achieve- ment of Third Grade Students." Unpublished Ph.D. dissertation, University of Kentucky, 1965. Cited in Dickenson, Walter A., et al. "A Humanistic Program for Change in a Large City School System." Journal of Humanistic ngchology, 10:120, 1970. Aspy, Donald N., and Hadlock, W. "The Effect of Empathy, Warmth, and Genuineness on Elementary Students' Reading Achievement." Unpublished Ph.D. dissertation, University of Florida, 1966. Dited in Dickenson, Walter A., et al. "A Humanistic Program for Change in a Large City School System." Journal Of Humanistic Psychology, 10:120, 1970. Brookover, Wilbur B.; Erickson, Edsel L.; and Joiner, Lee M. Relationship of Self-Concept to Achieve- ment in High School. East Lansing: Human Learning Research Institute, Michigan State University, 1967. Brookover, Wilbur B.; Patterson, Ann; and Thomas, Shailer. Self-Concept of Ability and School Achievement. East Lansing: Office of Research and Publications, Michigan State University, 1962. Brookover, Wilbur B., et a1. ImprovingtAcademic Achievement Through SeIf—Concept Enhancement. East Lansing: Bureau of EducatiOnal Research Services, Michigan State University, 1965. 112. 113. 114. 115. 120 Coleman, James S., et al. Equality of Educational Opportunity. Washington, D.C.: United States Government Printing Office, 1966. McMillan, Joseph H.; Erickson, Edsel L.; and Brookover, Wilbur B.‘ "School Racial Composition and Teacher Influence." Paper presented at the American Educational Research Association Meetings, 1968, Chicago. Michigan State University Agricultural Experiment Station. Calculation Of Least Squares (Regression) Problems on the LS Routine. STAT Series Description No.77, 1969. Undeutsch, U. "Auslese ffir und durch die HOheren Schulen." In Bericht fiber den Zweiundzwanzigsten Kongress der Deutschen Gesellscheft ffir Psyéhblogie. Gottingen, I960. APPENDICES 121 APPENDIX A SELF-CONCEPT OF ACADEMIC ABILITY SCALE IN GERMAN 122 SELF-CONCEPT OF ACADEMIC ABILITY SCALE IN GERMAN Ffir jede der folgenden Fragen kreise bitte den Buchstaben vor der Antwort ein, die am besten Deine Meinung widergibt. Gib immer nur jeweils eine Antwort. 1. Wie schatzt Du Dich ein, wenn Du Deine Fahigkeit, in der Schule zu lernen, mit der Deiner besten Freunde vergleichst? a. als b. als c. als d. als e. als der Beste fiber dem Durchschnitt stehend durchschnittlich unter dem Durchschnitt stehend der Schlechteste Wie schatzt Du Dich ein, wenn Du Deine Fahigkeit, in der Schule zu lernen, mit der Deiner Klassenkameraden vergleichst? a. als b. als c. als d. als e. als der Beste fiber dem Durchschnitt stehend durchschnittlich unter dem Durchschnitt stehend der Schlechteste Wenn Du an die Abgangsklasse denkst--wie wirst Du Deiner Meinung nach abschneiden? als als als als als der Beste fiber dem Durchschnitt stehend durchschnittlich unter dem Durchschnitt stehend der Schlechteste Glaubst Du, dass Du mit Erfolg bis zum Abschluss zum Gymnasium gehen, also das Abitur bestehen kannst? a. ja. bestimmt wahrscheinlich ja bin mir nicht sicher wahrscheinlich nicht nein 123 124 Wie wfirdest Du dabei abschneiden, verglichen mit den Leistungen Deiner Klassenkameraden? a. als der Beste b. als fiber dem Durchschnitt stehend c. als durchschnittlich d. als unter dem Durchschnitt stehend e. als der Schlechteste Um Doktor, Rechtsanwalt Oder Professor zu werden, muss man mindestens vier Jahre an einer Universitat studieren. Glaubst Du, dass Du das schaffen kannst? a. hOchstwahrscheinlich ja b. ich denke schon c. bin mir nicht sicher d. ich glaube nicht e. bestimmt nicht Denke bitte ffir einen Augenblick nicht daran, wie andere Menschen Deine Leistungen einschétzen. Wie schatzt Du selber Deine Leistungen ein? a. als sehr gut b. als gut c. als durchschnittlich d. als unter dem Durchschnitt stehend e. als erheblich unter dem Durchschnitt stehend Welche Zensuren glaubst Du bekommen zu kOnnen, wenn Du nur wolltest? a. vor allem Einsen b. vor allem Zweien c. vor allem Dreien d. vor allem Vieren e. vor allem Ffinfen APPENDIX B SELF-CONCEPT OF ABILITY--GENERAL (FORM A) 125 SELF-CONCEPT OF ABILITY--GENERAL* (FORM A) Circle the letter in front of the statement which best answers each question. 1. How do you rate yourself in school ability compared with your close friends? a. I am the best b. I am above average 0. I am average d. I am below average e. I am the poorest 2. How do you rate yourself in school ability compared with those in your class at school? a. I am among the best b. I am above average c. I am average d. I am below average e. I am among the poorest 3. Where do you think you would rank in your class in high school? a. among the best b. above average c. average d. below average e. among the poorest 4. DO you think you have the ability to complete college? a. yes, definitely b. yes, probably c. not sure either way d. probably not e. no * COpyright, Bureau of Educational Research, Michigan State University, 1962. 126 127 Where do you think you would rank in your class in college? a. among the best b. above average c. average d. below average e. among the poorest In order to become a doctor, lawyer, or university professor, work beyond four years of college is necessary. How likely do you think it is that you could complete such advanced work? a. very likely b. somewhat likely c. not sure either way d. unlikely e. most unlikely Forget for a moment how others grade your work. In your own Opinion how good do you think your work is? a. my b. my c. my d. my e. my What kind of work work work work work is is is is is excellent good average below average much below average grades do you think you are capable of getting? a. mostly b. mostly c. mostly d. mostly APPENDIX C PERCEIVED PARENTAL EVALUATION SCALE IN GERMAN 128 PERCEIVED PARENTAL EVALUATION SCALE IN GERMAN Bitte beantworte die folgenden Fragen so, wie Deine Eltern sie Deiner Meinung nach beantworten wfirden. Falls Du nicht bei Deinen Elternwohnst, antworte bitte ffir die Familie, bei der Du wohnst, Kreise bitte wieder ffir jede Frage den Buchstaben vor der Antwort ein, die am besten die Ansicht Deiner Eltern wiedergibt. l. Wenn Deine Eltern Dich mit den anderen Schfilern Deines Alters vergleichen wfirden, was glaubst Du, wie sie Deine Fahigkeit zu lernen einschfitzen? a. als b. als c. als d. als e. als einen der Besten fiber dem Durchschnitt stehend durchschnittlich unter dem Durchschnitt stehend einen der Schlechtesten 2. Wie glaubst Du schatzen Deine Eltern Dein wahrschein- liches Abschneiden in der AbgangsEIasse ein? a. unter den Besten b. als c. als d. als e. als fiber dem Durchschnitt stehend durchschnittlich unter dem Durchschnitt stehend einen der Schlechtesten 3. Glaubst Du, dass Deine Eltern der Meinung sind, Du kOnntest erfolgreich bis zum Abschluss zum Gymnasium gehen, also das Abitur bestehen? a. sie glauben, ich kOnnte es bestimmt schaffen b. sie glauben, ich kOnnte es wahrscheinlich schaffen c. bin mir nicht sicher d. sie glauben, ich kOnnte es wahrscheinlich nicht schaffen e. sie glauben, ich kOnnte es bestimmt nicht schaffen 129 130 4. Um Doktor, Rechtsanwalt Oder Professor zu werden, muss man mindestens vier Jahre an einer Universitfit studieren. Wfirden Deiner Meinung nach Deine Eltern es ffir wahrscheinlich halten, dass Du das schaffen kOnntest? a. sie glauben, ich kOnnte es bestimmt schaffen b. sie glauben, ich kOnnte es wahrscheinlich schaffen c. bin mir nicht sicher d. sie glauben, ich kOnnte es wahrscheinlich nicht schaffen e. sie glauben, ich kOnnte es bestimmt nicht schaffen 5. Was glaubst Du denken Deine Eltern, welche Zensuren Du im allgemeinen bekommen konntest, wenn Du nur wolltest? a. vor allem Einsen b. vor allem Zweien c. vor allem Dreien d. vor allem Vieren e. vor allem Ffinfen APPENDIX D PERCEIVED PARENTAL EVALUATION SCALE IN ENGLISH 131 PERCEIVED PARENTAL EVALUATION SCALE Please answer the following questions as you think your parents would answer them. If you are not living with your parents answer for the family with whom you are living. Circle the letter in front of the statement that best answers each question. 1. How do you think your PARENTS would rate your school ability compared with other students your age? a. Among the best b. Above average c. Average d. Below average e. Among the poorest 2. Where do you think your PARENTS would say you would rank in your high school graduating class? a. Among the best b. Above average c. Average d. Below average e. Among the poorest 3. Do you think that your PARENTS would say you have the ability to complete college? a. Yes, definitely b. Yes, probably c. Not sure either way d. Probably not e. Definitely not 4. In order to become a doctor, lawyer, or university professor, work beyond four years of college is neces- sary. How likely do you think your PARENTS would say it is that you could complete such advanced work? a. Very likely b. Somewhat likely c. Not sure either way d. Somewhat unlikely e. Very unlikely 132 133 5. What kind of grades do you think your PARENTS would say you are capable of getting in general? a. Mostly A's b. Mostly B's c. Mostly C's d. Mostly D's e. Mostly E's APPENDIX E PERCEIVED TEACHER EVALUATION SCALE IN GERMAN 134 PERCEIVED TEACHER EVALUATION SCALE IN GERMAN Denke bitte an den Lehrer, den Du am meisten magst. Beantworte bitte die folgenden Fragen so, wie dieser Lehrer sie Deiner Meinung nach beantworten wfirde. Kreise bitte wieder ffir jede Frage den Buchstaben vor der Antwort ein, die am besten die Ansicht dieses Lehrers wiedergibt. 1. Wie wird wohl dieser Lehrer Dein Fahigkeit, in der Schule zu lernen, einschatzen, wenn er Dich mit den anderen Schulern Deines Alters vergleicht? a. als einen der Besten b. als fiber dem Durchschnitt stehend c. als durchschnittlich d. als unter dem Durchschnitt stehend e. als einen der Schlechtesten. 2. Wie glaubst Du schatzt dieser Lehrer Dein wahrschein- liches Abschneiden in der AbgangsEIasse ein? a. unter den Besten b. als fiber dem Durchschnitt stehend c. als durchschnittlich d. als unter dem Durchschnitt stehend e. unter den Schlechtesten 3. Glaubst Du, dass dieser Lehrer der Meinung ist, Du konntest erfolgreich bis zum Abschluss ans Gymnasium gehen, also das Abitur bestehen? a. er glaubt, ich kOnnte es bestimmt schaffen b. er glaubt, ich kOnnte es wahrscheinlich schaffen c. bin mir nicht sicher d. er glaubt, ich kOnnte es wahrscheinlich nicht schaffen e. er glaubt, ich kOnnte es bestimmt nicht schaffen 135 136 4. Um Doktor, Rechtsanwalt Oder Professor zu werden, muss man mindestens vier Jahre an einer Universitat studieren. Wurde Deiner Meinung nach dieser Lehrer es ffir wahrscheinlich halten, dass Du das scHaffen kOnntest? a. b. c. d. e. er glaubt, ich kOnnte er glaubt, ich kOnnte schaffen bin mir nicht sicher er glaubt, ich kOnnte schaffen er glaubt, ich kOnnte schaffen es bestimmt schaffen es wahrscheinlich es wahrscheinlich nicht es bestimmt nicht 5. Was glaubst Du denkt dieser Lehrer, welche Zensuren Du im allgemeinen bekommen konntest, wenn Du nur wolltest? a. b. c. d. e. vor allem Einsen vor allem Zweien vor allem Dreien vor allem Vieren vor allem Ffinfen APPENDIX F PERCEIVED TEACHER EVALUATION SCALE IN ENGLISH 137 PERCEIVED TEACHER EVALUATION SCALE Think about your favorite teacher--the one you like best; the one you feel is most concerned about your schoolwork. Now answer the following questions as you think this TEACHER would answer them. Circle the letter in front of the statement which best answers each question. 1. How do you think this TEACHER would rate your school ability compared with other students your age? a. Among the best b. Above average c. Average d. Below average e. Among the poorest 2. Where do you think this TEACHER would say you would rank in your high school graduating class? a. Among the best b. Above average c. Average d. Below average e. Among the poorest 3. Do you think that this TEACHER would say you have the ability to complete college? a. Yes, definitely b. Yes, probably c. Not sure either way d. Probably not e. Definitely not 4. In order to become a doctor, lawyer, or university professor, work beyond four years of college is neces- sary. How likely do you think this TEACHER would say it is that you could complete such advanced work? a. Very likely b. Somewhat likely c. Not sure either way d. Somewhat unlikely e. Very unlikely 138 139 5. What kind of grades do you think this TEACHER would say you are capable of getting in general? a. Mostly A's b. Mostly B's c. Mostly C's d. Mostly D's e. Mostly E's APPENDIX G SCHEFFE POST-HOC CONTRASTS 140 SCHEFFE POST-HOC CONTRASTS TABLE l.--Mean Gymnasium SCA vs. mean Mittelschule SCA; mean Mittelschule SCA vs. mean VOIksschule SCA. “ _- 4- Standard . . . Contrast wi-Yk Yj 8 Error of $1 Sign1f1cant Gymnasium vs. Mittelschule 1.837 2.449 .354 Yes Mittelschule vs. Volksschule .666 2.449 .423 No Legend: 01 is the symbol for the contrast of interest. Yk and Yj represent the sample mean of the kth and jtt group. S is the Scheffé constant which is equal to /(J-1) F (.05) where J is the number of groups and N is the number of subjects. The significance level is .05. TABLE 2.--Mean Gymnasium PPEv vs. mean Mittleschutg PPEV mean Mitte so u e PPEV vs. mean VoIkSschule PPEV. Standard Contrast @i=Yk-Yj 8 Error of wi Sign1f1cant Gymnasium vs. Mittelschule 2.186 2.449 .276 Yes Mittelschule vs. Volkss5hule .732 2.449 .330 No Legend: See Table l. 141 142 TABLE 3.--Mean G mnasium PTEV vs. mean Mittelschule PTEV; mean MitteIschIe PTEV vs. mean VOIEsschule PTEV. A _— ;— StandardA . . . Contrast wi-Yk Yj 8 Error of $1 S1gn1f1cant Gymnasium vs. Mittelschule 1.927 2.449 .672 Yes Mittelschule vs. Volksschule .900 2.449 .804 Yes Legend: See Table 1. TABLE 4.--Mean Gymnasium SES vs. mean Mittelschule SES; mean MittelSchule SES vs. mean VOIksEEhule SES. StandardA Contrast wi=Yk-Yj S Error of 1: S1gn1f1cant Gymnasium vs. Mittelschule -.833 2.449 .078 Yes Mittelschule vs. Volksschule -.308 2.449 .094 Yes Legend: See Table 1. TABLE 5.--Mean Mittelschule GPA vs. mean Volksschule GPA; mean Gymnasium GPA vs. mean Volksschule GPA. . A _— '— Standard . . . Contrast wi-Yk Yj S Error of $1 S1gn1f1cant Mittelschule vs. Volksschule .138 2.449 .167 NO Gymnasium vs. Volksschule -.201 2.449 .142 Yes Legend: See Table 1. APPENDIX H STANDARD DEVIATIONS OF ALL VARIABLES FOR TOTAL SAMPLE AND FOR EACH BRANCH 143 STANDARD DEVIATIONS OF ALL VARIABLES FOR TOTAL SAMPLE AND FOR EACH BRANCH SES SCA PPEV PTEV GPA TOTAL Sample 1.00 4.09 3.32 3.28 .65 Gymnasium .99 3.92 3.06 3.03 .63 Mittelschule .76 3.12 2.66 2.57 .55 Volksschule .64 4.65 3.45 3.51 .71 144 APPENDIX I MEANS OF ALL VARIABLES FOR TOTAL SAMPLE AND FOR EACH BRANCH 145 MEANS OF ALL VARIABLES FOR TOTAL SAMPLE AND FOR EACH BRANCH SES SCA PPEV PTEV GPA TOTAL Sample 4.12 25.60 17.10 17.06 3.20 Gymnasium 3.67 26.63 18.25 18.14 3.08 Mittleschule 4.50 24.79 16.07 16.21 3.42 Volksschule 4.81 23.92 15.34 15.30 3.28 146 "I11111111111111.1111“